Method and system for constructing spatio-temporal dynamic atlas of engineering construction projects

Abstract

The present invention provides a method and system for constructing a spatio-temporal dynamic atlas of engineering construction projects, which relates to the field of railway tunnel engineering management, and includes: extracting construction data entities to generate an entity set; adding the entities in the entity set as nodes to the atlas and configuring attributes for each node; adding relationship edges to each node according to the logical structure between the construction data; adding time tags and space tags to each node to form a data atlas; associating each entity in the data atlas with a time chain and a space chain to form a data atlas attached to the spatio-temporal chain; combining a node merging strategy and a graph neural network to optimize the atlas structure; and dynamically modeling the changes of entities and relationships in the time dimension relying on a temporal graph neural network. The present invention constructs a dynamic change atlas of the relationships between construction data entities, enabling the spatio-temporal information and entity relationships of each process to be accurately represented and efficiently associated, which helps to improve the retrieval efficiency and management level.

Description

Technical Field

[0001] The present invention relates to the technical field of railway tunnel engineering management. Specifically, it relates to a method and system for constructing a spatio-temporal dynamic atlas of engineering construction projects. Background Art

[0002] The construction data of railway engineering projects is characterized by being numerous, miscellaneous, disorderly, and discrete, lacking a unified modeling and storage method. Especially in long and complex linear projects, a large amount of heterogeneous data is involved, such as geological forecasts, construction logs, progress ledgers, maintenance records, and experimental test reports. These data contain key information such as construction progress, time, and space. Traditional data recording carriers mainly process linear data and cannot effectively integrate spatio-temporal dynamic information, resulting in information disconnection and affecting construction decision-making and overall management and control. Although technologies based on RAG (Retrieval-Augmented Generation) can store data in a vector database, time and space information will be lost, leading to inaccurate retrieval results. Existing data atlas technologies can better connect various data through nodes and edges, but how to accurately construct an entity relationship atlas containing spatio-temporal information is still a technical problem. Therefore, there is an urgent need to propose a lossless and structured data storage and representation method. Summary of the Invention

[0003] The present invention aims to solve at least one of the technical problems existing in the prior art or related technologies, and discloses a method and system for constructing a spatio-temporal dynamic atlas of engineering construction projects, constructing a dynamic change atlas of the relationship between construction data entities, enabling the spatio-temporal information and entity relationships of each process to be accurately represented and efficiently associated, not only improving the retrieval efficiency but also precisely managing and tracking the dynamic changes during the construction process, providing an important basis for real-time monitoring, quality assessment, and progress optimization of the construction.

[0004] The first aspect of the present invention discloses a method for constructing a spatio-temporal dynamic atlas of engineering construction projects, including: extracting a construction data entity set: extracting construction data entities from construction process document materials to generate an entity set; constructing a data atlas: adding the entities in the entity set as nodes to the atlas and configuring attributes for each node; adding relationship edges to each node according to the logical structure between the construction data; adding time tags and space tags to each node to form a data atlas; attaching time chains and space chains: sorting the nodes according to the time information of the construction process to form a chain in the time dimension; classifying the nodes according to the construction location or mileage range to form a chain in the space dimension; using the time tags and space tags to associate each entity in the data atlas with the time chain and space chain to form a data atlas attached to the spatio-temporal chain; atlas compression: calculating the similarity between nodes, identifying similar nodes for node merging according to a set similarity threshold, and the node merging strategy includes: aggregating according to the time attribute or space attribute of the similar nodes, taking the similar nodes as child nodes and aggregating them into the same parent node; if there are differences in some attributes among the merged child nodes, combining the different attributes into multi-valued attributes; if there are multiple edges between the merged child nodes and the same node, retaining the weight information of the multiple edges; capturing the deep relationship between nodes and edges in the data atlas based on a graph neural network to optimize the weights of the nodes and edges; constructing a dynamic atlas: based on a temporal graph neural network, introducing a time embedding vector for each node in the data atlas, the time embedding vector is generated according to the historical feature sequence of the node and is updated iteratively over time; in the convolution process of each layer of the temporal graph neural network, using a time series aggregation mechanism to fuse the current state of the node with the information of historical time points; using a time-sensitive edge weight update mechanism to dynamically adjust the weight value of the edge, and by adding time information to the calculation of the edge, the temporal graph neural network can embed time factors in the weight of the edge; to ensure the continuity and consistency of the data atlas over time, the temporal graph neural network adopts a state transition mechanism to periodically insert new nodes and edges into the atlas, and the information of the new nodes is generated based on historical similar nodes and time embedding vectors; retraining the features of the nodes and edges according to the actual construction requirements so that the model can reflect the real-time state of the construction process.

[0005] According to the method for constructing a spatio-temporal dynamic atlas of engineering construction projects disclosed by the present invention, preferably, before the step of extracting the construction data entity set, it further includes: data preprocessing: performing unified normalization and structuring processing on the document materials generated during the construction process.

[0006] According to the method for constructing a spatio-temporal dynamic atlas of an engineering construction project disclosed by the present invention, preferably, after the step of constructing the dynamic atlas, it further includes: Dynamic atlas optimization: Optimize the temporal graph neural network by means of time window adjustment, weight adaptive update, and data incremental update to improve the performance and computational efficiency of the temporal graph neural network in a dynamic environment.

