Water conservancy project construction progress intelligent monitoring method and system combining BIM and AI

Abstract

The application provides a kind of water conservancy construction progress intelligent monitoring method and system combined with BIM and AI, it is related to water conservancy construction progress monitoring technical field, first obtain the BIM construction scene model of water conservancy and the field multi-source perception data of construction area, then the dynamic correlation processing of both, establish data mapping relationship, obtain dynamic correlation BIM model;Then call construction progress deduction AI model to process process progress correlation deduction to dynamic correlation BIM model, obtain process progress deduction result;Again, the progress difference is attributed to analysis, and the progress deviation attribution result is obtained;Finally, based on the progress deviation attribution result, adjust the process plan parameter in BIM construction scene model, generate construction progress adjustment instruction.The application realizes the intelligent monitoring and dynamic optimization of water conservancy construction progress, improves the monitoring efficiency and accuracy.

Description

Technical Field

[0001] This invention relates to the field of water conservancy project construction progress monitoring technology, and more specifically, to a method and system for intelligent monitoring of water conservancy project construction progress that combines BIM and AI. Background Technology

[0002] In the field of water conservancy engineering construction, effective monitoring and management of construction progress is a crucial link in ensuring the timely and high-quality completion of projects. Traditional methods for monitoring the construction progress of water conservancy projects mainly rely on regular manual inspections and experience-based judgment. This involves recording information such as the operational status of construction equipment and the progress of work processes on-site, and then comparing and analyzing this information with a pre-established construction plan. However, these methods have many limitations. On the one hand, manual inspections and recording not only consume a significant amount of manpower and time but are also easily affected by subjective factors, making it difficult to guarantee the accuracy and timeliness of data recording. On the other hand, traditional methods struggle to comprehensively and intuitively present the spatial characteristics of the water conservancy engineering construction scene and the complex hierarchical relationships between various work processes. Furthermore, they are unable to quickly and accurately analyze the causes and scope of impact of progress deviations that occur during construction, thus hindering timely and effective adjustments to the construction plan and easily leading to project delays and resource waste. With the continuous expansion of the scale and increasing complexity of water conservancy projects, traditional methods are no longer sufficient to meet the needs of modern water conservancy engineering construction progress monitoring. Summary of the Invention

[0003] In view of the aforementioned problems, and in conjunction with the first aspect of the present invention, embodiments of the present invention provide a method for intelligent monitoring of the construction progress of water conservancy projects combining BIM and AI, the method comprising: Acquire BIM construction scene models and on-site multi-source perception data of the construction area for water conservancy projects. The BIM construction scene model includes the hierarchical relationship of the work process, the work process planning cycle, and the spatial characteristics of the construction scene. The on-site multi-source perception data includes construction equipment operation information, work process execution status data, and scene environment data. The BIM construction scene model is dynamically associated with the on-site multi-source perception data. A data mapping relationship is established based on the spatial characteristics of the construction scene to obtain a dynamically associated BIM model that includes real-time process status annotations. The construction progress simulation AI model is invoked to perform process progress correlation simulation on the dynamically associated BIM model. Based on the process hierarchy, the progress impact link of each process is simulated, and the process progress simulation results including the difference between the actual progress and the planned progress of the process are obtained. By combining the process association information in the dynamic BIM model, the cause analysis of the schedule difference in the process schedule projection results is carried out to obtain the schedule deviation attribution results including the causes of the difference and the scope of the impact. Based on the attribution results of schedule deviations, the process planning parameters in the BIM construction scenario model are adjusted to generate construction schedule adjustment instructions that include the adjusted process cycle and equipment allocation path.

[0004] Furthermore, embodiments of the present invention also provide an intelligent monitoring system for the construction progress of water conservancy projects that combines BIM and AI, including: A processor; a machine-readable storage medium for storing machine-executable instructions of the processor; wherein the processor is configured to execute the above-described intelligent monitoring method for the construction progress of water conservancy projects combining BIM and AI by executing the machine-executable instructions.

[0005] In another aspect, embodiments of the present invention also provide a computer program product, the computer program product including machine-executable instructions, the machine-executable instructions being stored in a computer-readable storage medium, the processor of a computer device reading the machine-executable instructions from the computer-readable storage medium, the processor executing the machine-executable instructions, causing the computer device to execute the above-described intelligent monitoring method for the construction progress of water conservancy projects combining BIM and AI.

[0006] Based on the above, by acquiring the BIM construction scene model of the water conservancy project and the on-site multi-source perception data of the construction area, the two are dynamically correlated to construct a dynamically correlated BIM model containing real-time process status annotations. This model can present the real-time status of the water conservancy project construction scene and the complex relationships between various processes. Then, the construction progress prediction AI model is called to perform process progress correlation prediction on the dynamically correlated BIM model. Based on the process hierarchy relationship, the progress impact link of each process is predicted, and the process progress prediction results containing the difference between the actual progress and the planned progress of the process are obtained. Combining the process correlation information in the dynamically correlated BIM model, the progress difference in the process progress prediction results is analyzed for attribution. This can quickly and accurately determine the cause and scope of the difference. Based on the progress deviation attribution results, the process planning parameters in the BIM construction scene model are adjusted, and construction progress adjustment instructions containing the adjusted process cycle and equipment allocation path are generated. This realizes the dynamic optimization of construction progress and the rational allocation of resources, greatly improving the intelligence level and efficiency of water conservancy project construction progress monitoring, and effectively reducing the risk of project delays. Attached Figure Description

[0007] Figure 1 This is a schematic diagram of the execution flow of the intelligent monitoring method for water conservancy project construction progress that combines BIM and AI, provided in an embodiment of the present invention.

[0008] Figure 2 This is a schematic diagram of exemplary hardware and software components of the intelligent monitoring system for water conservancy project construction progress that combines BIM and AI, provided in an embodiment of the present invention. Detailed Implementation

[0009] The present invention will now be described in detail with reference to the accompanying drawings. Figure 1 This is a flowchart illustrating an embodiment of the intelligent monitoring method for the construction progress of water conservancy projects combining BIM and AI. The following is a detailed description of this intelligent monitoring method for the construction progress of water conservancy projects combining BIM and AI.

[0010] Step S110: Obtain the BIM construction scene model of the water conservancy project and the on-site multi-source perception data of the construction area. The BIM construction scene model includes the hierarchical relationship of the work process, the work process planning cycle and the spatial characteristics of the construction scene. The on-site multi-source perception data includes the operation information of construction equipment, the execution status data of the work process and the scene environment data.

[0011] In this embodiment, taking the construction of a water conservancy project as an example, the BIM construction scene model is constructed using professional BIM modeling software. This BIM construction scene model defines in detail the hierarchical relationship of each construction process in the water conservancy project. For example, the foundation excavation process is a prerequisite for the concrete pouring process. Each process has a corresponding planned start time and planned end time, and this time information constitutes the process planning cycle. At the same time, the BIM construction scene model also includes spatial features of the construction scene, such as the construction area being divided into different work zones, each zone having clear spatial boundaries, as well as information such as the working space range of construction equipment and the work space requirements of each process. The acquisition of multi-source sensing data on-site is achieved through various sensing devices deployed at the construction site. Construction equipment operation information can be collected through sensors installed on the construction equipment, such as the excavator's operating time, operating location coordinates, and fuel consumption during operation. Process execution status data can be obtained through monitoring cameras and sensors set up at the construction site, such as whether the concrete pouring process is currently in the vibration or curing stage, and the duration of the process. Scene environment data is collected through environmental monitoring equipment deployed at the construction site, such as temperature and humidity, light intensity, precipitation, and terrain change data in the work area. All of the above data will be uniformly transmitted to the data acquisition platform for preliminary aggregation and storage.

[0012] Step S120: Dynamically associate the BIM construction scene model with the on-site multi-source perception data, establish a data mapping relationship based on the spatial characteristics of the construction scene, and obtain a dynamically associated BIM model containing real-time process status annotations.

[0013] Step S121: Analyze the hierarchical structure of the BIM construction scene model, extract the process node set, process spatial distribution information and construction scene spatial features from the BIM construction scene model. The process node set includes the unique process code, planned start time and planned end time of each process. The construction scene spatial features include the construction area division boundary, equipment operation space range and process operation space requirements.

[0014] In this embodiment, a specialized BIM model parsing tool is used to analyze the hierarchical structure of the BIM construction scenario model of the aforementioned water conservancy project. First, all process nodes are extracted from the BIM model, forming a process node set. Each process node is assigned a unique process code; for example, the code for the foundation excavation process is GC001, and the code for the concrete pouring process is GC002, etc. Each process node also includes its corresponding planned start and end times. Then, the spatial distribution information of the processes is extracted, that is, the spatial location and distribution range of each process in the BIM model. For example, the three-dimensional coordinate range of the foundation excavation process's work area in the BIM model is from (x1, y1, z1) to (x2, y2, z2). Next, the spatial features of the construction scene are extracted. The construction area division boundary refers to dividing the entire water conservancy project construction area into different work areas. Each work area has a clear spatial boundary. For example, the boundary coordinates of work area A are (x3, y3, z3) to (x4, y4, z4). The equipment operation space range refers to the space required by different types of construction equipment when they are operating. For example, the operating radius range of a crane is defined in the BIM model as a spherical space with radius R centered on the equipment location. The process operation space requirement refers to the space required by each process when it is operating. For example, the operation space required by the concrete pouring process is a cuboid area with length, width, and height of L, W, and H, respectively.

[0015] Step S122: Classify and analyze the multi-source sensing data on site, dividing the multi-source sensing data on site into construction equipment operation information, process execution status data and scene environment data. Among them, the construction equipment operation information includes equipment code, operation location coordinates and operation duration, the process execution status data includes process code, current operation stage and operation duration, and scene environment data includes temperature and humidity, lighting conditions and terrain change data of the operation area.

[0016] In this embodiment, the multi-source sensing data collected from the construction site of the aforementioned water conservancy project is processed using data classification and analysis software. Based on the data's source and content, it is divided into three categories. For construction equipment operation information, different construction equipment is distinguished by identifying equipment codes. For example, the excavator with equipment code SB001 has its operating position coordinates obtained through its GPS positioning module, and its operating time is collected through the equipment's work timing system. For process execution status data, the corresponding process is associated with the process code. For example, the foundation excavation process with process code GC001 has its current operating stage determined by analyzing video images captured by monitoring cameras or sensor data, and its operating time is calculated using timestamps. For scene environment data, it is directly extracted from the data collected by environmental monitoring equipment. For example, the temperature and humidity of the work area are collected by temperature and humidity sensors, lighting conditions are collected by lighting sensors, and terrain change data is collected by terrain monitoring equipment. All of these data have corresponding collection timestamps to ensure data temporal consistency.

[0017] Step S123: Construct data association rules based on the spatial characteristics of the construction scene. Using the work process space requirements as a benchmark, match the equipment operation space range with the work process space requirements to establish the spatial association relationship between equipment codes and work process codes. At the same time, match the collection location coordinates of scene environment data with the work process space distribution information to establish the location association relationship between environmental data and work process codes.

[0018] In this embodiment, data association rules are constructed based on the extracted spatial features of the construction scene. Using the work process space requirements as a benchmark, the working space range of the construction equipment is checked to see if it overlaps with or contains the work process space requirements. For example, for the foundation excavation work process with work process code GC001, its working space requirement is a cuboid area, while the working space range of the excavator with equipment code SB001 is a spherical area centered on the equipment location. If this spherical area overlaps with the working space requirement area of ​​the foundation excavation work process, then the excavator is considered to have participated in the foundation excavation work, thus establishing a spatial association between equipment code SB001 and work process code GC001. For scene environment data, its collection location coordinates are compared with the work process spatial distribution information. If the collection location coordinates of the scene environment data fall within the spatial distribution range of a certain work process, then a positional association between the scene environment data and the corresponding work process code is established. For example, if the collection location coordinates of a temperature and humidity sensor fall within the spatial distribution range of the concrete pouring work process with work process code GC002, then the temperature and humidity data is associated with work process code GC002.

