Intelligent management system for power transmission and transformation project construction based on digital twinning

Abstract

The invention provides a power transmission and transformation project construction intelligent management system based on a digital twinborn technology. The system comprises a digital twin model construction unit, a data acquisition unit, a data processing and anomaly detection unit, a construction process intelligent optimization unit and a control unit. The digital twin model construction unit is responsible for constructing a three-dimensional model of a power transmission and transformation project. The data acquisition unit is configured at a construction site, collects sensor data about construction progress and quality, and maps and compares the sensor data with the digital twinborn model. The data processing and anomaly detection unit analyzes real-time data, identifies construction anomaly and potential safety hazards, and carries out positioning and analysis through a digital twin model. And the construction process intelligent optimization unit optimizes a construction path and resource allocation in real time, and simulates and verifies on a digital twin model. And the control unit provides a user interface, so that a user can monitor the construction progress in real time, receive an alarm and make a decision.

Description

Technical Field

[0001] The present invention relates to the technical field of power transmission and transformation project construction management, and in particular to a power transmission and transformation project construction intelligent management system based on digital twins. Background Art

[0002] In the current construction management practice of power transmission and transformation projects, there are a series of challenges and technical problems, mainly including construction progress control, resource optimization allocation, safety risk management, and the need for real-time monitoring and decision support. Traditional management methods often rely on offline supervision and post-data aggregation, which is not only inefficient, but also difficult to achieve immediate response and adjustment to dynamic changes on the construction site. In addition, due to the lack of effective data integration and analysis methods, anomalies and safety hazards in the construction process are difficult to be identified and handled in a timely manner, which brings potential risks to project quality and safety.

[0003] In recent years, the rise of digital twin technology has provided new ideas for solving the above problems. By creating virtual copies of physical entities, digital twins can map and simulate the state of entities in real time, providing strong data support and visual reference for management and decision-making. However, in the field of power transmission and transformation project construction, how to effectively apply digital twin technology to construction management, especially how to achieve real-time monitoring, intelligent optimization and anomaly detection of the construction process, is still a technical problem that needs to be solved urgently.

[0004] Therefore, the need to develop an intelligent management system for power transmission and transformation project construction based on digital twins is particularly urgent. Summary of the invention

[0005] The present application provides an intelligent management system for power transmission and transformation project construction based on digital twin technology to improve the management quality of power transmission and transformation project construction.

[0006] This application provides a smart management system for power transmission and transformation project construction based on digital twins, including:

[0007] A digital twin model construction unit, used to construct a three-dimensional digital twin model of the power transmission and transformation project, wherein the three-dimensional digital twin model is used to map and simulate the status of the construction site in real time, and provide accurate basic data and visual reference;

[0008] A data acquisition unit is configured at the construction site to collect sensor data during the construction process and map and compare the sensor data with the corresponding parts in the three-dimensional digital twin model to verify the construction progress and quality;

[0009] The data processing and anomaly detection unit is used to analyze and process the real-time collected data to identify construction anomalies and safety hazards, and use the digital twin model to locate and analyze anomalies;

[0010] An intelligent optimization unit for the construction process, which is used to optimize the construction path and resource allocation in real time, and simulate and verify the real-time optimization on the 3D digital twin model;

[0011] A control unit is used to provide a user interface, which allows the user to monitor the construction progress and status in real time, receive abnormal and safety alarms, and make decisions and adjustments based on the model data provided by the three-dimensional digital twin model by displaying the three-dimensional digital twin model.

[0012] Furthermore, the digital twin model construction unit uses a graph neural network combined with an adaptive meshing algorithm to construct a three-dimensional digital twin model of the power transmission and transformation project;

[0013] The input of the graph neural network is the structural data of the power transmission and transformation project; the structural data is represented in the form of a graph, where nodes represent entities in the power system and edges represent the connection relationship between entities; the output of the graph neural network is a graph representation, which includes the importance score of the node and the strength of the connection between the nodes; the graph neural network uses a multi-layer graph neural network to process the input of the graph neural network; at each layer, the feature information of the node is updated according to the information of its neighboring nodes; after multiple rounds of iterations, the features of the node fully integrate the information of the surrounding environment, thereby obtaining a graph representation that fully reflects the structural characteristics of the power transmission and transformation project;

[0014] The input of the adaptive meshing algorithm includes the graph representation output by the graph neural network and the original data of the three-dimensional model of the power transmission and transformation project; the output of the adaptive meshing algorithm is an optimized three-dimensional mesh model; the adaptive meshing algorithm first analyzes the graph representation, and determines the structurally complex areas or functionally critical areas based on the importance scores of the nodes in the graph representation and the strength of the edges between the nodes; dynamically adjusts the density of the three-dimensional model network, including increasing the mesh density for the determined structurally complex areas or functionally critical areas; and uses geometric modeling technology to optimize the mesh of the three-dimensional model to obtain an optimized three-dimensional mesh model.