[0007] According to the method for constructing a spatio-temporal dynamic atlas of an engineering construction project disclosed by the present invention, preferably, it further includes: By parsing the construction log, record the spatial changes as time series data; Map the state update of the spatial position to the corresponding time node; Whenever new construction log data is input, parse and map the spatial information involved in the log to a specific time point of the time atlas; By adding the state update of the spatial position to the slice of the time atlas, dynamically expand the features of each time node, so that the spatial changes are represented by the update of the time atlas.

[0008] According to the method for constructing a spatio-temporal dynamic atlas of an engineering construction project disclosed by the present invention, preferably, after the step of atlas compression, it further includes: Efficient retrieval strategy: By traversing the time chain, locate the data atlas slice of the target time point, and retrieve the construction status and material usage data included in the data atlas slice; Or by selecting the start and end time points on the time chain, locate and retrieve all relevant data atlas slices within the time interval, and perform aggregation analysis to generate the overall situation within the target time period; Or by traversing the spatial chain and finding the target construction site location, locate the relevant data atlas slice, and then combine the construction log to locate the time, so as to transform the spatial point problem into a time point problem for retrieval; Or set the start point and end point on the spatial chain, mark all construction sites within the range, identify the corresponding time period of each construction site on the time chain, obtain the time slices within the spatial range, and perform the aggregation operation of multiple slices to generate the overall analysis data of the construction situation within the construction site interval.

[0009] According to the method for constructing a spatio-temporal dynamic atlas of an engineering construction project disclosed by the present invention, preferably, the construction process data documents include: production notice, batching notice, ecological ledger, mix ratio report, smooth blasting self-evaluation form, tunnel visualized surrounding rock list, process writing, concrete curing record form, quality and safety inspection notice, and three-dimensional section scanning data.

[0010] According to the method for constructing the spatio-temporal dynamic atlas of engineering construction projects disclosed by the present invention, preferably, the construction data entities include: construction location, concrete, surrounding rock, over-excavated volume, over-limit volume, progress index, heading face excavation, primary support, invert primary support, invert filling, invert lining, lining arch and wall, prediction conclusion, primary support concrete, invert primary support concrete, invert filling concrete, invert lining concrete, lining arch and wall concrete, actual usage, stockpiling volume, over-consumption, drilling, charging and blasting, mucking, erecting and wire meshing, locking feet and anchor rods, inspection, shotcrete, and profiling.

[0011] The second aspect of the present invention discloses a spatio-temporal dynamic atlas construction system for engineering construction projects, including: a memory for storing program instructions; and a processor for calling the program instructions stored in the memory to implement the method for constructing the spatio-temporal dynamic atlas of engineering construction projects according to any one of the above technical solutions.

[0012] The beneficial effects of the present invention at least include: Based on the application of the spatio-temporal dynamic atlas, it can accurately manage and track the dynamic changes during the construction process, providing an important basis for real-time monitoring, quality assessment, and progress optimization of the construction; at the same time, the dynamic modeling technology of the spatio-temporal atlas supports the real-time update of the data atlas of railway construction projects, providing an efficient and flexible tool for the monitoring, management, and optimization of the construction process. Specifically, the atlas compression algorithm is a lossless compression, which significantly reduces the scale of the atlas while retaining the core information and avoiding the storage and processing of redundant information. For example, when conducting construction progress analysis, the spatio-temporal dynamic atlas can be directly generated based on the merged node information, saving memory space and accelerating the traversal and query efficiency of the graph structure. This compression method is particularly applicable to highly repetitive scenarios in railway construction, such as adjacent construction records and identical material usage records, enabling the system to significantly reduce the storage cost and consumption of computing resources while maintaining the integrity of information. Temporal GNN (Temporal Graph Neural Network) captures the changes of nodes and edges over time in the time dimension through hierarchical temporal learning, supports the dynamic update of the atlas, and maintains the consistency between the atlas and the construction site. Based on an efficient retrieval strategy, fast and accurate multi-dimensional queries can be realized on the highly compressed spatio-temporal dynamic atlas, greatly improving the management and monitoring capabilities of railway construction projects. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 Shows a schematic diagram of the overall architecture of the method for constructing the spatio-temporal dynamic atlas of engineering construction projects according to an embodiment of the present invention.

[0014] Figure 2 Shows a schematic block diagram of the spatio-temporal dynamic atlas construction system for engineering construction projects according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0015] In order to more clearly understand the above objects, features and advantages of the present invention, the present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments. Many specific details are set forth in the following description in order to fully understand the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the present invention is not limited to the limitations of the specific embodiments disclosed below.