[0019] Step S124: Based on the data association rules, the process execution status data is directly matched to the corresponding process node in the BIM construction scene model through the process code, the construction equipment operation information is matched to the corresponding process node through the spatial association relationship between the equipment code and the process code, and the scene environment data is matched to the corresponding process node through the location association relationship, thus forming a preliminary associated data set.

[0020] In this embodiment, the multi-source sensing data on site is matched according to the established data association rules. For process execution status data, since it contains process codes, it can be directly matched to the corresponding process node in the BIM construction scene model based on the process code. For example, the current operation stage and duration of the foundation excavation process with process code GC001 in the process execution status data will be directly matched to the corresponding foundation excavation process node in the BIM model. For construction equipment operation information, the equipment's operating location coordinates, operation duration, and other information are matched to the corresponding process node through the spatial association relationship between the equipment code and the process code. For example, the operation information of the excavator with equipment code SB001 will be matched to the foundation excavation process node with process code GC001, which has a spatial association relationship with it. For scene environment data, temperature, humidity, lighting conditions, and other data are matched to the corresponding process nodes through location association relationships. For example, temperature and humidity data collected at locations falling within the spatial distribution range of a concrete pouring process with process code GC002 will be matched to that concrete pouring process node, thus forming a preliminary associated data set. The data in this associated data set are all associated with the process nodes in the BIM model.

[0021] Step S125: Based on the timestamps of the multi-source sensing data collected on site, perform time synchronization processing on the preliminary associated data set to ensure that the equipment information, status information and environmental information associated with the same process node have a unified time reference, and generate the time-synchronized associated data set.

[0022] In this embodiment, due to potential time discrepancies in the data acquisition devices for multi-source sensing on-site, time synchronization processing is required for the initial associated data set. First, the acquisition timestamp of each multi-source sensing data point is obtained. Then, the timestamps of the data are calibrated using the planned time of the process in the BIM construction scenario model or the time of a unified time server as the benchmark. For example, the acquisition timestamp of construction equipment operation information may deviate from the acquisition timestamp of process execution status data. A time synchronization algorithm adjusts the timestamps of these data to the same time benchmark. In this way, the equipment information, status information, and environmental information associated with the same process node have a unified time benchmark, generating a time-synchronized associated data set.

[0023] Step S126: Embed the time-synchronized associated data set into the corresponding process node attributes of the BIM construction scene model in the form of real-time annotations. The real-time annotations include data type identifiers, collection timestamps, and specific data content, resulting in a dynamically associated BIM model containing real-time process status annotations.

[0024] In this embodiment, a BIM model editing tool is used to embed the time-synchronized associated data set into the corresponding process node attributes of the BIM construction scene model in the form of real-time annotations. For example, for the foundation excavation process node with process code GC001, its attributes will embed the operation information of the excavator with equipment code SB001, including operation duration, operation location coordinates, etc. This information will be marked as construction equipment operation information type, and will also include the collection timestamp and specific operation data content; process execution status data, such as the current operation stage and the operation duration, will also be embedded into the process node attributes and marked as process execution status data type; scene environment data, such as the temperature, humidity, and lighting conditions of the operation area of ​​the process, will also be embedded and marked as scene environment data type. In this way, the BIM construction scene model becomes a dynamically associated BIM model containing real-time process status annotations. Users can intuitively understand the real-time execution status of the process by viewing the process node attributes in the BIM model.

[0025] Step S127: Set up a real-time update mechanism for the dynamic association BIM model. Set the update cycle of the dynamic association BIM model according to the collection frequency of multi-source sensing data on site. Repeat the data classification and parsing, association matching and time synchronization steps in each update cycle to replace the old real-time annotation information in the dynamic association BIM model so that the dynamic association BIM model reflects the latest construction site status.

[0026] In this embodiment, the update cycle of the dynamically associated BIM model is set according to the collection frequency of multi-source sensing data on site. For example, if the collection frequency of construction equipment operation information and process execution status data is once per minute, and the collection frequency of scene environment data is once every five minutes, then the update cycle can be set to five minutes after comprehensive consideration. Within each update cycle, steps S122 to S125 are repeated, that is, the newly collected multi-source sensing data on site is classified, parsed, matched, and time-synchronized. Then, the processed data is embedded into the corresponding process node attributes of the BIM model in the form of real-time annotations, replacing the old real-time annotation information, thereby ensuring that the dynamically associated BIM model can reflect the latest construction site status in a timely manner.

[0027] Step S128: Construct a data index for the dynamically associated BIM model based on the process code, bind all real-time annotation information associated with each process node to the process code, and quickly query the equipment, status and environmental data of the corresponding node through the process code, thereby improving the access efficiency of the dynamically associated BIM model data.

[0028] In this embodiment, to improve the access efficiency of dynamically associated BIM model data, a data index is constructed based on process codes. Each process node has a unique process code, and all real-time annotation information associated with that process node is bound to that process code. For example, for the basic excavation process node with process code GC001, its associated equipment operation information, process execution status data, and scene environment data are all bound to this process code. When a user needs to query the relevant data for this process, they only need to enter the process code GC001 to quickly obtain all the real-time annotation information of that process node, including equipment operation information, process execution status, and environmental data, without having to traverse and query the entire BIM model, greatly improving data access efficiency.

[0029] Step S130: Call the construction progress simulation AI model to perform process progress correlation simulation on the dynamic associated BIM model, and simulate the progress influence link of each process based on the process hierarchy relationship to obtain the process progress simulation result including the difference between the actual progress and the planned progress of the process.

[0030] Step S131: Extract structured data from the dynamically associated BIM model to form a model input dataset. The model input dataset includes the process code, planned start time, planned end time, current work stage, completed work duration, work duration and work location coordinates of associated equipment, and associated scene environment data for each process node.

[0031] In this embodiment, structured data is extracted from the dynamically associated BIM model of the aforementioned water conservancy project using a data extraction tool. For example, for each work process node, information such as its work process code, planned start time, planned end time, current work stage, and accumulated work time is extracted. Simultaneously, information such as the work time and work location coordinates of the equipment associated with that work process node, as well as associated scene environment data, such as temperature, humidity, and lighting conditions, are extracted. This data is then organized and formatted to form a model input dataset, where each record corresponds to relevant data for a work process node.

[0032] Step S132: Input the model input dataset into the scene adaptation layer of the construction progress simulation AI model. Through the scene adaptation layer, combined with the spatial features of the construction scene in the dynamically associated BIM model, the input data is preprocessed in a scene-based manner to generate scene-adapted input data.

[0033] In this embodiment, the scenario adaptation layer of the construction progress simulation AI model combines the spatial features of the construction scenario in the dynamically associated BIM model to perform scenario-based preprocessing on the model input dataset. For example, for the construction scenario of a water conservancy project, the scenario adaptation layer considers spatial features such as the spatial division of the construction area, the equipment operation space range, and the process operation space requirements. It performs spatial normalization processing on the operation position coordinates in the input data, converting them into relative coordinates relative to the overall coordinate system of the construction area. At the same time, according to different operation zones, it partitions and labels the equipment operation information and process execution status data so that subsequent feature extraction and analysis can better combine the spatial features of the construction scenario. After scenario-based preprocessing, scenario-adapted input data is generated. This scenario-adapted input data is more consistent with the scenario characteristics of water conservancy project construction, which can improve the accuracy of subsequent model processing.

[0034] Step S133: Input the scene-adapted input data into the process feature extraction layer of the construction progress simulation AI model. The process feature extraction layer adopts a parallel feature extraction structure. The time feature extraction unit extracts the time dimension features of the planned time parameters and the duration of work done for each process to generate a process time feature vector. In addition, the association feature extraction unit extracts the features of the association between the process and the equipment and the association between the process and the environmental data to generate a process association feature vector.

[0035] In this embodiment, the scene-adapted input data is fed into the process feature extraction layer of the construction progress simulation AI model. The process feature extraction layer employs a parallel feature extraction structure, where the time feature extraction unit extracts time-dimensional features from time parameters such as the planned start time, planned end time, and already completed duration for each process. For example, by analyzing the changing trends of planned time parameters and already completed duration, the time progress features of the process are extracted, such as whether it is proceeding as planned, and the degree of being ahead of schedule or behind schedule. These features are combined into a process time feature vector, which is a multi-dimensional vector containing various feature information of the process in the time dimension. The association feature extraction unit extracts features from the association between processes and equipment, and the association between processes and environmental data. For the relationship between processes and equipment, we will analyze the matching degree between the equipment's operating time, operating location coordinates and processes, as well as the impact of changes in the number of equipment on processes. For the relationship between processes and environmental data, we will analyze the degree of influence of environmental parameters such as temperature, humidity, and lighting conditions on process operations, as well as the potential impact of the changing trends of environmental parameters on process progress. The above features will be combined into a process association feature vector, which contains a variety of feature information about the relationship between processes, equipment, and environment.

[0036] Step S134: Input the process time feature vector and process association feature vector into the feature fusion unit of the construction progress prediction AI model. Based on the attention mechanism, analyze the contribution of the process time feature vector and process association feature vector to the process progress prediction, dynamically allocate feature fusion weights, and generate a fused feature vector through weighted summation. The fused feature vector reflects both the time attribute and association attribute of the process.

[0037] In this embodiment, the process time feature vector and the process association feature vector are input into the feature fusion unit of the construction progress prediction AI model. The feature fusion unit analyzes the contribution of these two vectors to the process progress prediction based on an attention mechanism. The attention mechanism dynamically allocates feature fusion weights according to the importance of different features in the vectors to the process progress prediction. For example, for a certain process, if its time progress feature has a greater impact on the progress prediction, then the weight of the process time feature vector will be allocated higher; if the process-equipment association feature has a greater impact on the progress prediction, then the weight of the features related to equipment in the process association feature vector will be allocated higher. Then, the process time feature vector and the process association feature vector are fused through a weighted summation operation to generate a fused feature vector, which reflects both the time attribute and the association attribute of the process.

[0038] Step S135: Extract the process hierarchy from the dynamic BIM model and construct a process association matrix. The process association matrix uses process codes as rows and columns. The elements of the process association matrix represent the dependency relationship between the corresponding row process and the corresponding column process. If the row process is the predecessor of the column process, the element value is 1, otherwise it is 0. The sequential execution logic between each process is presented through the process association matrix.

[0039] In this embodiment, the hierarchical relationships of construction processes are extracted from the dynamically associated BIM model, and then a process association matrix is ​​constructed. Taking the construction process of a water conservancy project as an example, the foundation excavation process with process code GC001 is a prerequisite process for the concrete pouring process with process code GC002. Therefore, in the process association matrix, the element with row process GC001 and column process GC002 has a value of 1, while the element with row process GC002 and column process GC001 has a value of 0. Through the above method, the dependencies between all processes are presented in matrix form, showing the sequential execution logic between each process.

[0040] Step S136: Input the fused feature vector and process correlation matrix into the correlation inference layer of the construction progress inference AI model, construct the process correlation inference network based on graph neural network, use the fused feature vector as the network node feature, use the process correlation matrix as the node connection relationship, and generate process progress influence link data by iterating through multiple rounds of iterations to infer the impact of the progress changes of each process on its related processes.