[0015] Furthermore, the graph neural network uses an aggregation function to synthesize all neighbor node information of node v in the graph neural network. The aggregation function AGGREGATE avg (v) This is achieved by calculating the average value of the neighbor node feature vectors using the following formula 1:

[0016]

[0017] Among them, N(v) is the set of neighbor nodes of node v, h u is the feature vector of neighbor node u;

[0018] Using the output of the aggregation function as input, the feature vector of node v is updated through linear transformation and nonlinear activation function ReLU using the following formula 2:

[0019] h′ v =ReLU(W AGGREGATE avg (v)+b) (2)

[0020] Among them, h′ v is the updated feature vector of node v; W and b are the weight matrix and bias vector in the graph neural network layer, respectively.

[0021] Furthermore, the graph neural network is implemented using a three-layer graph convolutional neural network, where the feature dimensions of the output of each layer of the graph convolutional neural network are 64, 64, and 32 respectively.

[0022] Furthermore, the data processing and anomaly detection unit uses the following formula 3 to identify the safety hazard level H of the construction site: risk :

[0023]

[0024] Among them, P ef,i represents the probability that the i-th data point in the construction site is in an abnormal state; I ds,j represents the quantitative value of the j-th construction status indicator provided by the digital twin model; w i YesP ef,i The weight of u j isI ds,j The weight of n is the total number of data points in the construction site; m is the total number of construction status indicators provided by the digital twin model; α and β are adjustment coefficients;

[0025] Furthermore, the probability that the i-th data point in the construction site is in an abnormal state is calculated by the following steps:

[0026] Preprocessing the sensor data related to the i-th data point in the construction site collected from the data acquisition unit to obtain preprocessed data; wherein the preprocessing includes data cleaning, standardization and feature extraction;

[0027] The preprocessed data is input into a trained support vector machine to obtain the probability that the i-th data point in the construction site is in an abnormal state.

[0028] Furthermore, the construction process intelligent optimization unit is specifically used for:

[0029] Monitor the status and progress of the construction site and update the digital twin model in real time through sensor data collected by the data acquisition unit;

[0030] Simulate multiple construction paths and resource allocation schemes in the updated digital twin model to obtain the expected performance of each construction path and resource allocation scheme under specific construction conditions, including expected cost, time and quality indicators;

[0031] According to the performance, the trained decision tree model is applied to select the optimal construction path and resource allocation scheme from a variety of construction paths and resource allocation schemes.

[0032] Furthermore, the input of the decision tree model includes weather conditions at the construction site, personnel distribution, equipment usage status, material inventory, each construction path and resource allocation plan, and the predicted performance of each construction path and resource allocation plan in the updated digital twin model; the output of the decision tree model is the optimal construction path and resource allocation plan.

[0033] This application has the following beneficial technical effects:

[0034] (1) By building a three-dimensional digital twin model of the power transmission and transformation project, the system can reflect the status of the construction site in real time and provide accurate basic data and visual reference for construction management. This real-time mapping and simulation capability enables managers to instantly understand the construction progress and on-site conditions and adjust construction plans and strategies in a timely manner, greatly improving the flexibility and responsiveness of construction management.

[0035] (2) The deployment of data acquisition units ensures the comprehensive collection of key data during the construction process. By mapping and comparing the collected data with the digital twin model, the system can effectively verify the construction progress and quality, detect deviations and problems in a timely manner, and ensure that construction activities are carried out according to the established plan and quality standards.

[0036] (3) The application of data processing and anomaly detection units has improved the ability to identify construction anomalies and safety hazards. Using digital twin models to locate and analyze anomalies can accurately find the specific location of the problem, providing a basis for taking targeted solutions, thereby effectively improving the level of construction safety management.

[0037] (4) The intelligent optimization unit of the construction process can optimize the construction path and resource allocation in real time based on the real-time data and information provided by the digital twin model. This optimization is not only based on the current construction status, but also predicts future construction needs and challenges, thereby achieving the optimal allocation of resources and the optimal planning of the construction path, significantly improving construction efficiency and resource utilization.

[0038] (5) The user interface provided by the control unit allows managers to monitor construction progress and status in real time, receive abnormal and safety alerts, and make decisions and adjustments based on model data provided by the 3D digital twin model. This intuitive and interactive decision support tool not only enhances the user experience, but also improves the quality and efficiency of construction management decisions. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 It is a schematic diagram of a smart management system for power transmission and transformation project construction based on digital twins provided in the first embodiment of the present application. DETAILED DESCRIPTION

[0040] Many specific details are described in the following description to facilitate a full understanding of the present application. However, the present application can be implemented in many other ways than those described herein, and those skilled in the art can make similar generalizations without violating the connotation of the present application, so the present application is not limited by the specific implementation disclosed below.