[0016] Disclosed according to an embodiment of the present invention is a method for constructing a spatio-temporal dynamic atlas of an engineering construction project, including:

[0017] Extracting a construction data entity set: extracting construction data entities from construction process data documents to generate an entity set;

[0018] Constructing a data atlas: adding the entities in the entity set as nodes to the atlas and configuring attributes for each node; adding relationship edges for each node according to the logical structure between the construction data; adding time tags and space tags to each node to form a data atlas;

[0019] Attaching the time chain and the space chain: sorting the nodes according to the time information of the construction process to form a chain in the time dimension; classifying the nodes according to the construction location or mileage range to form a chain in the space dimension; using the time tags and space tags to associate each entity in the data atlas with the time chain and the space chain to form a data atlas attached to the spatio-temporal chain;

[0020] Atlas compression: calculating the similarity between nodes, identifying similar nodes according to a set similarity threshold for node merging, and the node merging strategy includes: aggregating according to the time attribute or space attribute of the similar nodes, taking the similar nodes as child nodes and aggregating them into the same parent node; if there are differences in some attributes among the merged child nodes, combining the different attributes into multi-valued attributes; if there are multiple edges between the merged child nodes and the same node, retaining the weight information of the multiple edges; capturing the deep relationship between the nodes and edges in the data atlas based on a graph neural network to optimize the weights of the nodes and edges;

[0021] Efficient Retrieval Strategy: By traversing the time chain, locate the data graph slice at the target time point and retrieve the construction status and material usage data contained in the data graph slice; or by selecting the start and end time points on the time chain, locate and retrieve all relevant data graph slices within the time interval, and perform aggregation analysis to generate the overall situation within the target time period; or by traversing the space chain and finding the target construction site location, locate the relevant data graph slice, and then combine the construction log to locate the time, so as to transform the space point problem into a time point problem for retrieval; or set the start point and end point on the space chain, mark all construction sites within the range, identify the corresponding time period of each construction site on the time chain, obtain the time slices within the space range, and perform the aggregation operation of multiple slices to generate the overall analysis data of the construction situation within the construction site interval;

[0022] Constructing a Dynamic Graph: Based on the temporal graph neural network, introduce a time embedding vector for each node in the data graph. The time embedding vector is generated according to the historical feature sequence of the node and is iteratively updated over time; during the convolution process of each layer of the temporal graph neural network, use the time series aggregation mechanism to fuse the current state of the node with the information at historical time points; use the time-sensitive edge weight update mechanism to dynamically adjust the edge weight value. By adding time information to the calculation of the edge, the temporal graph neural network can embed time factors in the edge weight; to ensure the continuity and consistency of the data graph over time, the temporal graph neural network adopts a state transition mechanism to regularly insert new nodes and edges into the graph. The information of the new nodes is generated based on historical similar nodes and time embedding vectors; retrain the features of nodes and edges according to the actual construction requirements to ensure that the model always reflects the real-time state of the construction process.

[0023] As Figure 1 shown, the specific implementation methods of the spatio-temporal dynamic graph construction method for the engineering construction project disclosed in the above embodiments include the following aspects:

[0024] S1, Domain Modeling: For the specific scenario of railway tunnel construction, construct an engineering construction data graph attached to time and space;

[0025] S2, Graph Compression Algorithm: Combine the node merging strategy of multi-timestamp and attribute aggregation and the graph neural network (GNN) to perform deep compression modeling on the spatio-temporal graph and efficiently optimize the graph structure;

[0026] S3, Efficient Retrieval: Based on the construction log medium, rely on temporal efficient retrieval;

[0027] S4, Dynamic Update Algorithm: Rely on the temporal graph neural network (Temporal GNN) to dynamically model the changes of entities and relationships in the time dimension.

[0028] The above S1 includes: In the complex environment of railway tunnel construction, there is a large amount of data generated during the construction process and it has high structural requirements. These data include but are not limited to: production notice, batching notice, ecological ledger, mix ratio report, smooth blasting self-evaluation form, tunnel visualized surrounding rock list, process writing, concrete curing record form, quality and safety inspection notice, and three-dimensional section scanning, etc. First, perform unified standardization and structuring processing on these data to meet the needs of subsequent analysis and model construction. S1 specifically includes:

[0029] S1.1 Key scenario analysis:

[0030] For the key scenarios during the construction process, four main analysis objectives are defined:

[0031] Judgment of the timing of secondary lining construction: In the drill and blast method construction of railway tunnels, combined with surrounding rock monitoring data and construction logs, conduct comprehensive analysis and cross-verification of the compliance of secondary lining construction to ensure the safety and stability of the structure, optimize the project progress, reduce project costs, and meet the requirements of specifications and regulations.

[0032] Analysis of concrete consumption: Accurately analyze the demand and actual consumption of concrete to optimize resource allocation and cost control.

[0033] Analysis of concrete quality: In the drill and blast method construction of railway tunnels, track and evaluate the quality of secondary lining concrete by analyzing the compressive strength and rebound strength of specimens, identify problems in the design and construction processes, so as to improve construction quality and reduce economic costs.

[0034] Analysis of construction progress: Real-time monitor the construction progress, adjust the construction plan and resource allocation in a timely manner to ensure the project is completed on schedule.

[0035] S1.2 Entity extraction and data graph construction:

[0036] For the selected scenarios and construction data under the scenarios, through expert analysis, some examples of the entity set obtained are as follows: construction location, concrete, surrounding rock, over-excavated volume, over-limit volume, progress index, face excavation, primary support, invert primary support, invert filling, invert lining, lining arch wall, forecast conclusion, primary support concrete, invert primary support concrete, invert filling concrete, invert lining concrete, lining arch wall concrete, actual consumption, batch volume, over-consumption, drilling, charging and blasting, mucking, setting up the frame and hanging the mesh, locking feet and bolts, inspection, shotcrete, scraping the section

[0037] Based on the above entity set, the construction process of the data graph includes the following steps: According to the extracted entity set, create the core entity objects in the graph. For example, entities such as "construction mileage", "excavation method", "equipment type", etc. will be added as nodes to the graph; then assign the corresponding attributes to each entity object. For example, the attributes of the "maintenance record" entity may include maintenance method, temperature, humidity, etc.; the attributes of the "inverted arch primary support" entity may include the front mileage, footage, etc.; then, according to the logical structure in the construction data, with the guidance of experts, add relationship edges to relevant entities; then bind the entity objects with time and space tags. For example, mark the start date and end date of the "inverted arch primary support concrete" entity, and at the same time mark the construction site where it is located.