[0041] Step S1361: Construct the basic architecture of the process association inference network based on graph neural network. The basic architecture includes an input layer, a graph convolutional layer, a node update layer and an output layer. The input layer is used to receive the fused feature vector and the process association matrix. The graph convolutional layer is used to realize the neighborhood aggregation of node features. The node update layer is used to update the feature representation of each node. The output layer is used to output the process progress impact link data.

[0042] In this embodiment, the correlation inference layer of the construction progress prediction AI model is based on a graph neural network to construct the process correlation inference network. The input layer receives the fused feature vector and the process correlation matrix and passes the data to the graph convolutional layer. The graph convolutional layer determines the neighboring nodes of each process node based on the process correlation matrix, and then performs neighborhood aggregation on the node features. That is, it integrates the node's own fused feature vector with the fused feature vectors of its neighboring nodes to capture the feature interaction information between nodes. The node update layer updates the feature representation of each node based on the output of the graph convolutional layer and the progress change information of the process, so that the node features can reflect the latest state after the process progress changes. The output layer is responsible for converting the processed node features into process progress impact link data and outputting this data.

[0043] Step S1362: The fused feature vector is input into the input layer of the process association inference network. After feature standardization, it is distributed to each node of the process association inference network. Each node corresponds to a process code, and the initial feature value of the node is the fused feature vector corresponding to the process code.

[0044] In this embodiment, the fused feature vector is input into the input layer of the process association inference network, and then undergoes feature standardization to adjust the numerical range of the feature vector to a suitable interval, such as [0, 1]. The processed fused feature vector is then assigned to each node of the process association inference network. Each node corresponds to a process code, and the initial feature value of the node is the fused feature vector corresponding to that process code. For example, the initial feature value of the node with process code GC001 is the fused feature vector corresponding to that process, which includes feature information such as the time attribute and association attribute of the process.

[0045] Step S1363: After inputting the process association matrix into the input layer of the process association inference network, it is converted into the adjacency matrix of the process association inference network. The element values ​​in the adjacency matrix represent the connection strength between the corresponding nodes. For process nodes with dependencies, the connection strength is set according to the degree of dependency. The connection strength of the preceding process to the subsequent process is higher than the connection strength of the subsequent process to the preceding process.

[0046] In this embodiment, the process association matrix is ​​input into the input layer of the process association inference network and then converted into an adjacency matrix. The element values ​​in the adjacency matrix represent the connection strength between corresponding nodes. For process nodes with dependencies, the connection strength is set according to the tightness of the dependency. For example, the foundation excavation process with process code GC001 is a prerequisite process for the concrete pouring process with process code GC002, and the progress changes of the foundation excavation process have a significant impact on the concrete pouring process. Therefore, the connection strength between them will be set to a higher value, while the connection strength between the concrete pouring process and the foundation excavation process will be set to a lower value because the progress changes of the concrete pouring process have a relatively smaller impact on the foundation excavation process.

[0047] Step S1364: In the graph convolutional layer of the process association inference network, for each node, the set of neighboring nodes of the node is selected according to the adjacency matrix. The set of neighboring nodes consists of other process nodes that have a dependency relationship with the node. The node's own features and the features of neighboring nodes are aggregated by the graph convolution operator to generate neighborhood aggregated features. The aggregation operation is used to capture the feature interaction information between the node and its neighboring nodes.

[0048] In this embodiment, in the graph convolutional layer of the process association inference network, for each node, the set of neighboring nodes is selected based on the adjacency matrix. For example, for the concrete pouring process node with process code GC002, its set of neighboring nodes may include the foundation excavation process node with process code GC001 and the rebar tying process node with process code GC003 (assuming that the rebar tying process is a subsequent process of the concrete pouring process). Then, the graph convolution operator performs an aggregation operation on the fused feature vector of the node itself and the fused feature vectors of the neighboring nodes. The aggregation operation can be a weighted summation of the above feature vectors or other feature fusion methods to capture the feature interaction information between the node and its neighboring nodes and generate neighborhood aggregated features.

[0049] Step S1365: Input the neighborhood aggregation features into the node update layer of the process association inference network, perform nonlinear transformation through the activation function, and update the feature representation of the node by combining the change in the operation time of the corresponding process, so that the node features reflect the latest state after the change in process progress.

[0050] In this embodiment, neighborhood aggregation features are input into the node update layer of the process association inference network, and then nonlinearly transformed using an activation function, such as ReLU or Sigmoid. Simultaneously, the node's feature representation is updated by incorporating the change in the completed operation time of the corresponding process. For example, for the concrete pouring process node coded GC002, if its completed operation time has increased by a certain period, this change will be taken into account when updating the node's feature representation, ensuring that the node features reflect the latest state after the change in the process's progress, thus facilitating subsequent inference.

[0051] Step S1366: Set the number of iterations for the correlation deduction. The number of iterations is determined according to the total number of procedures in the water conservancy project. In each iteration, the neighbor node screening, feature aggregation and node update steps are repeatedly executed so that the features of each node are gradually transmitted to its associated nodes, realizing the chain deduction of the progress impact.

[0052] In this embodiment, the number of iterations for the correlation deduction is determined based on the total number of procedures in the water conservancy project. For example, if the total number of construction procedures in a water conservancy hub project is 50, then the number of iterations can be set to 10. During each iteration, steps S1364 and S1365, namely the neighbor node filtering, feature aggregation, and node update steps, are repeatedly executed, so that the features of each node are gradually transmitted to its associated nodes, realizing the chain deduction of the progress impact. For example, in the first iteration, the feature change of the foundation excavation procedure node with procedure code GC001 will be transmitted to its neighbor node, the concrete pouring procedure node with procedure code GC002; in the second iteration, the feature change of the concrete pouring procedure node with procedure code GC002 will be transmitted to its neighbor node, the rebar tying procedure node with procedure code GC003, and so on, until the set number of iterations is completed.

[0053] Step S1367: After each iteration, the feature change of each node is extracted through the output layer of the process association inference network. Nodes with feature change exceeding a preset threshold are marked as affected nodes. The feature change reflects the degree to which the process corresponding to the node is affected by the progress changes of other related processes.

[0054] In this embodiment, after each iteration, the feature change of each node is extracted through the output layer of the process association inference network. The feature change reflects the degree to which the process corresponding to the node is affected by the progress changes of other related processes. Then, nodes with feature changes exceeding a preset threshold are marked as affected nodes. For example, the preset threshold is 0.2, and the feature change of the concrete pouring process node with process code GC002 is 0.3, which exceeds the preset threshold, so this node will be marked as an affected node.

[0055] Step S1368: For each affected node, trace back the transmission path of its characteristic changes, determine the source process node and intermediate transmission nodes that affect it, and form a process progress influence link with the source process as the starting point and the affected node as the ending point. The process progress influence link includes the process code of each node and the order of influence transmission.

[0056] In this embodiment, for each affected node, the transmission path of its characteristic changes is traced back. For example, if the affected node is the rebar tying process node with process code GC003, by analyzing the transmission process of its characteristic changes, the source process node affecting it is determined to be the foundation excavation process node with process code GC001, and the intermediate transmission node is the concrete pouring process node with process code GC002. Then, a process progress influence link is formed, starting from the foundation excavation process node with process code GC001 and ending at the rebar tying process node with process code GC003. This process progress influence link includes the process code of each node and the order of influence transmission, i.e., GC001→GC002→GC003.

[0057] Step S1369: Deduplicate all process progress impact links to obtain the final process progress impact link data. Associate and store the process progress impact link data with the corresponding node feature changes so that each link contains impact degree information.

[0058] In this embodiment, all process progress impact links are deduplicated to remove duplicate links. For example, there may be multiple identical process progress impact links; after deduplication, the final process progress impact link data is obtained. Then, the process progress impact link data is associated and stored with the corresponding node feature changes, so that each link contains impact degree information. For example, the node feature changes corresponding to the process progress impact link GC001→GC002→GC003 are 0.3, 0.2, and 0.1, respectively. This information is associated and stored for subsequent analysis and processing.

[0059] Step S137: In the correlation simulation layer of the construction progress simulation AI model, for each process node, combined with its already completed work time and current work stage, and referring to the historical work cycle data of the same process in similar water conservancy projects, the actual completion time of the process is predicted. By comparing the actual completion time with the planned end time, the time difference between the actual progress and the planned progress of the process is calculated.

[0060] In this embodiment, within the correlation inference layer of the construction progress prediction AI model, for each process node, the actual completion time of the process is predicted by combining its accumulated work time and current work stage, and referencing historical work cycle data of similar processes in similar water conservancy projects. For example, for the concrete pouring process with process code GC002, its accumulated work time is T1, and its current work stage is the vibration stage. Referring to historical work cycle data of concrete pouring processes in similar water conservancy projects, this process also needs to go through a curing stage after the vibration stage, and the curing stage usually lasts for T2. Therefore, the predicted actual completion time of this process is the current time plus T2. Then, by comparing the actual completion time with the planned end time, the time difference value between the actual progress and the planned progress of the process is calculated. If the actual completion time is later than the planned end time, the time difference value is positive; otherwise, it is negative.

[0061] Step S138: Combining the process progress impact data and the time difference values ​​of each process, obtain the final process progress projection result that includes the difference between the actual and planned progress of the process.

[0062] In this embodiment, the process progress impact chain data and the time difference values ​​of each process are combined to obtain the final process progress projection result. For example, in the process progress impact chain GC001→GC002→GC003, the time difference value of the foundation excavation process with process code GC001 is +2 days (indicating that the actual progress is 2 days behind the planned progress), the time difference value of the concrete pouring process with process code GC002 is +1 day, and the time difference value of the rebar tying process with process code GC003 is +0.5 days. This information is integrated into the process progress projection result, clearly showing the difference between the actual progress and the planned progress of each process and the influence relationship between them.

[0063] Step S140: Combining the process association information in the dynamic BIM model, perform attribution analysis on the schedule differences in the process schedule projection results to obtain schedule deviation attribution results that include the causes of the differences and the scope of their impact.

[0064] Step S141: Extract the schedule difference data from the process schedule projection results, filter out the deviation processes with non-zero time difference values, and form a deviation process set. The deviation process set includes the process code, actual completion time prediction value, planned end time, and time difference value of the deviation process.

[0065] In this embodiment, progress difference data is extracted from the process progress projection results, and then deviation processes with non-zero time difference values ​​are filtered out to form a deviation process set. For example, in the process progress projection results, the time difference value of the foundation excavation process with process code GC001 is +2 days, the time difference value of the concrete pouring process with process code GC002 is +1 day, and the time difference value of the rebar tying process with process code GC003 is +0.5 days. These processes have non-zero time difference values, so they will be filtered out to form a deviation process set. This deviation process set includes the process code, actual completion time prediction value, planned end time, and time difference value of these deviation processes.

[0066] Step S142: Extract the association information corresponding to the deviation process from the dynamic association BIM model. The association information includes the associated construction equipment operation information, scene environment data and process association relationship. The construction equipment operation information includes the equipment operation time, operation parameters and operation location changes. The scene environment data includes the environmental parameter changes during the operation of the deviation process. The process association relationship includes the information of the preceding and subsequent processes of the deviation process.

[0067] In this embodiment, the associated information corresponding to the deviation process is extracted from the dynamically associated BIM model. Taking the concrete pouring process with process code GC002 as an example, its associated construction equipment operation information includes the operation time, operation parameters (such as pumping pressure, pumping volume, etc.) and operation location changes of the concrete pump truck participating in the process; the scene environment data includes the temperature and humidity changes, light intensity changes, precipitation conditions and terrain change data during the operation of the process; the process association relationship includes the information of the preceding process (the foundation excavation process with process code GC001) and the subsequent process (the rebar binding process with process code GC003).