[0041] The first embodiment of the present application provides a smart management system for power transmission and transformation project construction based on digital twins. Figure 1 , which is a schematic diagram of the first embodiment of the present application. Figure 1 The first embodiment of the present application provides a detailed description of an intelligent management system for power transmission and transformation project construction based on digital twins.

[0042] The digital twin-based intelligent management system for power transmission and transformation project construction includes a digital twin model construction unit 101, a data acquisition unit 102, a data processing and anomaly detection unit 103, a construction process intelligent optimization unit 104 and a control unit 105.

[0043] Furthermore, the digital twin model construction unit uses a graph neural network combined with an adaptive meshing algorithm to construct a three-dimensional digital twin model of the power transmission and transformation project;

[0044] The input of the graph neural network is the structural data of the power transmission and transformation project; the structural data is represented in the form of a graph, where nodes represent entities in the power system and edges represent the connection relationship between entities; the output of the graph neural network is a graph representation, which includes the importance score of the node and the strength of the connection between the nodes; the graph neural network uses a multi-layer graph neural network to process the input of the graph neural network; at each layer, the feature information of the node is updated according to the information of its neighboring nodes; after multiple rounds of iterations, the features of the node fully integrate the information of the surrounding environment, thereby obtaining a graph representation that fully reflects the structural characteristics of the power transmission and transformation project;

[0045] The input of the adaptive meshing algorithm includes the graph representation output by the graph neural network and the original data of the three-dimensional model of the power transmission and transformation project; the output of the adaptive meshing algorithm is an optimized three-dimensional mesh model; the adaptive meshing algorithm first analyzes the graph representation to determine the structurally complex areas or functionally critical areas; dynamically adjusts the density of the three-dimensional model network, including increasing the mesh density for the determined structurally complex areas or functionally critical areas; and uses geometric modeling technology to optimize the mesh of the three-dimensional model to obtain an optimized three-dimensional mesh model.

[0046] Furthermore, the graph neural network uses an aggregation function to integrate all neighbor node information of a node. The aggregation function AGGREGATE avg (v) This is achieved by calculating the average value of the neighbor node feature vectors using the following formula 1:

[0047]

[0048] Among them, N(v) is the set of neighbor nodes of node v, h u is the feature vector of neighbor node u;

[0049] Using the output of the aggregation function as input, the feature vector of node v is updated through linear transformation and nonlinear activation function ReLU using the following formula 2:

[0050] h′ v =ReLU(W AGGREGATE avg (v)+b) (2)

[0051] Among them, h′ v is the updated feature vector of node v; W and b are the weight matrix and bias vector in the graph neural network layer, respectively.

[0052] The digital twin model construction unit 101 is used to construct a three-dimensional digital twin model of the power transmission and transformation project. The three-dimensional digital twin model is used to map and simulate the status of the construction site in real time and provide accurate basic data and visual reference.

[0053] In this embodiment, the digital twin model construction unit 101 is a core component developed based on the digital twin technology, and is intended to provide a highly accurate and dynamic three-dimensional digital twin model for power transmission and transformation projects. The model can not only map and simulate the state of the construction site in real time, but also provide accurate basic data and necessary visual references for construction management. The following is a detailed description of the digital twin model construction unit 101.

[0054] The development and implementation of the digital twin model building unit 101 is based on advanced technologies such as graph neural network (GNN) and adaptive mesh partitioning algorithm (AGDA). First, the unit uses graph neural network to process complex network structure data of power transmission and transformation projects, including but not limited to the location, size and connection relationship of wires, towers, transformers and other related facilities. The application of graph neural network can ensure the accuracy and efficiency of the model when processing large-scale and highly complex data.

[0055] Next, the adaptive meshing algorithm (AGDA) is used to refine and optimize the performance of the 3D model. The AGDA algorithm dynamically adjusts the density of the mesh according to the complexity and importance of each part of the model to achieve higher visual accuracy and simulation accuracy. This means that there will be more detailed meshing in key parts of the model (such as important structural nodes or key construction areas), providing more accurate data and visual effects.

[0056] The digital twin model building unit 101 also includes a data integration and synchronization mechanism that ensures that the data collected in real time from the construction site can be accurately mapped and integrated into the digital twin model. This includes processing and analyzing the sensor data collected from the data acquisition unit 102, and then mapping and comparing these data with the corresponding parts in the three-dimensional digital twin model. This process not only supports real-time verification of construction progress and quality, but also provides data support for anomaly detection and construction optimization.