[0038] S1.3 Attachment of time-space chain:

[0039] Regarding the time and space sensitivity of railway tunnel construction data, through the following operation steps, the graph is concatenated to a one-way chain of time and space to achieve the spatio-temporal binding of data and avoid loss and interference:

[0040] (1) Extract all entity nodes and their associated attributes and relationships from the constructed data graph.

[0041] (2) Sort the entities according to the time nodes in the construction process (such as the start date and end date of construction, etc.) to construct a chain in the time dimension.

[0042] (3) Divide the entities into corresponding space nodes according to the construction location or mileage range. For example, use construction sites, mileage segments, etc. as the classification basis.

[0043] (4) Use time and space tags to associate the sorted entities with their corresponding time nodes and space nodes to form a spatio-temporal serialized chain relationship.

[0044] (5) Integrate the processed spatio-temporal association relationship with other information in the graph and output a data graph containing a complete time-space chain.

[0045] The above S2 includes: Relying on the spatio-temporal sequence chain to construct a graph in railway construction projects can ensure the accuracy of the spatio-temporal information of the data and facilitate the systematic archiving of the data. However, the huge amount of data generated during the construction process and the high similarity between different graphs also cause a waste of resources. For example, multiple similar entity nodes may be generated for the same construction part recorded at different time points, and the relationships of these nodes (such as material usage, construction process status, etc.) may also highly overlap. To solve the contradiction between accuracy and practicality, a graph compression algorithm based on graph neural network (GNN) is proposed to optimize the storage space, improve the data processing efficiency, and accelerate the process of data query and analysis. S2 specifically includes:

[0046] S2.1 Redundant data identification and removal:

[0047] First, a comparison and merging algorithm based on node similarity is proposed. The core idea of this algorithm is to automatically identify and merge similar nodes by calculating the similarity between nodes according to a set similarity threshold. The specific steps are as follows:

[0048] (1) Node similarity calculation: The features of each node can be regarded as a vector (for example, including node attributes such as construction location, material type, timestamp, construction status, etc.). A vector similarity calculation method is used to measure the similarity between two nodes. The similarity formula is:

[0049]

[0050] where and are the feature vectors of two nodes to be compared, is the dimension of the feature vector, and i represents the i-th node.

[0051] (2) Comparison of similar nodes and threshold determination: Set a similarity threshold , if the similarity of two nodes , then they are considered redundant nodes. The setting of this threshold can be adjusted according to the specific characteristics of the data. For example, in a group of nodes with high similarity, appropriately reduce the value to enhance the merging effect.

[0052] (3) Node merging: When two nodes are determined to be redundant nodes, the system will merge these nodes into a new node and merge its edges. The weights of the merged edges will be weighted and averaged according to the importance of the edge relationship to retain the representativeness of the edges. For example, if node and node both have the same type of edges with node , then the weight of the merged edge can be calculated as:

[0053]

[0054] where and are the edge weights before merging, and the merged weight is their average value.

[0055] For example: Suppose there are two nodes and , which represent the states of different construction sites and have the following characteristics: Node : Location is "Construction Site 1", timestamp is "October 1, 2024", material type is "concrete", and status is "primary support" node : The location is "Construction Site 1.5", the timestamp is "October 1, 2024", the material type is "concrete", and the status is "primary support". Due to nodes and being exactly the same in terms of time, material, and status, and the construction locations being very close, they can be regarded as redundant nodes. Assuming the similarity threshold is 0.9, and and have a similarity of 0.95, then the merging condition is met. The system will merge and into a new node , and the spatial attribute can be represented by "Construction Site 1 - 1.5", and all associated edges will be merged.

[0056] Optimization effect: This node comparison and merging algorithm significantly reduces the scale of the graph while retaining the core information, avoiding the storage and processing of redundant information. For example, when conducting construction progress analysis, a spatio - temporal dynamic graph can be directly generated based on the merged node information, saving memory space and accelerating the traversal and query efficiency of the graph structure. This compression method is particularly applicable to highly repetitive scenarios in railway construction, such as adjacent construction records, identical material usage records, etc., enabling the system to significantly reduce the storage cost and consumption of computing resources while maintaining information integrity.

[0057] S2.2 Node Merging Strategy

[0058] In the above process, in order to retain the independent information of different nodes while performing effective lossless compression, a node merging strategy that combines multiple timestamps and attribute aggregation will be adopted. This strategy can not only retain the specific information of each node but also distinguish the status and attributes within different time - space segments during query. The node merging strategy specifically includes:

[0059] 1. When the core attributes such as the spatial (temporal) location and material type of two nodes are the same, while the time (space) stamp is slightly different, they are regarded as "child nodes" in the time (space) order, rather than completely redundant nodes. The system aggregates these child nodes onto a "parent node" and retains all time (space) stamp information within the "parent node" so that the specific time (space) points can be distinguished during query. For example, if nodes and are both located at "Construction Part 1", have the same material type "concrete" and status "primary support", but the timestamps are "October 1, 2024" and "October 2, 2024" respectively, then the two timestamps can be retained in the merged node and represented as [October 1, 2024, October 2, 2024].