[0068] Step S143: Construct a schedule deviation attribution index system. The schedule deviation attribution index system includes equipment impact index, environmental impact index, process-related impact index, and self-execution index. The equipment impact index is used to measure the impact of equipment operating status on process progress. The environmental impact index is used to measure the impact of changes in environmental parameters on process progress. The process-related impact index is used to measure the impact of the progress of the preceding process on the current deviation process. The self-execution index is used to measure the impact of the process's own execution efficiency on the schedule.

[0069] In this embodiment, a schedule deviation attribution index system is constructed. Equipment impact index measures the impact of equipment operating status on process progress by analyzing factors such as the operating time of construction equipment, the stability of operating parameters, and equipment failure records. Environmental impact index measures the impact of changes in environmental parameters such as temperature and humidity, light intensity, precipitation, and terrain during the operation period on process progress. Process correlation impact index measures the impact of the progress of preceding processes on the current deviation process by analyzing factors such as the time difference value of preceding processes and the dependence strength between preceding processes and the current deviation process. Self-execution index measures the impact of the process's own execution efficiency on progress by analyzing factors such as the ratio of the completed operation time of the process to the planned time of the current operation stage, and the difference rate between the process's execution efficiency and the average execution efficiency of similar processes.

[0070] Step S144: Input the correlation information of the deviation process into the schedule deviation attribution AI model. The schedule deviation attribution AI model includes an index calculation module and a cause identification module. The index calculation module quantifies the correlation information according to the schedule deviation attribution index system and calculates the specific values ​​of each index. The equipment impact index is calculated by matching the equipment operation time with the operation time of the process. The environmental impact index is obtained by calculating the relative deviation between the actual value and the standard value of each environmental parameter and weighting and summing according to the degree of impact.

[0071] Step S1441: After receiving the correlation information of the deviation process through the index calculation module, the correlation information is preprocessed to convert the unstructured environmental description data into structured environmental parameter values ​​and the equipment operation status description into equipment operation parameter values.

[0072] In this embodiment, after receiving the associated information of the deviation process, the index calculation module will perform data preprocessing on the above information. For example, for unstructured environmental description data, such as "the humidity in the work area is high", it will be converted into structured environmental parameter values, such as humidity as a specific percentage value; for equipment operation status descriptions, such as "the concrete pump truck is operating normally", it will be converted into equipment operation parameter values, such as pumping pressure as a specific value, pumping volume as a specific value, etc.

[0073] Step S1442: For the equipment impact index, extract the construction equipment operation information from the associated information. The construction equipment operation information includes equipment operation time, operation parameter stability and equipment failure records. Calculate the overlap between the equipment operation time and the operation time of the deviation process. At the same time, calculate the deviation rate between the equipment operation parameters and the equipment standard operation parameters. If there are equipment failure records, add the failure impact coefficient. The final value of the equipment impact index is the comprehensive calculation result of overlap, deviation rate and failure impact coefficient. The higher the overlap, the higher the equipment participation. The lower the deviation rate, the more stable the equipment operation status.

[0074] In this embodiment, for equipment impact indicators, construction equipment operation information is extracted from the associated information. Taking the concrete pouring process with process code GC002 as an example, the associated concrete pump truck's operation time is T3, and the deviation process's completed operation time is T4. The overlap between the equipment operation time and the deviation process's completed operation time is calculated. The overlap is calculated by dividing the intersection time of T3 and T4 by the total time of T4. Simultaneously, the deviation rate between equipment operation parameters such as pumping pressure and pumping volume and the equipment's standard operation parameters is calculated. The deviation rate is calculated as |actual operation parameter - standard operation parameter| / standard operation parameter. If the concrete pump truck has a fault record, such as a fault occurring once during operation, a fault impact coefficient is added. The magnitude of the fault impact coefficient is determined based on the severity and duration of the fault. The final value of the equipment impact indicator is the comprehensive calculation result of the overlap, deviation rate, and fault impact coefficient. The impact of equipment operation status on process progress is measured using the above methods.

[0075] Step S1443: For environmental impact indicators, extract scene environmental data from the associated information. Scene environmental data includes temperature and humidity changes, light intensity changes, precipitation, and terrain changes during the operation. Query the standard operating environment parameter range corresponding to the deviation process. For each environmental parameter, calculate the relative deviation between the actual value and the standard value. The relative deviation is calculated as |actual value - standard value| / standard value range width, where the standard value range width is the difference between the upper and lower limits of the standard parameter. For standard parameters without a range, use the standard value itself as the denominator to calculate the relative deviation. At the same time, set weights according to the degree of influence of environmental parameters on the operation process. The weight of precipitation is higher than that of temperature and humidity changes. The final value of the environmental impact indicator is the weighted sum of the relative deviations of each environmental parameter and their corresponding weights.

[0076] In this embodiment, scene environment data is extracted from the associated information for environmental impact indicators. Taking the concrete pouring process with process code GC002 as an example, the data on temperature and humidity changes, light intensity changes, precipitation, and terrain changes during the operation are extracted. The standard operating environment parameter range corresponding to this process is queried, for example, the standard temperature range is 20℃-30℃, and the standard humidity range is 40%-60%. For each environmental parameter, the relative deviation between the actual value and the standard value is calculated. For example, if the actual temperature is 25℃ and the standard value range is 20℃-30℃, then the standard value range width is 10℃, and the relative deviation is |25-25| / 10=0 (assuming the standard value is 25℃). If it is precipitation, the standard value is no precipitation, and the actual value is precipitation, then the relative deviation is calculated as |actual value - standard value| / standard value (here the standard value is 0, and the actual value is 1, indicating precipitation). Simultaneously, weights are assigned based on the degree of impact of environmental parameters on the operational process. Precipitation has a higher weight than temperature and humidity changes; for example, precipitation has a weight of 0.4, temperature and humidity changes have a weight of 0.2, light intensity changes have a weight of 0.2, and topographic changes have a weight of 0.2. The final value of the environmental impact index is the weighted sum of the relative deviations of each environmental parameter and their corresponding weights. This method is used to measure the impact of environmental parameter changes on the process progress.

[0077] Step S1444: For the process association impact index, extract the information of the preceding process from the association information, obtain the time difference value of the preceding process from the process progress projection results, and calculate the ratio of the time difference value of the preceding process to the time difference value of the deviation process. The larger the ratio, the greater the impact of the preceding process on the deviation process. At the same time, combine the dependency strength in the process association relationship. The higher the dependency strength, the greater the weight. The final value of the process association impact index is the product of the ratio and the dependency strength.

[0078] In this embodiment, for the process association impact index, the information of the preceding process is extracted from the association information. Taking the concrete pouring process with process code GC002 as an example, its preceding process is the foundation excavation process with process code GC001. From the process progress projection results, the time difference value of the preceding process is +2 days, and the time difference value of the deviation process is +1 day, and the calculated ratio is 2 / 1=2. At the same time, combined with the dependence strength in the process association relationship, assuming that the dependence strength between the foundation excavation process and the concrete pouring process is 0.8, then the final value of the process association impact index is 2*0.8=1.6. The impact of the progress of the preceding process on the current deviation process is measured in the above way.

[0079] Step S1445: For the self-performing indicators, extract the completed time, current stage and planned cycle of the deviation process, calculate the ratio of completed time to the planned time of the current stage. If the ratio is greater than 1, it means that the self-performing efficiency is lower than the plan. At the same time, refer to the performance efficiency data of similar processes under the same environmental and equipment conditions, calculate the difference rate between the performance efficiency of the deviation process and the average performance efficiency of similar processes. The final value of the self-performing indicators is the comprehensive result of the ratio and the difference rate.

[0080] In this embodiment, for the self-performance indicators, the accumulated work time, current work stage, and planned work cycle of the deviation process are extracted. Taking the concrete pouring process with process code GC002 as an example, its accumulated work time is T5, the current work stage is the vibration stage, and the planned time for the vibration stage in the planned work cycle is T6. The ratio is calculated as T5 / T6. If the ratio is greater than 1, it indicates that the self-performance efficiency is lower than the plan. At the same time, referring to the performance efficiency data of similar processes under the same environmental and equipment conditions, the difference rate between the performance efficiency of the deviation process and the average performance efficiency of similar processes is calculated. The difference rate is calculated as (performance efficiency of deviation process - average performance efficiency of similar processes) / average performance efficiency of similar processes. The final value of the self-performance indicator is the combined result of the ratio and the difference rate. The impact of the process's own performance efficiency on the progress is measured through the above method.

[0081] Step S1446: Normalize the calculated values ​​of each indicator and establish the correspondence between the indicator values ​​and the degree of influence. Organize the normalized values ​​of each indicator and the corresponding names of the influencing factors into an indicator quantification result table. Output the indicator quantification result table to the cause identification module as the input feature for deviation cause identification.

[0082] In this embodiment, the calculated values ​​of equipment impact indicators, environmental impact indicators, process-related impact indicators, and self-execution indicators are normalized, adjusting their value range to [0, 1]. Then, a correspondence between indicator values ​​and their degree of influence is established; for example, the closer the indicator value is to 1, the greater the impact of that influencing factor on schedule deviation. The normalized indicator values ​​and their corresponding influencing factor names are compiled into an indicator quantification result table. For example, the equipment impact indicator value is 0.6, the environmental impact indicator value is 0.3, the process-related impact indicator value is 0.8, and the self-execution indicator value is 0.4, with the influencing factor names being equipment operation problems, environmental change problems, impact of preceding process deviations, and self-execution efficiency problems, respectively. This indicator quantification result table is then output to the cause identification module as input features for deviation cause identification.

[0083] Step S145: The cause identification module uses a multi-class logistic regression algorithm, taking the calculated values ​​of each indicator as input features, and analyzes the contribution of each influencing factor through the trained model parameters to determine the cause of the schedule deviation. If the equipment influence indicator value is the largest, the cause is the equipment operation problem; if the environment influence indicator value is the largest, the cause is the environmental change problem; if the process association influence indicator value is the largest, the cause is the deviation of the preceding process; if the self-execution indicator value is the largest, the cause is the self-execution efficiency problem.

[0084] In this embodiment, the cause identification module employs a multi-class logistic regression algorithm, using the calculated values ​​of each indicator in the indicator quantification result table obtained in step S1446 as input features. This multi-class logistic regression model, with its trained parameters, is capable of analyzing the contribution of each influencing factor. For example, if the input features are a value of 0.6 for the equipment impact indicator, 0.3 for the environmental impact indicator, 0.8 for the process-related impact indicator, and 0.4 for the self-execution indicator, the model will analyze the contribution of the influencing factors corresponding to these values. Since the process-related impact indicator has the largest value, the cause of the schedule deviation is determined to be the deviation of the preceding process.

[0085] Step S146: Combine the process progress impact link data in the process progress simulation results to determine the impact range of the deviation process on subsequent processes. The impact range includes the code of the subsequent process affected by the deviation process, the expected time difference value of each subsequent process, and the impact transmission path.

[0086] In this embodiment, the impact range of the deviated process on subsequent processes is determined by combining the process progress impact chain data in the process progress simulation results. Taking the concrete pouring process with process code GC002 as an example, its process progress impact chain data is GC002→GC003→GC004 (assuming GC003 is the rebar tying process and GC004 is the formwork removal process). Then, the subsequent processes affected by this deviated process are coded as GC003 and GC004. The expected time difference value of each subsequent process is obtained from the process progress simulation results. For example, the expected time difference value of GC003 is +0.5 days, and the expected time difference value of GC004 is +0.3 days. The impact transmission path is GC002→GC003→GC004.