[0057] To ensure the real-time update and accuracy of the 3D digital twin model, the construction unit 101 uses an efficient data processing framework and algorithm to process and synchronize large amounts of real-time data. In addition, the unit is highly configurable, allowing the parameters and strategies of model construction to be adjusted according to specific engineering needs and characteristics.

[0058] The application of graph neural network (GNN) in the three-dimensional digital twin model construction unit is explained in detail below.

[0059] The input of the graph neural network (GNN) mainly includes the structural data of the power transmission and transformation project, which is represented in the form of a graph. The nodes represent the entities in the power system (such as towers, transformers, etc.), and the edges represent the connection relationship between entities (such as cable connections). In addition, each node and edge can also carry additional attribute information, such as location coordinates, size, material type, etc.

[0060] The output of GNN processing is a highly accurate graph representation that contains enhanced feature information, such as the importance score of nodes, the strength of connections between nodes, etc. These enhanced features are used to guide the subsequent adaptive meshing algorithm to ensure that the model has a higher level of detail in key areas.

[0061] The implementation of GNN involves converting the structural data of the power transmission and transformation project into a graph data structure and processing it through a multi-layer graph neural network. At each layer, the feature information of the node is updated according to the information of its neighboring nodes. This process is completed through a specific aggregation function (such as sum, average or maximum value). After multiple rounds of iterations, the features of each node will fully integrate the information of the surrounding environment, thereby obtaining a graph representation that fully reflects the structural characteristics of the power transmission and transformation project.

[0062] The implementation of GNN specifically includes the following implementation steps:

[0063] 1. Detailed format and preprocessing of GNN input data:

[0064] The attribute information of nodes and edges should adopt the following data format:

[0065] Node attributes: `{'id':'N1','type':'Tower','position':[longitude,latitude],'height':50,'material':'Steel'}`. The position coordinates are expressed in longitude and latitude, and the height unit is meter.

[0066] Edge attributes: `{'source':'N1','target':'N2','type':'Cable','length':100,'load_capacity':200}`. The length unit is meter and the load capacity unit is kilowatt.

[0067] Each node is represented as a dictionary containing key information such as the node's unique identifier (id), type (such as "Tower" or "Transformer"), location (an array expressed in the form of longitude and latitude), height (in meters), material type, etc.

[0068] Each edge is also represented as a dictionary, containing information such as the source node (source), the target node (target), the connection type (such as "Cable"), the length (in meters), and the load capacity (in kilowatts).

[0069] The preprocessing includes the following steps:

[0070] Data cleaning: Remove nodes and edges that are missing key attributes (such as location or type).

[0071] Standardization: Convert all location coordinates into a unified Cartesian coordinate system to ensure compatibility between different data sources.

[0072] 2. Selection and implementation of aggregate functions:

[0073] We choose the average aggregation function as an example. For node v, its aggregation function is defined as:

[0074]

[0075] Among them, N(v) is the set of neighbor nodes of node v, h u is the feature vector of neighbor node u.

[0076] Feature update mechanism:

[0077] Node feature update uses linear transformation and ReLU activation function:

[0078] h′ v =ReLU(W AGGREGATE avg (v)+b)

[0079] Among them, W and b are the weight matrix and bias vector respectively, and ReLU is the activation function.

[0080] 3. Specific design of multi-layer GNN structure:

[0081] Network depth and layer design:

[0082] GNN uses a 3-layer graph convolutional neural network GCN, and the output feature dimensions of each layer are 64, 64, and 32 respectively. The output of the last layer is used for subsequent processing.

[0083] Parameter tuning and regularization strategies:

[0084] The Adam optimizer was used, the initial learning rate was set to 0.01, and dropout regularization was applied with a ratio of 0.5 to reduce overfitting.

[0085] 4. Training process and optimization algorithm:

[0086] The training steps of graph neural network (GNN) can be summarized into the following main steps, aiming to enable GNN to effectively learn and capture the structural characteristics of graph data and the complex relationships between nodes:

[0087] Prepare graph data: First, you need to prepare and preprocess the graph data, which includes defining the nodes and edges in the graph, as well as their features and labels. For the features of nodes and edges, some preprocessing operations such as normalization may be required. At the same time, depending on the specific task (such as node classification, graph classification, link prediction, etc.), the graph data may also need to be divided into training sets, validation sets, and test sets.

[0088] Define the GNN model structure: Next, define the GNN model structure, including selecting the appropriate graph convolution layer type (such as GCN, GAT, etc.), determining the number of layers of the model and the output dimension of each layer. In addition, it is necessary to define the forward propagation rules of the model, that is, how to update the feature representation of the node by aggregating the information of neighboring nodes.