[0060] 2. If the merged child nodes differ in some attributes, the system will retain these attributes as multi-valued attributes. For example, different material usage amounts or different monitoring values can be stored separately as time series data and accurately called by timestamp during query. For the attribute of "concrete usage amount" in the example, if the concrete usage amounts of node and are 50 tons and 60 tons respectively, then in the merged node the concrete usage amount will be recorded as [October 1, 2024: 50 tons, October 2, 2024: 60 tons].

[0061] 3. Multi-weight merging of edges: If there are multiple edges between the merged nodes and the same node, the weight information of multiple edges will be retained to distinguish the relationship weights at different time points during query. For example, if nodes and respectively have different construction activity records with node , after merging, the timestamp and activity type will be retained on the edge, and the corresponding activity records can be accessed by timestamp when needed.

[0062] Effect of the merging strategy: Through the above node merging strategy with multiple timestamps and multiple attributes, both effective graph compression is achieved and the traceability and distinguishability of information at different time periods and with different attributes are ensured. This method can significantly optimize the storage efficiency of spatio-temporal dynamic graphs and at the same time provide more flexible query support.

[0063] S2.3 Feature learning:

[0064] After removing redundant information and merging similar nodes in the graph, to further optimize the model's expressive ability and feature weights, a graph neural network (GNN) is introduced for feature training. Through feature learning, the model can capture the deep relationships between nodes and edges in the graph, achieve weight optimization for nodes and edges, and thus extract key information more effectively. The specific steps are as follows:

[0065] Capturing local graph structure information: During feature training, first use a graph convolutional network (GCN) to capture the local connection features between entities in the graph. GCN integrates the information of adjacent nodes into the current node through neighborhood aggregation operations in each layer, so that the features of each node are not only its own information but also the context information of its neighborhood structure. This process of feature aggregation can effectively extract the local structure features of the graph and deepen the understanding of complex graphs layer by layer. The feature update formula of GCN is

[0066]

[0067] where represents the The node feature matrix of the layer is the adjacency matrix of the graph is the degree matrix is the weight matrix for training is the non - linear activation function. Through GCN, the model can aggregate the information of neighboring nodes layer by layer, effectively capture the local features of the graph spectrum, and enable the model to better understand the connection characteristics and structural relationships between nodes.

[0068] Assignment of relationship edge weights: In addition to the feature aggregation of nodes, the model also needs to dynamically assign weights to each edge to identify and highlight important relationships. For this purpose, the Graph Attention Network (GAT) is introduced to dynamically calculate the weights of the edges. By introducing the attention mechanism, GAT enables the model to assign adaptive weights between neighboring nodes, thereby enhancing the representation ability of important relationships. Its attention weight calculation formula is:

[0069]

[0070] where is the weight of the edge between node and is the learnable weight matrix represents the vector concatenation operation T represents the attention mechanism parameter j represents the input feature vector of the j - th node i represents the input feature vector of the i - th node. The edge weights obtained after Softmax regularization are used to weighted - aggregate node features, thereby enhancing the discrimination and representation ability of important relationships. Through this process, the model can distinguish different node - to - node relationships according to the edge weights and achieve fine - grained modeling of the graph spectrum.

[0071] Modeling and prediction of relationships between entities: After capturing the features of nodes and edges, the entity embeddings are provided as inputs to the GNN to learn and predict the relationship types between entities. By optimizing the adjustment of feature embeddings and weights, the model can effectively capture the complex relationship information in the graph spectrum. This process ensures that the relationship information in the complex network structure is accurately modeled, while improving the prediction ability of the model and making the relationship representation more refined and accurate. Specifically, by training the model to optimize the loss function:

[0072]

[0073] where represents the true label ​Denoted as the predicted value, by minimizing the loss function, the model can effectively learn the potential relationship types between entities. This optimization process ensures the accurate modeling of relationship information in complex network structures, while enhancing the prediction ability and making the relationship representation more refined and accurate.

[0074] Effect of feature training: Further compress and optimize the redundant information in the spatio-temporal dynamic graph, improve the effectiveness of graph representation and the accuracy of retrieval, and provide more efficient data support for the dynamic monitoring, progress tracking and resource optimization of railway construction projects.

[0075] The above S3 includes: After lossless compression of the graph data, the volume and storage space of the data graph are significantly optimized, thus accelerating the overall speed of retrieval. The problem retrieval process for railway construction scenarios usually highly depends on the precise positioning of time and space. Therefore, the retrieval problems are classified into the following types to systematically process and improve the accuracy of queries:

[0076] 1. Time point problem:

[0077] The time point problem refers to the data query for a specific time point, usually used to query the construction status, material consumption or project progress at a certain moment. For example, query the initial support concrete consumption of a certain construction part on a specific date. Since the compressed graph retains multiple child node information and timestamps at each time node, the system can directly locate the node on the time chain and extract relevant information.