[0087] Step S147: Collect the process code, time difference value, cause of deviation, scope of impact and path of impact for the deviation process to form schedule deviation attribution data.

[0088] In this embodiment, relevant information on the deviation process is collected. Taking the concrete pouring process with process code GC002 as an example, its process code is GC002, the time difference value is +1 day, the cause of the deviation is the influence of the deviation of the preceding process, the scope of the influence is the subsequent processes GC003 and GC004, the expected time difference values ​​are +0.5 days and +0.3 days respectively, and the influence transmission path is GC002→GC003→GC004. The above information is collected to form schedule deviation attribution data.

[0089] Step S148: Organize the schedule deviation attribution data according to the execution order of the deviation process, establish an independent attribution entry for each deviation process, and obtain the schedule deviation attribution result containing the cause of the difference and the scope of influence. The attribution entry contains complete deviation information, cause analysis and description of the scope of influence.

[0090] In this embodiment, the schedule deviation attribution data is organized according to the execution order of the deviation processes. For example, if the execution order of the deviation processes is GC001, GC002, and GC003, then an independent attribution entry will be created for each deviation process in this order. Taking the concrete pouring process of GC002 as an example, its attribution entry includes complete deviation information, cause analysis, and description of the scope of influence, such as process code GC002, time difference value +1 day, the impact of the deviation on the preceding process, the scope of influence on subsequent processes GC003 and GC004, the expected time difference values ​​of +0.5 days and +0.3 days respectively, and the impact propagation path GC002→GC003→GC004. This yields a schedule deviation attribution result that includes the cause of the deviation and the scope of influence.

[0091] Step S149: Add an urgency indicator to each attribution item in the schedule deviation attribution results. The urgency is determined based on the importance and scope of the deviation process. The importance is extracted from the process hierarchy relationship in the dynamically associated BIM model. The larger the scope of influence, the higher the urgency.

[0092] In this embodiment, an urgency level indicator is added to each attribution item in the schedule deviation attribution results. Taking the concrete pouring process with process code GC002 as an example, its importance is extracted from the process hierarchy in the dynamically associated BIM model. Assuming this process is a critical process with high importance, and its influence scope includes two subsequent processes, its influence scope is large, so the urgency level indicator is high. By adding an appropriate urgency level indicator to each attribution item in this way, subsequent schedule adjustments can prioritize addressing deviation processes with high urgency.

[0093] Step S150: Adjust the process planning parameters in the BIM construction scenario model based on the schedule deviation attribution results, and generate a construction schedule adjustment instruction that includes the adjusted process cycle and equipment allocation path.

[0094] Step S151: Extract key information from the schedule deviation attribution results, sort the deviation processes in descending order of urgency level, and determine the priority order for schedule adjustment. The key information includes the deviation process code, deviation cause, scope of impact, urgency level indicator, and time difference value of each deviation process.

[0095] In this embodiment, key information is extracted from the schedule deviation attribution results. Taking the schedule deviation attribution results of a water conservancy project as an example, the deviation processes include GC001, GC002, GC003, etc. Their deviation process codes, deviation causes, scope of impact, urgency level indicators, and time difference values ​​are extracted. Then, the above deviation processes are sorted in descending order of urgency level indicators. For example, deviation process GC002, which has a high urgency level indicator, is placed first, and deviation process GC001, which has a medium urgency level indicator, is placed later, thereby determining the priority order for schedule adjustments.

[0096] Step S152: For each priority deviation process, formulate a targeted preliminary adjustment strategy based on the cause of the deviation, and input the preliminary adjustment strategy into the progress simulation module of the dynamic association BIM model. Based on the process association relationship and resource constraints of the BIM construction scenario model, simulate the process execution after the implementation of the adjustment strategy. In this process, update the resource configuration parameters in the dynamic association BIM model according to the resource adjustment suggestions in the adjustment strategy, and then recalculate the operation cycle of each process to predict the new planned start time and new planned end time of the deviation process and the affected process.

[0097] Step S1521: Analyze the resource adjustment parameters, time adjustment suggestions and process optimization schemes in the preliminary adjustment strategy, and extract basic data from the dynamic BIM model. The basic data includes the process relationship matrix, the standard operation cycle of each process, the total amount of construction resources and the spatial constraints of the construction scene. The total amount of construction resources includes the number of equipment, the number of personnel and the amount of material reserves. The spatial constraints of the construction scene include the maximum operation capacity of each work area.

[0098] In this embodiment, for the prioritized deviation process GC002, a preliminary adjustment strategy is formed based on the impact of deviations in preceding processes, such as increasing the number of construction equipment and adjusting the operation time. The resource adjustment parameters in this preliminary adjustment strategy are analyzed, such as the need to add a concrete pump truck; time adjustment suggestions, such as extending the operation time of the concrete pouring process; and process optimization schemes, such as optimizing the construction process of concrete pouring. Simultaneously, basic data is extracted from the dynamically linked BIM model, including a process relationship matrix showing the dependencies between processes; the standard operating cycle of each process, such as a standard operating cycle of 10 days for the concrete pouring process; total construction resource information, such as 10 concrete pump trucks, 200 personnel, and 1000 cubic meters of concrete reserves; and spatial constraints of the construction scene, such as the maximum operating capacity of each work area being able to accommodate 5 concrete pump trucks simultaneously.

[0099] Step S1522: Based on the resource adjustment parameters in the preliminary adjustment strategy, update the allocation status of construction resources. Based on the updated resource allocation status and the process association matrix, calculate the execution cycle of each process and determine the key process path of the water conservancy project. The key process path refers to the process sequence with the smallest total float.

[0100] In this embodiment, the allocation status of construction resources is updated according to the resource adjustment parameters in the initial adjustment strategy. For example, after adding a concrete pump truck, the number of equipment in the concrete pouring process becomes 2, and the updated resource allocation status is recorded. Then, based on the updated resource allocation status and the process association matrix, the execution cycle of each process is calculated. For example, due to the increase in the number of equipment and improved work efficiency, the execution cycle of the concrete pouring process is shortened to 8 days. By calculating the total float of each process, the critical process path of the water conservancy project is determined. The critical process path is the sequence of processes with the smallest total float, such as the sequence of processes consisting of foundation excavation → concrete pouring → rebar tying → formwork removal, which has the smallest total float and should be prioritized to ensure its progress.

[0101] Step S1523: For the deviation processes on the critical process path, recalculate their operation cycle based on the operation efficiency optimization suggestions in the adjustment strategy and the resource allocation.

[0102] In this embodiment, for the deviation process GC002 on the critical process path, the operation cycle is recalculated based on the work efficiency optimization suggestions in the adjustment strategy, such as optimizing the concrete pouring construction process, combined with resource allocation (equipment quantity of 2 concrete pump trucks, increase in personnel, etc.). For example, after optimizing the construction process, the work efficiency of the concrete pouring process is improved, and the operation cycle is shortened from the original 10 days to 8 days. The recalculated operation cycle will be recorded.

[0103] Step S1524: Calculate the new plan start time and new plan end time of each process in the order of process association. The new plan end time of the preceding process is used as the benchmark for the new plan start time of the subsequent process. If there are multiple preceding processes in the subsequent process, the new plan end time of the latest completed preceding process is used as its new plan start time.

[0104] In this embodiment, the new planned start time and new planned end time of each process are calculated sequentially according to the order of their relationships. Taking the critical process path of foundation excavation → concrete pouring → rebar tying → formwork removal as an example, the new planned start time of the foundation excavation process is the original planned start time, and the new planned end time is the original planned end time plus or minus the adjusted time difference; the new planned start time of the concrete pouring process is the new planned end time of the foundation excavation process, and the new planned end time is the new planned end time of the foundation excavation process plus the recalculated work cycle; the new planned start time of the rebar tying process is the new planned end time of the concrete pouring process, and the new planned end time is the new planned end time of the concrete pouring process plus its standard work cycle; the new planned start time of the formwork removal process is the new planned end time of the rebar tying process, and the new planned end time is the new planned end time of the rebar tying process plus its standard work cycle. If a subsequent process has multiple preceding processes, such as a process having two preceding processes, then the new planned end time of the latest completed preceding process is used as its new planned start time.

[0105] Step S1525: During the time calculation process, in conjunction with the spatial constraints of the construction scenario, check whether there are simultaneous operations of processes within the same work area. If so, adjust the new planned start time of one of the processes, thereby outputting the new planned start time, new planned end time, adjusted process cycle, and resource consumption details for each process. The resource consumption details include the number of equipment, personnel, and materials required for each process after the adjustment.

[0106] In this embodiment, during the time calculation process, the spatial constraints of the construction scenario are considered to check whether there are simultaneous operations within the same work area. For example, both the concrete pouring and rebar tying operations are performed in work area A, which has a maximum capacity of accommodating 5 concrete pump trucks simultaneously. The concrete pouring operation requires 2 concrete pump trucks, and the rebar tying operation requires 3 cranes, so there is no equipment conflict. However, if the operation times of the two operations overlap and the space in the work area is limited, it may affect construction efficiency. Therefore, it is necessary to adjust the new planned start time of one of the operations. Assuming the new planned start time for the concrete pouring operation is day 1 and the new planned end time is day 8, and the new planned start time for the rebar tying operation is day 8 and the new planned end time is day 12, this avoids simultaneous operations. Then output the new plan start time, new plan end time, adjusted process cycle and resource consumption details for each process. The resource consumption details include the number of equipment, personnel and materials required for each process after the adjustment. For example, the adjusted number of equipment required for the concrete pouring process is 2 concrete pump trucks, 50 personnel and 800 cubic meters of concrete.

[0107] Step S1526: Compare the time calculation results with the original planned time to generate a schedule adjustment simulation report. The schedule adjustment simulation report includes the differences in process time before and after the adjustment, changes in resource allocation, and possible conflict prompts.

[0108] In this embodiment, the calculated time is compared with the original planned time to generate a schedule adjustment simulation report. For example, the original planned start time for the concrete pouring process was day 1, and the original planned end time was day 10. The adjusted new planned start time is day 1, and the new planned end time is day 8, a time difference of 2 days. The resource allocation changes include an increase of 1 concrete pump truck, an increase of 10 personnel, and a decrease of 200 cubic meters of concrete in material usage. The report also includes potential conflict alerts, such as whether there are resource requirements for other processes that conflict with the adjusted resource allocation.

[0109] Step S1527: Feed back the schedule adjustment simulation report to the adjustment strategy formation stage. If there are conflict prompts in the schedule adjustment simulation report, the preliminary adjustment strategy needs to be re-optimized.

[0110] In this embodiment, the schedule adjustment simulation report is fed back to the adjustment strategy formation stage. If there are conflict warnings in the report, such as conflicts between the adjusted resource allocation and the resource requirements of other processes, the initial adjustment strategy needs to be re-optimized. For example, the number of equipment added for the concrete pouring process may be reduced, or the operation time of other processes may be adjusted to resolve the conflict. Then, the schedule simulation module of the dynamically associated BIM model is input again for simulation until there are no more conflict warnings.

[0111] Step S153: Extract the output of the progress simulation module of the dynamically associated BIM model to obtain the adjusted process time parameters. The adjusted process time parameters include the new planned start time, new planned end time and adjusted process cycle of each process. At the same time, check whether there are any conflicts in the adjusted process plan. The conflict types include process time overlap conflict and logical sequence conflict. Process time overlap conflict is when the same resource is occupied by multiple processes in the same time period. Logical sequence conflict is when the start time of the subsequent process is earlier than the end time of the preceding process.