[0089] Choose loss function and optimizer: According to the specific task that GNN is to solve, choose a suitable loss function (such as cross entropy loss, mean square error loss, etc.). At the same time, choose an optimization algorithm (usually a variant of gradient descent, such as SGD, Adam, etc.) to adjust the model parameters to minimize the loss function.

[0090] Training model: In the training phase, the graph data in the training set is input into the GNN model, and the optimizer is used to adjust the model parameters according to the feedback of the loss function, so that the model can learn the structural characteristics of the graph data and the relationship between nodes. Usually, this process will be carried out for multiple training cycles (epochs), and each cycle will traverse all the data in the training set once.

[0091] Validation and parameter tuning: During the training process, the performance of the model is regularly evaluated on the validation set to monitor the generalization ability of the model. Based on the validation results, the hyperparameters of the model (such as learning rate, number of layers, hidden layer dimensions, etc.) may need to be adjusted to achieve better performance.

[0092] Testing and evaluation: After training, the performance of the GNN model is evaluated on the test set to verify the model's generalization ability for unseen data. The performance evaluation indicators vary depending on the task and may include accuracy, recall, F1 score, etc.

[0093] Model deployment and application: Once the model performs well on the test set, it can be deployed to actual applications, such as obtaining a graphical representation that fully reflects the structural characteristics of the power transmission and transformation project.

[0094] 5. Application of output features:

[0095] The output features of GNN (node ​​importance scores) are used as input to the adaptive meshing algorithm to optimize mesh density. The node importance scores are used for resource allocation and priority adjustment to assist construction decision-making.

[0096] 6. Performance evaluation and verification:

[0097] The model performance is evaluated using precision, recall, and F1 score, with special attention paid to the recognition accuracy of high-importance nodes.

[0098] The validation process uses a 5-fold cross-validation method to ensure the generalization ability of the model on different data subsets.

[0099] The last layer output of the 3-layer GCN above is used for subsequent processing. Extracting information about node importance and connection strength from the 32-dimensional features output by the last layer can be achieved through a series of carefully designed post-processing steps. The following is a detailed description of how to perform these steps:

[0100] 1. Feature Dimensionality Reduction:

[0101] Dimensionality reduction is a common method for processing high-dimensional data, which aims to reduce the dimension of the data while retaining the important information of the original data as much as possible. For the 32-dimensional features output by the last layer of GNN, principal component analysis (PCA), t-distributed stochastic neighbor embedding (t-SNE) or unified manifold approximation and projection (UMAP) can be used for dimensionality reduction.

[0102] 2. Clustering:

[0103] Clustering is an exploratory data analysis technique used to discover natural groupings in data. Clustering the reduced features can help identify groups of nodes with similar characteristics, which may represent nodes with similar importance or play similar roles in the network.

[0104] 3. Importance score extraction:

[0105] The importance score of a node can be extracted by combining the clustering results with the original 32-dimensional features. For example, the importance of a node can be evaluated based on its centrality in a specific cluster (such as how far it is from the cluster center), or the feature values ​​of certain dimensions can be directly used as an indicator of importance.

[0106] 4. Connection strength assessment:

[0107] The strength of the connection between nodes can be evaluated by comparing the feature vectors of the node pair. For example, the cosine similarity of the feature vectors of two nodes can be calculated as an indicator of the connection strength.

[0108] Through the above steps, information about node importance and connection strength can be extracted and emphasized from the output of GNN, providing a basis for subsequent applications such as network optimization, fault prediction, etc.

[0109] The application of the adaptive grid generation algorithm (AGDA) is described in detail below.

[0110] The input of the AGDA algorithm includes the graph representation of the GNN output, especially the importance score of the node and the strength evaluation of the edge. In addition, the original data of the 3D model of the power transmission and transformation project is also required, including information such as geometry and spatial position. The original data of the 3D model of the power transmission and transformation project can come from the engineering design software.

[0111] During the planning and design phase of power transmission and transformation projects, engineers use professional engineering design software (such as AutoCAD, Revit, PLS-CADD, etc.) to create a 3D model of the power system. These software can provide detailed information such as geometry, structural layout, material properties, etc., forming the original data of the 3D model.

[0112] The output of the AGDA algorithm is an optimized 3D mesh model, where the density of the mesh is dynamically adjusted according to the complexity and importance of each part of the model. In areas with complex structures or high criticality, the mesh density will be increased so that the model can more accurately express the details of these areas.