[0078] Implementation method: By traversing the time chain, quickly locate the graph slice at the target time point and retrieve the construction status and material usage data contained therein. For the graph compression structure, the nodes on the time chain of the system retain detailed spatio-temporal association information for each time point, greatly improving the efficiency of the retrieval operation.

[0079] 2. Time period problem:

[0080] The time period problem refers to the query of construction data within a certain time interval. This kind of retrieval is usually used to analyze the construction progress, resource consumption trend, etc. within a certain period. For example, analyze the initial support concrete consumption of a certain tunnel within one month. The time period retrieval can provide detailed information within the time span.

[0081] Implementation method: By selecting the start and end time points on the time chain, the system can quickly locate and retrieve all relevant graph slices within this time interval. In each slice, the detailed construction information at this time point can be extracted and aggregated for analysis to generate the overall situation of the time period.

[0082] 3. Space point problem:

[0083] Spatial point problems are queries targeting specific geographical locations, usually used to understand the specific construction status or resource usage of a certain construction site. For example: Analyze the quantity of initial support concrete used at the working point DK*69+627 of the main tunnel entrance. For such problems, the working point can be located by traversing on the spatial chain. However, by using the construction log as a medium, the problem can be mapped onto the time chain, usually corresponding to one or more consecutive slices of the time chain graph. Therefore, the problem is respectively transformed into problems (1) or (2) for retrieval, greatly improving the retrieval efficiency and accuracy.

[0084] Implementation method: During the graph compression process, the spatial chain retains the spatial location and construction status information of each working point. By traversing the spatial chain and finding the location of the target working point, the system can quickly locate the relevant graph slices, and then combine the construction log to locate the time, thus achieving the dual mapping of time and space.

[0085] 4. Spatial segment problems:

[0086] Spatial segment problems are queries targeting a specific mileage range. For example: Analyze the quantity of concrete used for the full-section initial support within the mileage range of DK*69+593 - DK*69+597 at the working point of the main tunnel entrance. This type of query involves multiple time node information at multiple spatial positions. Therefore, it can be transformed into a many-to-many mapping problem and retrieved by traversing the spatial and time chains simultaneously.

[0087] Implementation method: For spatial segment queries, first locate the starting and ending spatial points, and mark all working points within the range through the spatial chain. Next, identify the corresponding time periods of each working point on the time chain, obtain the time slices within this spatial range, and perform an aggregation operation on multiple slices to generate the overall analysis data of the construction situation within this mileage range.

[0088] Effect of efficient retrieval: By introducing an efficient retrieval strategy, fast and accurate multi-dimensional queries can be achieved on a highly compressed spatio-temporal dynamic graph, greatly improving the management and monitoring capabilities of railway construction projects.

[0089] The above S4 includes: So far, the spatio-temporal atlas is still static and applicable to archived construction data. However, the dynamic nature of railway construction projects leads to continuous update and evolution of data over time, making the static atlas unable to meet the need for continuously updated construction data in long-term ongoing projects. Therefore, in order to enable the system to process dynamic data and capture the atlas information that changes over time during the construction process, a Temporal Graph Neural Network (Temporal GNN) is introduced to achieve dynamic modeling of the spatio-temporal atlas. Temporal GNN can efficiently capture the evolving characteristics of entities and relationships over time, support efficient modeling of the time dimension, thereby constructing a time-sensitive knowledge graph to ensure that the system can always reflect the latest state of the construction site. Specifically, it includes the following steps:

[0090] S4.1, Temporal Modeling Requirement Analysis:

[0091] During the railway construction process, the time-dependence of data is reflected in multiple aspects: Real-time changes in construction status: For example, the progress of the support structure, the usage of materials, etc. will change continuously over time. Continuous introduction of new entities: The addition of new construction parts or engineering projects requires the system to dynamically expand the atlas. Relationship updates: As the construction progresses, the process relationships, construction sequences, and time-dependent relationships also change. Therefore, Temporal GNN captures the changes of nodes and edges over time through hierarchical temporal learning, supports dynamic updates of the atlas, and maintains the consistency between the atlas and the construction site.

[0092] S4.2, Dynamic Feature Modeling Process of Temporal GNN:

[0093] Node Time Embedding: For each node (such as construction parts, material types, etc.), the system introduces a time embedding vector to represent the time sensitivity of the node. The time embedding vector is generated based on the historical feature sequence of the node and is iteratively updated over time. Through time embedding, the model can reflect the time-varying characteristics of the node in the atlas.

[0094] Among them, is the feature of the node at time , is the adjacency matrix, is the weight matrix at the current time, represents the aggregation process of the historical features of the node, is the historical window length.

[0095] Temporal update of edge weights: Using a time-sensitive edge weight update mechanism, the edge weight values are dynamically adjusted to reflect the importance of the relationship between nodes over time. By incorporating time information into the edge calculation, Temporal GNN can embed time factors into the edge weights. For example, let the change of edge weight over time be:

[0096]

[0097] where, (t) represents time t, is the time of node and node between the edge weights, T represents the attention mechanism parameters, j represents the input feature vector of the j-th node, i represents the input feature vector of the i-th node.

[0098] State transition and update mechanism: To ensure the continuity and consistency of the graph over time, Temporal GNN adopts a state transition mechanism to periodically insert new node and edge information into the graph. The information of new nodes is generated based on historical similar nodes and time embeddings, enabling the newly added nodes to quickly find their appropriate positions in the model. At the same time, the features of nodes and edges are retrained according to the actual construction requirements to ensure that the model always reflects the real-time state of the construction process.