[0112] In this embodiment, the output of the progress simulation module of the dynamically associated BIM model is extracted to obtain the adjusted process time parameters, such as the new planned start time, new planned end time, and adjusted process cycle for each process. Then, the adjusted process schedule is checked for conflicts. Process time overlap conflicts refer to the same resource being occupied by multiple processes within the same time period, such as a concrete pump truck being allocated to two processes within the same time period. Logical sequence conflicts refer to the start time of a subsequent process being earlier than the end time of a preceding process, such as the new planned start time of the rebar tying process being earlier than the new planned end time of the concrete pouring process. If conflicts exist, the adjustment strategy needs further optimization.

[0113] Step S154: If there is a conflict in the work process plan, optimize the preliminary adjustment strategy based on the spatial characteristics of the construction scene and the total resource information of the dynamic association BIM model. According to the optimized adjustment strategy, determine the final adjusted work process cycle, update the planned start time and planned end time of the corresponding work process node in the dynamic association BIM model, and at the same time, plan the equipment allocation path based on the spatial characteristics of the construction scene and the current location information of the equipment in the dynamic association BIM model. The equipment allocation path includes the starting position coordinates of the equipment, the target work process position coordinates, the route and the estimated allocation time. Through path optimization, the equipment can reach the target work area efficiently.

[0114] For example, step S1541: extract construction scene spatial feature data from the dynamically associated BIM model, and convert the construction scene spatial feature data into a digital map under a spatial coordinate system. The digital map includes the coordinate range of each area, the width and traffic capacity of the road, and the location and size of obstacles. The construction scene spatial feature data includes the topographic data of the construction area, road distribution data, work area boundary data, and obstacle distribution data.

[0115] In this embodiment, spatial feature data of the construction scene is extracted from the dynamically associated BIM model. Taking a water conservancy project as an example, the spatial feature data of the construction scene includes topographic data of the construction area, such as the elevation changes; road distribution data, such as the direction, width, and traffic capacity of construction roads; work area boundary data, such as the spatial boundaries of each work zone; and obstacle distribution data, such as the location and size of obstacles such as buildings and trees at the construction site. The above data is converted into a digital map in a spatial coordinate system, such as a Cartesian coordinate system. The coordinate range of each area is clearly marked, the width and traffic capacity of roads are represented in numerical form, and the location and size of obstacles are also clearly presented on the map.

[0116] Step S1542: Extract the current location information of the device, which includes the device's real-time coordinates, device type, device size, and device movement speed parameters.

[0117] In this embodiment, the current location information of the equipment is extracted. Taking a concrete pump truck as an example, its real-time coordinates can be obtained by a GPS positioning device installed on the equipment. The equipment type is a concrete pump truck, the equipment dimensions are length L, width W, and height H, and the equipment moving speed parameter is V (unit: meters / minute).

[0118] Step S1543: Obtain the work location coordinates of the target process from the adjusted process cycle information. The work location coordinates are the center coordinates of the work area of ​​the target process in the dynamic association BIM model. At the same time, determine the latest time for the equipment to arrive at the target work location. The latest time is the new planned start time of the target process minus the equipment preparation time.

[0119] In this embodiment, the work location coordinates of the target process are obtained from the adjusted process cycle information. Taking the concrete pouring process as an example, its work location coordinates are the center coordinates (X, Y, Z) of the work area, which are defined in the spatial coordinate system of the dynamically associated BIM model. Simultaneously, the latest time for the equipment to arrive at the target work location is determined. If the new planned start time for the target process is T, and the equipment preparation time is t (including equipment startup, inspection, etc.), then the latest time is Tt.

[0120] Step S1544: Input the digital map, the current coordinates of the equipment, the coordinates of the target work location, the equipment moving speed parameters and the latest arrival time into the equipment path planning AI model, and build a path search mechanism based on the A* algorithm, with path length and traffic efficiency as optimization objectives.

[0121] In this embodiment, the digital map, the current coordinates of the device (X1, Y1, Z1), the coordinates of the target work location (X, Y, Z), the device's moving speed parameter V, and the latest arrival time Tt are input into the device path planning AI model. This device path planning AI model constructs a path search mechanism based on the A* algorithm. The A* algorithm comprehensively considers path length and traffic efficiency as optimization objectives, prioritizing paths with shorter lengths and higher traffic efficiency.

[0122] Step S1545: Based on the road distribution in the digital map, search for all feasible paths from the current coordinates of the equipment to the coordinates of the target work location. Feasible paths refer to paths that can meet the equipment size requirements and avoid obstacles.

[0123] In this embodiment, based on the road distribution in the digital map, all feasible paths are searched from the current coordinates (X1, Y1, Z1) of the equipment to the coordinates (X, Y, Z) of the target work location. For example, the road distribution in the digital map includes main roads, side roads, etc. The path search algorithm checks whether each road can meet the size requirements for the concrete pump truck, that is, whether the width of the road is greater than the width of the equipment, whether the height is greater than the height of the equipment, and avoids the location of obstacles, such as buildings, trees, etc., thereby filtering out all feasible paths.

[0124] Step S1546: Evaluate all feasible paths. Evaluation indicators include total path length, number of obstacles on the path, road capacity, and estimated travel time. Calculate the evaluation score for each feasible path. The higher the evaluation score, the better the path. The estimated travel time is calculated based on the path length and equipment moving speed parameters. For road sections with low road capacity, a travel time correction value needs to be added.

[0125] In this embodiment, all feasible paths are evaluated, and the evaluation indicators include the total path length, the number of obstacles on the path, the road capacity, and the estimated travel time. The total path length can be obtained by calculating the sum of the lengths of each segment on the path; the number of obstacles on the path is obtained by counting the number of obstacles the path passes through; the road capacity is expressed in numerical form, such as the capacity of a certain road segment being 100 vehicles / hour; the estimated travel time is calculated based on the path length and the equipment moving speed parameter, and the formula is: Estimated travel time = Total path length / Equipment moving speed parameter. For road segments with low road capacity, a travel time correction value needs to be added. For example, if the capacity of a certain road segment is 50 vehicles / hour, which is lower than the standard capacity of 100 vehicles / hour, then the travel time correction value is: Path length / (Equipment moving speed parameter * (50 / 100)) - Path length / Equipment moving speed parameter. Then, the evaluation score for each feasible path is calculated. The evaluation score can be calculated by weighting and summing the evaluation indicators. The weights are set according to their importance. For example, the weight of the total path length is 0.3, the weight of the number of obstacles is 0.2, the weight of the road capacity is 0.3, and the weight of the estimated travel time is 0.2. The higher the evaluation score, the better the path.

[0126] Step S1547: Select the path with the highest evaluation score as the preliminary equipment dispatch path. The preliminary equipment dispatch path contains multiple consecutive coordinate points to form the route of the equipment. At the same time, calculate the total length of the preliminary equipment dispatch path and the estimated dispatch time. The estimated dispatch time is the total path length divided by the equipment moving speed parameter, plus the possible waiting time in the preliminary equipment dispatch path.

[0127] In this embodiment, the path with the highest evaluation score is selected as the initial equipment dispatch path. This initial equipment dispatch path contains multiple consecutive coordinate points, such as (X1, Y1, Z1) → (X2, Y2, Z2) → … → (X, Y, Z), forming the route of the equipment. Simultaneously, the total length of this initial equipment dispatch path is calculated, which is the sum of the distances between each coordinate point. The estimated dispatch time is the total path length divided by the equipment moving speed parameter, plus the possible waiting time in the initial equipment dispatch path. For example, in some sections, waiting for other equipment to pass is required, with a waiting time of t1. Therefore, the estimated dispatch time is (total path length / V) + t1.

[0128] Step S1548: Perform a conflict check between the preliminary equipment allocation path and other equipment allocation plans in the dynamically associated BIM model. Check whether other equipment passes through the key sections of the preliminary equipment allocation path within the same time period. If there is a path conflict, adjust the passage time of the preliminary equipment allocation path or search for an alternative path.

[0129] In this embodiment, a conflict check is performed between the preliminary equipment dispatch route and other equipment dispatch plans in the dynamically associated BIM model. For example, it checks whether other concrete pump trucks or cranes are traveling on key sections of the preliminary equipment dispatch route within the same time period. If a route conflict exists, the travel time of the preliminary equipment dispatch route needs to be adjusted, such as by advancing or delaying the departure time of the equipment, or by searching for alternative routes to avoid the conflict.

[0130] Step S1549: Based on the latest arrival time of the target process, check whether the estimated allocation time of the preliminary equipment allocation path meets the requirements. If the estimated allocation time is less than the difference between the latest arrival time and the current time, then the preliminary equipment allocation path meets the requirements.

[0131] In this embodiment, the estimated allocation time of the preliminary equipment allocation path is checked based on the latest arrival time Tt of the target process to see if it meets the requirements. The current time is t0, and the difference between the latest arrival time and the current time is (Tt) - t0. If the estimated allocation time is less than this difference, the preliminary equipment allocation path meets the requirements and can be used as the final equipment allocation path. If the estimated allocation time is greater than this difference, the preliminary equipment allocation path needs to be re-optimized, for example, by selecting a shorter path or increasing the equipment's moving speed (if possible).

[0132] Step S15410: Integrate the starting position coordinates of the equipment, the target process operation position coordinates, and the information of the preliminary equipment allocation path that meets the requirements into equipment allocation path information, associate the equipment allocation path information with the corresponding target process code, and convert the equipment allocation path information into visual path data, embedding it into the corresponding area of ​​the dynamically associated BIM model.

[0133] In this embodiment, the starting coordinates (X1, Y1, Z1) of the equipment, the coordinates of the target operation location (X, Y, Z), and the information of the preliminary equipment dispatching path that meets the requirements are integrated into equipment dispatching path information. This equipment dispatching path information includes the coordinates of the route, the total length, and the estimated dispatching time. Then, the equipment dispatching path information is associated with the corresponding target operation code (e.g., GC002) so that the corresponding equipment dispatching path information can be quickly retrieved in the dynamically associated BIM model by operation code. Simultaneously, the equipment dispatching path information is converted into visual path data; for example, the equipment dispatching path is marked with colored lines in the BIM model and embedded in the corresponding area of ​​the dynamically associated BIM model for easy viewing and following by construction personnel.

[0134] Step S155: Collect the adjusted process cycle information, equipment allocation path information, specific implementation steps of the adjustment strategy, responsible execution team and execution time node, form a construction progress adjustment instruction, and associate and bind the construction progress adjustment instruction with the dynamic association BIM model. The adjustment instruction is associated with the corresponding process node through process coding, so that construction personnel can click on the process node in the dynamic association BIM model to view the corresponding adjustment content.

[0135] In this embodiment, adjusted process cycle information is collected, such as the new planned start time, new planned end time, and adjusted process cycle for each process; equipment allocation path information, such as the allocation path of concrete pump trucks; specific implementation steps of the adjustment strategy, such as increasing the number of equipment or optimizing construction technology; responsible execution team, such as the second construction team being responsible for the concrete pouring process; and execution time nodes, such as the implementation time of the adjustment strategy being the second day. This information is integrated to form a construction progress adjustment instruction. Then, this construction progress adjustment instruction is associated and bound to the dynamically linked BIM model. The association between the adjustment instruction and the corresponding process node is achieved through process codes. For example, the concrete pouring process node with process code GC002 is associated and bound to the corresponding construction progress adjustment instruction. When construction personnel click on this process node, they can view the corresponding adjustment content, including the adjusted process cycle, equipment allocation path, specific implementation steps of the adjustment strategy, responsible execution team, and execution time nodes, thereby adjusting the construction progress according to the instruction requirements.