[0113] The AGDA algorithm first evaluates the complexity and importance of each part based on the node scores and edge evaluations in the graph representation provided by GNN. The algorithm then dynamically adjusts the density of the mesh division based on the evaluation results, using finer meshes to represent areas with high scores (i.e., important or complex). This process involves geometric modeling techniques and mesh optimization algorithms to ensure that the model is represented with the most appropriate accuracy in different areas.

[0114] The implementation of the AGDA algorithm can be divided into the following steps:

[0115] (1) Evaluating complexity and importance: First, the graph representation output by the GNN is analyzed, especially the importance scores of nodes and the strength of edges between nodes. This step involves analyzing the scores of each node and its adjacent edges in the graph to determine which areas in the 3D model mesh are structurally complex or functionally critical.

[0116] (2) Dynamically adjust mesh density: Based on the evaluation results of the previous step, the AGDA algorithm will dynamically adjust the density of the 3D model mesh. For areas that are assessed as important or complex, the algorithm will increase the mesh density to more finely represent the details of these areas. Specifically, this may involve calculating the minimum mesh size required for each area to ensure sufficient accuracy.

[0117] (3) Mesh optimization: Use geometric modeling techniques to optimize the mesh of the 3D model. This includes applying dynamically adjusted mesh density parameters to each part of the model and regenerating the mesh to ensure that the model's details are optimized while maintaining the overall accuracy of the model.

[0118] (4) Output optimized model: Finally, the AGDA algorithm outputs an optimized three-dimensional mesh model. This model will reflect the detailed geometric structure of the power transmission and transformation project and have a higher mesh density in key and complex areas, thereby improving the expression accuracy and practicality of the overall model.

[0119] The data acquisition unit 102 is configured at the construction site to collect sensor data during the construction process, and map and compare the sensor data with the corresponding parts in the three-dimensional digital twin model to verify the construction progress and quality.

[0120] In the smart management system provided in this embodiment, the data acquisition unit 102 plays a vital role. It is the basis for realizing the digitalization and intelligence of the construction management of power transmission and transformation projects. The main responsibility of the data acquisition unit 102 is to collect various sensor data in real time at the construction site. These data cover various information during the construction process, including but not limited to temperature, humidity, vibration, noise level, equipment operation status, and the location of construction personnel. The collected data reflects the real-time status of the construction site and provides original and real-time data support for construction management.

[0121] The specific implementation of the data acquisition unit 102 configured at the construction site involves the deployment and management of multiple sensors. These sensors may be fixedly installed on key equipment, or they may be carried by construction workers or installed on mobile devices. In order to ensure the comprehensiveness and accuracy of the data, the selection of sensors, deployment locations, collection frequency, etc. need to be determined according to the specific conditions and management requirements of the construction site.

[0122] When implementing the data acquisition unit 102, the following key aspects need to be considered:

[0123] 1. Sensor selection and configuration: Choose the right type of sensor based on the specific needs of the construction site. For example, temperature and humidity sensors are used to monitor environmental conditions, vibration sensors are used to monitor the operating status of equipment, and GPS sensors are used to track the location of construction personnel and equipment. The configuration of sensors needs to ensure that they can cover key areas and equipment on the construction site to obtain comprehensive data.

[0124] 2. Data transmission: Considering that the environment of the construction site may be relatively complex, the data acquisition unit 102 needs to adopt a reliable data transmission method, such as wireless network, Bluetooth or other applicable communication technology, to ensure stable transmission of real-time data.

[0125] 3. Data preprocessing: The collected raw data may contain noise or incomplete information. Therefore, before the data is transmitted to the subsequent processing unit, necessary preprocessing, such as data cleaning, formatting and preliminary analysis, is required to improve the quality and usability of the data.

[0126] 4. Mapping with the three-dimensional digital twin model: The data collected by the data acquisition unit 102 needs to be mapped and compared with the corresponding parts in the digital twin model. This process involves data correlation analysis and fusion, and it is necessary to ensure the consistency and accurate mapping between the data and the model in order to accurately reflect the real-time status of the construction site and support subsequent analysis and decision-making.

[0127] Through the above implementation steps, the data acquisition unit 102 can be realized to provide real-time and accurate data support for the intelligent management system of power transmission and transformation project construction. This can not only improve the efficiency and safety of construction management, but also provide data basis for decision-making during the construction process, and further promote the development of construction management towards digitalization and intelligence.

[0128] The data processing and anomaly detection unit 103 is used to analyze and process the data collected in real time to identify construction anomalies and safety hazards, and use the digital twin model to locate and analyze anomalies.

[0129] In the smart management system, the data processing and anomaly detection unit 103 is a key component, which is responsible for processing the real-time data collected from the data acquisition unit 102 and analyzing the data to identify possible construction anomalies and safety hazards. In addition, the unit is also responsible for locating and analyzing abnormal situations using the digital twin model, providing important support for construction management and decision-making.