[0099] S4.3, Optimization of the dynamic model:

[0100] To achieve real-time and effective management of dynamic data during the construction process, the Temporal GNN model incorporates various optimization measures in its design to improve the model's performance and computational efficiency in a dynamic environment. Specifically, it includes time window adjustment, weight adaptive update, and data incremental update.

[0101] 1. Dynamic adjustment of the time window:

[0102] Definition: The time window length is an important parameter in Temporal GNN, which determines the amount of historical information used by the model in each state update. Different time window lengths have different impacts on real-time performance and computational efficiency.

[0103] Function: In practical applications, the spatio-temporal dynamic characteristics of railway construction projects change with the construction stage. For example, at key construction nodes (such as tunnel excavation or the installation of support structures), the demand for real-time monitoring is high, and at this time, the time window can be set shorter to capture construction changes more promptly; while in the regular construction stage, appropriately lengthening the time window can aggregate more historical data and improve the stability of the model.

[0104] Adjustment strategy:

[0105] Short time window: Suitable for key node monitoring and real-time response. Through a small time span, the system can more quickly reflect the state changes within a short time.

[0106] Long time window: Suitable for global monitoring, especially for macro tasks such as analyzing material consumption trends and progress prediction. The long time window can reduce the model update frequency and improve the calculation efficiency.

[0107] Through a flexible time window adjustment mechanism, Temporal GNN can achieve a balance between real-time performance and efficiency at different construction stages.

[0108] 2. Weight adaptive update:

[0109] Definition: Weight adaptive update refers to that in Temporal GNN, the model can dynamically adjust the weights of edges and nodes according to changes in the time dimension to make them more in line with the characteristics of time series data.

[0110] Function: In the railway construction drawing spectrum, the relationship strengths of different nodes and edges change over time. For example, the relationship of certain construction parts may be particularly important at a specific stage, and the weight of this relationship will gradually weaken after the construction stage ends. Therefore, the model needs to update the weights of edges and nodes based on changes in the time dimension through an adaptive mechanism to reflect the importance and time sensitivity of the relationship.

[0111] Update method:

[0112] Time decay mechanism: Apply time decay to outdated information or data with a long history, gradually reducing the weights of these nodes and edges, so that the model can focus on recent data.

[0113] Adaptive learning: The model automatically adjusts the weights according to the time dimension during training, so that nodes and edges with a more recent time occupy a greater proportion in the prediction.

[0114] Through the adaptive update of weights, the model can capture important changes in nodes and relationships in the time dimension, fully reflecting the time-sensitive characteristics of the construction process.

[0115] 3. Data incremental local update:

[0116] Definition: Incremental update means that every time the data is updated, the system only trains on the newly added nodes and edges, without having to recalculate the entire graph. This update method is applicable to scenarios where construction data is continuously increasing and can effectively reduce the calculation cost.

[0117] Function: During the construction process, the system needs to frequently update the newly added nodes (such as newly built construction parts) or edges (such as newly added material consumption records) in the graph. However, if the entire graph is retrained every time the data is updated, it will lead to excessive calculation overhead and low efficiency. Therefore, the incremental update mechanism can significantly improve the update efficiency of the graph, enabling the system to quickly adapt to changes in new data.

[0118] Implementation method:

[0119] Local training: In incremental update, only the newly added nodes, edges, and their adjacent nodes are trained, enabling new data to quickly integrate into the existing graph without affecting the global structure.

[0120] Dynamic graph merging: For the local graph after each incremental update, the system will perform global merging after a period of time to ensure the structural integrity and consistency of the graph.

[0121] Through data incremental update, the system can efficiently process newly added spatio-temporal data, avoid repeated calculations, and significantly improve the adaptability to dynamic data and calculation efficiency.

[0122] S4.4 Spatio-temporal mapping strategy:

[0123] To simplify the tracking of spatial changes and avoid frequent updates of the spatial chain, a spatio-temporal mapping strategy is introduced to integrate spatial changes into the time graph, thereby centralizing all data updates for management in the time dimension.

[0124] Implementation mechanism:

[0125] Recording spatial changes in the construction log: The construction log usually records the operation activities, status updates, and resource usage of each construction part in chronological order. By systematically parsing the construction log, spatial changes can be recorded as time series data. The status updates of these spatial positions will be automatically mapped to the corresponding time nodes.

[0126] Mapping and Updating of the Temporal-Spatial Map: Whenever new construction log data is input, the system will parse and map the spatial information involved in the log (such as construction parts or specific locations) to specific time points on the temporal-spatial map. By adding status updates of spatial locations to the slices of the temporal-spatial map, the system can dynamically expand the features of each time node, enabling the representation of spatial changes through the update of the temporal-spatial map. For example, if the construction log records that a shotcrete operation was completed at location "DK*69+627" on "May 1, 2024", this operation will be mapped to the slice of the temporal-spatial map for "May 1, 2024", enabling the system to comprehensively track the construction status from a temporal perspective.