[0136] Based on the same inventive concept, please refer to Figure 2 This paper shows a schematic block diagram of the structure of a water conservancy project construction progress intelligent monitoring system 100 combining BIM and AI, which is provided in an embodiment of this application for executing the above-described intelligent monitoring method for water conservancy project construction progress combining BIM and AI. The water conservancy project construction progress intelligent monitoring system 100 combining BIM and AI may include a communication unit 110, a machine-readable storage medium 120, and a processor 130.

[0137] In this embodiment, both the machine-readable storage medium 120 and the processor 130 are located within the intelligent monitoring system 100 for the construction progress of water conservancy projects integrating BIM and AI, and are separately configured. However, it should be understood that the machine-readable storage medium 120 may also be independent of the intelligent monitoring system 100 for the construction progress of water conservancy projects integrating BIM and AI, and may be accessed by the processor 130 through a bus interface. Alternatively, the machine-readable storage medium 120 may also be integrated into the processor 130 and may communicate and interact with external systems through the communication unit 110.

[0138] The processor 130 is the control center of the intelligent monitoring system 100 for the construction progress of water conservancy projects integrating BIM and AI. It connects various parts of the system via various interfaces and lines, and performs overall monitoring by running or executing software programs and / or modules stored in the machine-readable storage medium 120, and by calling data stored in the machine-readable storage medium 120, executes various functions and processes data of the intelligent monitoring system 100. Optionally, the processor 130 may include one or more processing cores; for example, the processor 130 may integrate an application processor and a modem processor, wherein the application processor handles the operating system, user interface, and applications, and the modem processor handles wireless communication. It is understood that the modem processor may also not be integrated into the processor. The machine-readable storage medium 120 is used to store machine-executable instructions for executing the solution of this application, and the processor 130 is used to execute the machine-executable instructions stored in the machine-readable storage medium 120 to realize the intelligent monitoring method for water conservancy project construction progress that combines BIM and AI provided in the aforementioned method embodiment.

[0139] It should be noted that, in order to simplify the description of the present invention and thus help to understand one or more embodiments of the invention, multiple features may sometimes be grouped into one embodiment, drawing or description thereof in the foregoing description of the embodiments of the present invention.

Claims

1. A method for intelligent monitoring of construction progress in water conservancy projects combining BIM and AI, characterized in that, The method includes: Acquire BIM construction scene models and on-site multi-source perception data of the construction area for water conservancy projects. The BIM construction scene model includes the hierarchical relationship of the work process, the work process planning cycle, and the spatial characteristics of the construction scene. The on-site multi-source perception data includes construction equipment operation information, work process execution status data, and scene environment data. The BIM construction scene model is dynamically associated with the on-site multi-source perception data. A data mapping relationship is established based on the spatial characteristics of the construction scene to obtain a dynamically associated BIM model that includes real-time process status annotations. The construction progress simulation AI model is invoked to perform process progress correlation simulation on the dynamically associated BIM model. Based on the process hierarchy, the progress impact link of each process is simulated, and the process progress simulation results including the difference between the actual progress and the planned progress of the process are obtained. By combining the process association information in the dynamic BIM model, the cause analysis of the schedule difference in the process schedule projection results is carried out to obtain the schedule deviation attribution results including the causes of the difference and the scope of the impact. Based on the attribution results of schedule deviations, the process planning parameters in the BIM construction scenario model are adjusted to generate construction schedule adjustment instructions that include the adjusted process cycle and equipment allocation path.

2. The intelligent monitoring method for the construction progress of water conservancy projects combining BIM and AI according to claim 1, characterized in that, The process of dynamically associating the BIM construction scene model with multi-source sensing data on site, establishing a data mapping relationship based on the spatial characteristics of the construction scene, and obtaining a dynamically associated BIM model containing real-time process status annotations includes: The hierarchical structure of the BIM construction scene model is analyzed, and the process node set, process spatial distribution information and construction scene spatial features in the BIM construction scene model are extracted. The process node set includes the unique process code, planned start time and planned end time of each process. The construction scene spatial features include the construction area division boundary, equipment operation space range and process operation space requirements. The on-site multi-source sensing data is classified and analyzed, and divided into construction equipment operation information, process execution status data and scene environment data. The construction equipment operation information includes equipment code, operation location coordinates and operation duration; the process execution status data includes process code, current operation stage and operation duration; and the scene environment data includes temperature and humidity, lighting conditions and terrain change data of the operation area. Based on the spatial characteristics of the construction scene, data association rules are constructed. Taking the spatial requirements of the operation process as the benchmark, the equipment operation space range is matched with the spatial requirements of the operation process to establish the spatial association relationship between equipment code and process code. At the same time, the collection location coordinates of the scene environment data are matched with the spatial distribution information of the operation process to establish the location association relationship between the environmental data and the process code. Based on the data association rules, the process execution status data is directly matched to the corresponding process node in the BIM construction scene model through the process code, the construction equipment operation information is matched to the corresponding process node through the spatial association relationship between the equipment code and the process code, and the scene environment data is matched to the corresponding process node through the location association relationship, thus forming a preliminary associated data set; Based on the collection timestamps of multi-source sensing data on site, the preliminary associated data set is time-synchronized to ensure that the equipment information, status information and environmental information associated with the same process node have a unified time reference, and to generate a time-synchronized associated data set. The time-synchronized associated data set is embedded into the corresponding process node attributes of the BIM construction scene model in the form of real-time annotations. The real-time annotations include data type identifiers, collection timestamps, and specific data content, resulting in a dynamic associated BIM model containing real-time process status annotations. Set up a real-time update mechanism for the dynamic association BIM model. Set the update cycle of the dynamic association BIM model according to the collection frequency of multi-source sensing data on site. Repeat the data classification and parsing, association matching and time synchronization steps in each update cycle to replace the old real-time annotation information in the dynamic association BIM model so that the dynamic association BIM model reflects the latest construction site status. A data index for dynamically linked BIM models is built based on process codes. All real-time annotation information associated with each process node is bound to the process code. The equipment, status and environmental data of the corresponding node can be quickly queried through the process code, thereby improving the access efficiency of dynamically linked BIM model data.

3. The intelligent monitoring method for the construction progress of water conservancy projects combining BIM and AI according to claim 1, characterized in that, The process involves calling the construction progress simulation AI model to perform process progress correlation simulation on the dynamically associated BIM model. Based on the process hierarchy, the progress impact chain of each process is simulated, resulting in process progress simulation results that include the difference between the actual and planned progress of each process, including: Extract structured data from the dynamically associated BIM model to form a model input dataset. The model input dataset includes the process code, planned start time, planned end time, current work stage, completed work duration, work duration and work location coordinates of associated equipment, and associated scene environment data for each process node. The model input dataset is input into the scene adaptation layer of the construction progress extrapolation AI model. The scene adaptation layer combines the spatial features of the construction scene in the dynamically associated BIM model to perform scene-based preprocessing on the input data, generating scene-adapted input data. The scene-adapted input data is input into the process feature extraction layer of the construction progress simulation AI model. The process feature extraction layer adopts a parallel feature extraction structure. The time feature extraction unit extracts the time dimension features of the planned time parameters and the duration of work done for each process to generate a process time feature vector. In addition, the association feature extraction unit extracts the features of the association between the process and the equipment and the association between the process and the environmental data to generate a process association feature vector. The process time feature vector and process association feature vector are input into the feature fusion unit of the construction progress prediction AI model. Based on the attention mechanism, the contribution of the process time feature vector and process association feature vector to the process progress prediction is analyzed, the feature fusion weights are dynamically allocated, and the fused feature vector is generated by weighted summation. The fused feature vector reflects both the time attribute and the association attribute of the process. Extract the process hierarchy from the dynamic BIM model and construct a process association matrix. The process association matrix uses process codes as rows and columns. The elements of the process association matrix represent the dependency relationship between the corresponding row process and the corresponding column process. If the row process is the predecessor process of the column process, the element value is 1, otherwise it is 0. The sequential execution logic between each process is presented through the process association matrix. The fused feature vector and the process correlation matrix are input into the correlation inference layer of the construction progress inference AI model. The process correlation inference network is constructed based on graph neural network. The fused feature vector is used as the network node feature and the process correlation matrix is ​​used as the node connection relationship. Through multiple rounds of iteration, the impact of the progress change of each process on its related processes is inferred, and process progress impact link data is generated. In the correlation inference layer of the construction progress prediction AI model, for each process node, the actual completion time of the process is predicted by combining its already completed work time and current work stage and referring to the historical work cycle data of the same process in similar water conservancy projects. By comparing the actual completion time with the planned end time, the time difference between the actual progress and the planned progress of the process is calculated. By combining the process progress impact data and the time difference values ​​of each process, the final process progress projection result, which includes the difference between the actual progress and the planned progress of the process, is obtained.

4. The intelligent monitoring method for the construction progress of water conservancy projects combining BIM and AI according to claim 3, characterized in that, The process involves inputting the fused feature vector and the process correlation matrix into the correlation inference layer of the construction progress prediction AI model. A process correlation inference network is constructed based on a graph neural network, using the fused feature vector as the network node feature and the process correlation matrix as the node connection relationship. Through multiple iterations, the impact of progress changes in each process on its associated processes is predicted, generating process progress impact chain data, including: The basic architecture for constructing a process association inference network based on graph neural networks includes an input layer, a graph convolutional layer, a node update layer, and an output layer. The input layer is used to receive the fused feature vector and the process association matrix. The graph convolutional layer is used to realize the neighborhood aggregation of node features. The node update layer is used to update the feature representation of each node. The output layer is used to output the process progress impact link data. The fused feature vector is input into the input layer of the process association inference network. After feature standardization, it is distributed to each node of the process association inference network. Each node corresponds to a process code, and the initial feature value of the node is the fused feature vector corresponding to the process code. After the process association matrix is ​​input into the input layer of the process association inference network, it is converted into the adjacency matrix of the process association inference network. The element values ​​in the adjacency matrix represent the connection strength between the corresponding nodes. For process nodes with dependencies, the connection strength is set according to the degree of dependency. The connection strength of the preceding process to the subsequent process is higher than the connection strength of the subsequent process to the preceding process. In the graph convolutional layer of the process association inference network, for each node, the set of neighboring nodes of the node is selected according to the adjacency matrix. The set of neighboring nodes consists of other process nodes that have a dependency relationship with the node. The graph convolution operator is used to perform aggregation operation on the node's own features and the features of neighboring nodes to generate neighborhood aggregation features. The aggregation operation is used to capture the feature interaction information between the node and its neighboring nodes. The neighborhood aggregation features are input into the node update layer of the process association inference network. The nonlinear transformation is performed through the activation function. Combined with the change in the operation time of the corresponding process, the feature representation of the node is updated so that the node features reflect the latest state after the change in process progress. The number of iterations for the correlation deduction is set, which is determined according to the total number of procedures in the water conservancy project. In each iteration, the steps of neighbor node screening, feature aggregation and node update are repeatedly executed so that the features of each node are gradually transmitted to its associated nodes, thereby realizing the chain deduction of the progress impact. After each iteration, the feature change of each node is extracted through the output layer of the process association inference network. Nodes whose feature change exceeds a preset threshold are marked as affected nodes. The feature change reflects the degree to which the process corresponding to the node is affected by the progress changes of other related processes. For each affected node, trace back the transmission path of its characteristic changes, identify the source process node and intermediate transmission nodes that affect it, and form a process progress influence link with the source process as the starting point and the affected node as the ending point. The process progress influence link includes the process code of each node and the order of influence transmission. All process progress impact links are deduplicated to obtain the final process progress impact link data. The process progress impact link data is then associated with and stored with the corresponding node feature changes, so that each link contains impact degree information.