[0130] The process of implementing the data processing and anomaly detection unit 103 involves several important steps, which are described in detail below.

[0131] First, a data processing process needs to be established to process and analyze the real-time data collected from the construction site. This includes steps such as data preprocessing, feature extraction, and cleaning, aiming to improve the quality of the data and the accuracy of the analysis. Preprocessing may include operations such as removing noise, filling missing values, and normalizing data. Through these processes, it can be ensured that subsequent analysis and anomaly detection are based on accurate and reliable data.

[0132] Next, design and implement anomaly detection algorithms that can identify potential construction anomalies and safety hazards based on the processed data. This may include setting thresholds, pattern recognition, machine learning, and other technical means. For example, statistical analysis methods can be used to identify outliers in the data, or machine learning models can be used to predict and identify abnormal patterns. The key is to select or develop detection algorithms that are suitable for the specific conditions and needs of the construction site.

[0133] In addition, the data processing and anomaly detection unit 103 also needs to be tightly integrated with the digital twin model. This means that once an anomaly is detected, the unit is not only able to report the abnormal event, but also use the digital twin model to locate and conduct in-depth analysis of the abnormal situation. This may involve matching the abnormal data with the corresponding entities or processes in the model, and using the model to simulate and analyze to determine the cause and possible impact of the anomaly.

[0134] To achieve this integration, technicians need to ensure that the digital twin model has sufficient detail and accuracy to reflect the actual situation on the construction site. In addition, appropriate interfaces and protocols need to be developed so that the data processing and anomaly detection units can effectively access and utilize the digital twin model.

[0135] Finally, in order to support construction management and decision-making, the data processing and anomaly detection unit should also be able to feed back the detection results to the construction team in an appropriate form. This may include automatically generating reports, sending alarms, or displaying abnormal information directly on the user interface. Therefore, the unit also needs to have the ability to interact with the control unit and the user interface.

[0136] Furthermore, the data processing and anomaly detection unit uses the following formula 3 to identify the safety hazard level H of the construction site: risk :

[0137]

[0138] Among them, P ef,i represents the probability that the i-th data point in the construction site is in an abnormal state; I ds,h represents the quantitative value of the j-th construction status indicator provided by the digital twin model; w i YesP ef,i The weight of u j isI ds,j The weight of n is the total number of data points in the construction site; m is the total number of construction status indicators provided by the digital twin model; α and β are adjustment coefficients;

[0139] Furthermore, the probability that the i-th data point in the construction site is in an abnormal state is calculated by the following steps:

[0140] Preprocessing the sensor data related to the i-th data point in the construction site collected from the data acquisition unit to obtain preprocessed data; wherein the preprocessing includes data cleaning, standardization and feature extraction;

[0141] The preprocessed data is input into a trained support vector machine to obtain the probability that the i-th data point in the construction site is in an abnormal state.

[0142] In this embodiment, a method for evaluating the safety hazard level H of a construction site is proposed. risk The formula is:

[0143]

[0144] P ef,i represents the probability that the ith data point belongs to an abnormal state. This is determined by analyzing the sensor data collected at the construction site, and the specific method will be explained in detail below.

[0145] I dsd,h It is the quantitative value of the jth construction status indicator provided by the digital twin model. These indicators may include equipment temperature, vibration level, noise level, etc., which are obtained by real-time mapping and simulation of the construction site by the digital twin model.

[0146] w i is the weight of the probability of abnormal state of the ith data point, reflecting the relative importance of the data point in the safety hazard assessment. The weight can be determined based on historical data analysis or expert experience.

[0147] u j is the weight of the jth construction status indicator, which indicates the contribution of this status indicator in assessing the safety hazard level.

[0148] n is the total number of data points in the construction site, and m is the total number of construction status indicators provided by the digital twin model. These two values ​​depend on the monitoring scope of the actual construction site and the level of detail of the digital twin model.

[0149] α and β are adjustment coefficients used to balance the influence of abnormal state probability and construction state index in safety hazard assessment. The selection of these coefficients can be adjusted through optimization algorithm to achieve the best assessment effect.

[0150] To obtain the probability P that the i-th data point in the construction site is in an abnormal state ef,i , you need to perform the following steps:

[0151] 1. Data preprocessing: First, the sensor data related to the ith data point is collected from the data acquisition unit. The data is preprocessed, including data cleaning (removing erroneous or incomplete data records), standardization (converting data into a uniform format or range), and feature extraction (extracting key features from raw data that help identify anomalies).