[0127] As Figure 2 shown, according to another embodiment of the present invention, a spatio-temporal dynamic map construction system 200 for engineering construction projects is also disclosed, including: a memory 201 for storing program instructions; a processor 202 for calling the program instructions stored in the memory to implement the spatio-temporal dynamic map construction method for engineering construction projects as described in the above embodiments.

[0128] All or part of the steps in the various methods of the above embodiments can be completed by a program controlling related hardware. The program can be stored in a readable storage medium, which includes read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically-erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM), or other optical disc memories, magnetic disk memories, tape memories, or any other readable medium capable of carrying or storing data.

[0129] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention can have various changes and modifications. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for constructing a spatiotemporal dynamic graph of an engineering construction project, characterized in that: include: Extract construction data entity set: extract construction data entities from construction process data documents and generate entity sets; Constructing a data graph: adding entities in the entity set as nodes to the graph and configuring attributes for each node; adding relationship edges to each node according to the logical structure between the construction data; adding time labels and space labels to each node, thereby forming a data graph; Attaching the time chain and the space chain: sorting the nodes in time according to the time information of the construction process to form a chain in the time dimension; classifying the nodes according to the construction location or mileage range to form a chain in the space dimension; using the time label and the space label, associating each entity in the data map with the time chain and the space chain to form a data map attached to the time and space chain; Graph compression: Calculate the similarity between nodes, identify similar nodes according to the set similarity threshold for node merging, and the node merging strategy includes: aggregating similar nodes according to their time attributes or spatial attributes, and aggregating similar nodes as child nodes into the same parent node; if the merged child nodes have differences in some attributes, then combine the difference attributes into multi-valued attributes; if the merged child nodes have multiple edges with the same node, then retain the weight information of multiple edges; optimize the weights of nodes and edges in the data graph based on the graph neural network; Efficient retrieval strategy: by traversing the time chain, locating the data map slice at the target time point, and retrieving the construction status and material usage data contained in the data map slice; or by selecting the start and end time points on the time chain, locating and retrieving all relevant data map slices within the time interval, and performing aggregation analysis to generate the overall situation within the target time period; or by traversing the space chain and finding the target work point location, locating the relevant data map slice, and then combining the construction log to locate the time, thereby converting the space point problem into a time point problem for retrieval; or setting the start point and end point on the space chain, marking all work points within the range, identifying the corresponding time period of each work point on the time chain, obtaining the time slice within the space range, and performing aggregation operations on multiple slices to generate overall analysis data of the construction situation within the work point interval; Constructing a dynamic graph: Based on the time-series graph neural network, a time embedding vector is introduced for each node in the data graph. The time embedding vector is generated according to the historical feature sequence of the node and is iteratively updated over time. In the convolution process of each layer of the time-series graph neural network, the time series aggregation mechanism is used to fuse the current state of the node with the information of the historical time point. The time-sensitive edge weight update mechanism is used to dynamically adjust the weight value of the edge. The time-series graph neural network adopts a state transfer mechanism to regularly insert new nodes and edges into the graph. The information of the new node is generated based on historical similar nodes and time embedding vectors.

2. The method for constructing a spatiotemporal dynamic graph of an engineering construction project according to claim 1, characterized in that: Before the step of extracting the construction data entity set, it also includes: Data preprocessing: Unified standardization and structuring of data documents generated during the construction process.

3. The method for constructing a spatiotemporal dynamic graph of an engineering construction project according to claim 1, characterized in that: After the step of building the dynamic graph, it also includes: Dynamic graph optimization: The time-series graph neural network is optimized by means of time window adjustment, weight adaptive update and data incremental update to improve the performance and computational efficiency of the time-series graph neural network in a dynamic environment.

4. The method for constructing a spatiotemporal dynamic graph of an engineering construction project according to claim 3, characterized in that: Also includes: By parsing the construction log, the spatial changes are recorded as time series data; the status updates of the spatial locations are mapped to the corresponding time nodes; Whenever new construction log data is input, the spatial information involved in the log is parsed and mapped to a specific time point in the time map. By adding the state update of the spatial position to the slice of the time map, the characteristics of each time node are dynamically expanded, so that the spatial changes can be represented by the update of the time map.

5. The method for constructing a spatiotemporal dynamic graph of an engineering construction project according to any one of claims 1 to 4, characterized in that: The construction process data documents include: production notice, batching notice, ecological ledger, mix ratio report, smooth blasting self-evaluation form, tunnel visualization surrounding rock list, process description, concrete maintenance record form, quality and safety inspection notice and three-dimensional section scanning data.

6. The method for constructing a spatiotemporal dynamic graph of an engineering construction project according to any one of claims 1 to 4, characterized in that: The construction data entities include: construction site, concrete, surrounding rock, over-excavation volume, over-limit volume, progress index, face excavation, initial support, initial support of invert, invert filling, invert lining, lining arch wall, forecast conclusion, initial support concrete, initial support concrete of invert, invert filling concrete, invert lining concrete, lining arch wall concrete, actual usage, disk release, excess consumption, drilling, charging and blasting, slag discharge, erection and mesh hanging, locking feet and anchor rods, inspection report, shotcrete, and cross-section.

7. The spatiotemporal dynamic graph construction system for engineering construction projects is characterized by: include: A memory for storing program instructions; A processor is used to call the program instructions stored in the memory to implement the method for constructing a spatiotemporal dynamic graph of an engineering construction project as described in any one of claims 1 to 6.

Images