5. The intelligent monitoring method for the construction progress of water conservancy projects combining BIM and AI according to claim 1, characterized in that, The process association information in the dynamically associated BIM model is combined to perform attribution analysis on the schedule differences in the process schedule projection results, and obtain schedule deviation attribution results including the causes of the differences and the scope of their impact, including: Extract the schedule difference data from the process schedule projection results, filter out the deviation processes with non-zero time difference values, and form a deviation process set. The deviation process set includes the process code, actual completion time prediction value, planned end time, and time difference value of the deviation process. The correlation information corresponding to the deviation process is extracted from the dynamic BIM model. The correlation information includes the associated construction equipment operation information, scene environment data and process correlation relationship. The construction equipment operation information includes the equipment operation time, operation parameters and operation location changes. The scene environment data includes the environmental parameter changes during the operation of the deviation process. The process correlation relationship includes the information of the preceding and subsequent processes of the deviation process. A schedule deviation attribution index system is constructed, which includes equipment impact index, environmental impact index, process-related impact index, and self-execution index. The equipment impact index is used to measure the impact of equipment operating status on process progress, the environmental impact index is used to measure the impact of changes in environmental parameters on process progress, the process-related impact index is used to measure the impact of the progress of preceding processes on the current deviation process, and the self-execution index is used to measure the impact of the process's own execution efficiency on the schedule. The associated information of the deviation process is input into the schedule deviation attribution AI model. The schedule deviation attribution AI model includes an index calculation module and a cause identification module. The index calculation module quantifies the associated information according to the schedule deviation attribution index system and calculates the specific values ​​of each index. The equipment impact index is calculated by matching the equipment operation time with the operation time of the process. The environmental impact index is obtained by calculating the relative deviation between the actual value and the standard value of each environmental parameter and weighting and summing according to the degree of impact. The cause identification module uses a multi-class logistic regression algorithm, taking the calculated values ​​of each indicator as input features. It analyzes the contribution of each influencing factor through the trained model parameters to determine the cause of the schedule deviation. If the equipment influence indicator value is the largest, the cause is the equipment operation problem; if the environment influence indicator value is the largest, the cause is the environmental change problem; if the process association influence indicator value is the largest, the cause is the deviation of the preceding process; if the self-execution indicator value is the largest, the cause is the self-execution efficiency problem. Based on the process schedule impact chain data in the process schedule simulation results, determine the impact range of the deviation process on subsequent processes. The impact range includes the code of the subsequent process affected by the deviation process, the expected time difference value of each subsequent process, and the impact transmission path. Collect the process code, time difference value, cause of deviation, scope of impact and path of impact of the deviation process to form schedule deviation attribution data; The schedule deviation attribution data is organized according to the execution order of the deviation process, and an independent attribution entry is established for each deviation process to obtain the schedule deviation attribution results containing the cause of the difference and the scope of impact. The attribution entry contains complete deviation information, cause analysis and description of the scope of impact. Add an urgency indicator to each attribution item in the schedule deviation attribution results. The urgency is determined based on the importance of the deviation process and the size of its impact. The importance is extracted from the process hierarchy relationship in the dynamically associated BIM model. The larger the impact, the higher the urgency.

6. The intelligent monitoring method for the construction progress of water conservancy projects combining BIM and AI according to claim 5, characterized in that, The process involves inputting the associated information of the deviation process into the schedule deviation attribution AI model. This model includes an indicator calculation module and a cause identification module. The indicator calculation module quantifies the associated information based on the schedule deviation attribution indicator system, calculating the specific values ​​of each indicator, including: After receiving the correlation information of the deviation process through the index calculation module, the correlation information is preprocessed to convert unstructured environmental description data into structured environmental parameter values ​​and equipment operation status description into equipment operation parameter values. For equipment impact indicators, construction equipment operation information is extracted from the associated information. Construction equipment operation information includes equipment operation time, operation parameter stability and equipment failure records. The overlap between equipment operation time and deviation process operation time is calculated. At the same time, the deviation rate between equipment operation parameters and equipment standard operation parameters is calculated. If there are equipment failure records, a failure impact coefficient is added. The final value of the equipment impact indicator is the comprehensive calculation result of overlap, deviation rate and failure impact coefficient. The higher the overlap, the higher the equipment participation. The lower the deviation rate, the more stable the equipment operation status. For environmental impact indicators, scene environmental data is extracted from the associated information. Scene environmental data includes temperature and humidity changes, light intensity changes, precipitation and terrain changes during the operation. The standard operating environment parameter range corresponding to the deviation process is queried. For each environmental parameter, the relative deviation between the actual value and the standard value is calculated. The relative deviation is calculated as |actual value - standard value| / standard value range width, where the standard value range width is the difference between the upper and lower limits of the standard parameter. For standard parameters without a range, the standard value itself is used as the denominator to calculate the relative deviation. At the same time, weights are set according to the degree of influence of environmental parameters on the operation process. The weight of precipitation is higher than that of temperature and humidity changes. The final value of the environmental impact indicator is the weighted sum of the relative deviations of each environmental parameter and their corresponding weights. For the process association impact index, the information of the preceding process is extracted from the association information, the time difference value of the preceding process is obtained from the process progress projection results, and the ratio of the time difference value of the preceding process to the time difference value of the deviation process is calculated. The larger the ratio, the greater the impact of the preceding process on the deviation process. At the same time, the dependence strength in the process association relationship is combined. The higher the dependence strength, the greater the weight. The final value of the process association impact index is the product of the ratio and the dependence strength. For its own performance indicators, the system extracts the completed time, current stage, and planned cycle of the deviation process, calculates the ratio of completed time to the planned time for the current stage, and if the ratio is greater than 1, it indicates that its own performance efficiency is lower than the plan. At the same time, the system refers to the performance efficiency data of similar processes under the same environmental and equipment conditions, calculates the difference rate between the performance efficiency of the deviation process and the average performance efficiency of similar processes, and the final value of its own performance indicators is the comprehensive result of the ratio and the difference rate. The calculated index values ​​are normalized, and a correspondence between index values ​​and the degree of influence is established. The normalized index values ​​and the corresponding influencing factor names are compiled into an index quantification result table. The index quantification result table is output to the cause identification module as input features for deviation cause identification.

7. The intelligent monitoring method for the construction progress of water conservancy projects combining BIM and AI according to claim 1, characterized in that, The process planning parameters in the BIM construction scenario model are adjusted based on the schedule deviation attribution results, generating a construction schedule adjustment instruction that includes the adjusted process cycle and equipment allocation path, including: Extract key information from the schedule deviation attribution results, sort the deviation processes in descending order of urgency, and determine the priority order for schedule adjustments. The key information includes the deviation process code, deviation cause, scope of impact, urgency level indicator, and time difference value of each deviation process. For each priority deviation process, a targeted preliminary adjustment strategy is formed based on the cause of the deviation. The preliminary adjustment strategy is input into the progress simulation module of the dynamic BIM model. Based on the process association relationship and resource constraints of the BIM construction scenario model, the process execution after the implementation of the adjustment strategy is simulated. In this process, the resource configuration parameters in the dynamic BIM model are updated according to the resource adjustment suggestions in the adjustment strategy. Then, the operation cycle of each process is recalculated, and the new planned start time and new planned end time of the deviation process and the affected process are predicted. Extract the output of the progress simulation module of the dynamically associated BIM model to obtain the adjusted process time parameters. The adjusted process time parameters include the new planned start time, new planned end time and adjusted process cycle of each process. At the same time, check whether there are any conflicts in the adjusted process plan. The conflict types include process time overlap conflict and logical sequence conflict. Process time overlap conflict is when the same resource is occupied by multiple processes in the same time period. Logical sequence conflict is when the start time of the subsequent process is earlier than the end time of the preceding process. If there are conflicts in the work process schedule, the initial adjustment strategy is optimized based on the spatial characteristics of the construction scene and the total amount of resources in the dynamic BIM model. Based on the optimized adjustment strategy, the final adjusted work process cycle is determined, and the planned start time and planned end time of the corresponding work process node in the dynamic BIM model are updated. At the same time, based on the spatial characteristics of the construction scene and the current location information of the equipment in the dynamic BIM model, the equipment dispatch path is planned. The equipment dispatch path includes the starting position coordinates of the equipment, the target work process position coordinates, the route, and the estimated dispatch time. Through path optimization, the equipment can reach the target work area efficiently. Collect adjusted process cycle information, equipment allocation path information, specific implementation steps of adjustment strategies, responsible execution teams and execution time nodes to form construction progress adjustment instructions. Link and bind the construction progress adjustment instructions with the dynamic BIM model. The process code realizes the association between the adjustment instructions and the corresponding process nodes, so that construction personnel can click on the process nodes in the dynamic BIM model to view the corresponding adjustment content.

8. The intelligent monitoring method for the construction progress of water conservancy projects combining BIM and AI according to claim 7, characterized in that, The progress simulation module, which inputs the initial adjustment strategy into the dynamically associated BIM model, simulates the process execution after the adjustment strategy is implemented, based on the process relationships and resource constraints of the BIM construction scenario model. This includes: The resource adjustment parameters, time adjustment suggestions, and process optimization schemes in the preliminary adjustment strategy are analyzed. At the same time, basic data is extracted from the dynamic BIM model. The basic data includes the process relationship matrix, the standard operation cycle of each process, the total amount of construction resources, and the spatial constraints of the construction scene. The total amount of construction resources includes the number of equipment, the number of personnel, and the amount of material reserves. The spatial constraints of the construction scene include the maximum operation capacity of each work area. Based on the resource adjustment parameters in the initial adjustment strategy, the allocation status of construction resources is updated. Based on the updated resource allocation status and the process association matrix, the execution cycle of each process is calculated, and the key process path of the water conservancy project is determined. The key process path refers to the process sequence with the smallest total float. For deviation processes on key process paths, their operation cycles are recalculated based on the work efficiency optimization suggestions in the adjustment strategy and in combination with resource allocation. According to the order of the process relationships, calculate the new plan start time and new plan end time of each process in turn. The new plan end time of the preceding process is used as the benchmark for the new plan start time of the subsequent process. If there are multiple preceding processes for the subsequent process, the new plan end time of the latest completed preceding process is used as its new plan start time. During the time calculation process, combined with the spatial constraints of the construction scenario, it is checked whether there are simultaneous operations of processes in the same work area. If so, the new planned start time of one of the processes is adjusted, thereby outputting the new planned start time, new planned end time, adjusted process cycle and resource consumption details for each process. The resource consumption details include the number of equipment, personnel and materials required for each process after adjustment. The time calculation results are compared with the original planned time to generate a schedule adjustment simulation report. The schedule adjustment simulation report includes the differences in process time before and after the adjustment, changes in resource allocation, and possible conflict warnings. The schedule adjustment simulation report will be fed back to the adjustment strategy formation stage. If there are conflict warnings in the schedule adjustment simulation report, the initial adjustment strategy needs to be re-optimized.

9. A smart monitoring system for the construction progress of water conservancy projects combining BIM and AI, characterized in that, include: processor; A machine-readable storage medium for storing machine-executable instructions of the processor; The processor is configured to execute the intelligent monitoring method for water conservancy project construction progress combining BIM and AI as described in any one of claims 1 to 8 by executing the machine-executable instructions.

10. A computer program product, characterized in that, The computer program product includes machine-executable instructions stored in a computer-readable storage medium. The processor of the computer device reads the machine-executable instructions from the computer-readable storage medium and executes the machine-executable instructions, causing the computer device to perform the intelligent monitoring method for water conservancy project construction progress combining BIM and AI according to any one of claims 1 to 8.

Images