[0152] 2. Abnormal probability calculation: Input the preprocessed data into a trained support vector machine (SVM) model. SVM is a supervised learning algorithm that can learn and distinguish between normal and abnormal states based on historical data. The model will output a probability value, namely P ef,i, indicating the possibility that the i-th data point belongs to an abnormal state. Since SVM is a common model, its construction and training process will not be described in detail here.

[0153] Through the above steps, this embodiment provides a method that comprehensively considers real-time sensor data and digital twin model status indicators to evaluate the safety hazard level of the construction site. This method can help the construction management team to identify and respond to potential safety risks in a timely manner and improve the safety management level of the construction site.

[0154] When the construction site safety hazard level indicates a higher risk, the management team can conduct a deeper analysis of the data to identify the specific factors that led to the high risk assessment. This may involve going back to the raw data or using more granular analysis tools to examine specific data points or areas.

[0155] The safety hazard level of the construction site can be used as part of regular assessments, combined with an immediate feedback mechanism, to help the management team adjust safety measures and resource allocation in a timely manner, and even prevent potential safety hazards before specific problems arise.

[0156] Through the above detailed description, the data processing and anomaly detection unit 103 can be implemented to provide effective data analysis and anomaly detection support for the intelligent management system for power transmission and transformation project construction, thereby improving construction safety and efficiency.

[0157] The construction process intelligent optimization unit 104 is used to optimize the construction path and resource allocation in real time, and simulate and verify the real-time optimization on the three-dimensional digital twin model.

[0158] In the intelligent management system for power transmission and transformation project construction, the construction process intelligent optimization unit 104 is a key component, whose main responsibility is to implement real-time optimization of construction paths and resource allocation, as well as to simulate and verify these optimization measures on the three-dimensional digital twin model. Through the operation of the intelligent optimization unit, the system can improve construction efficiency, reduce resource waste, and ensure construction quality and safety.

[0159] The implementation details of the construction process intelligent optimization unit 104 are as follows:

[0160] The construction process intelligent optimization unit 104 is designed to support dynamic adjustment of the construction path and real-time optimization of resource allocation. It identifies potential problems and optimization opportunities in the construction process by analyzing the three-dimensional digital twin model and the real-time collected construction site data. For example, by analyzing the model data, the intelligent optimization unit can propose a more efficient construction path, or suggest optimizing the allocation of human and mechanical resources within a specific time.

[0161] Realizing the intelligent optimization unit of the construction process involves multiple technical links:

[0162] 1. Data integration and analysis: First, the data collected from the data acquisition unit 102 needs to be integrated with the corresponding parts in the digital twin model. This includes construction progress data, resource usage, and on-site environmental conditions. Through advanced data analysis techniques, such as machine learning algorithms, the unit can extract useful information from these data to support optimization decisions.

[0163] 2. Optimization algorithm design: The intelligent optimization unit uses advanced optimization algorithms, such as genetic algorithms, particle swarm optimization or other applicable algorithms, to find the optimal construction path and resource allocation plan. The design of the algorithm needs to consider multiple constraints, such as construction time limit, resource availability, safety regulations, etc.

[0164] 3. Simulation and verification: Through the three-dimensional digital twin model, the intelligent optimization unit can not only propose optimization solutions, but also simulate on the model to predict the actual effect of the optimization measures. This step is crucial to verify the feasibility and effect of the optimization solution.

[0165] 4. Feedback mechanism: The intelligent optimization unit also includes a feedback mechanism, which continuously adjusts and optimizes the algorithm parameters based on the comparative analysis of simulation results and actual construction conditions to improve the accuracy and applicability of the optimization scheme.

[0166] In terms of operation, the construction process intelligent optimization unit will regularly receive the latest construction data from the data acquisition unit, analyze it in combination with the 3D digital twin model, and identify optimization opportunities. Subsequently, the unit will automatically generate optimization plans and verify the effectiveness of these plans through simulation. Finally, based on the simulation results, the system will provide specific optimization suggestions to the construction management team for decision-making reference.

[0167] Furthermore, the construction process intelligent optimization unit is specifically used for:

[0168] Monitor the status and progress of the construction site and update the digital twin model in real time through sensor data collected by the data acquisition unit;

[0169] Simulate multiple construction paths and resource allocation schemes in the updated digital twin model to obtain the expected performance of each construction path and resource allocation scheme under specific construction conditions, including expected cost, time and quality indicators;

[0170] According to the performance, the trained decision tree model is applied to select the optimal construction path and resource allocation scheme from a variety of construction paths and resource allocation schemes.

[0171] Furthermore, the input of the decision tree model includes weather conditions at the construction site, personnel distribution, equipment usage status, material inventory, each construction path and resource allocation plan, and the predicted performance of each construction path and resource allocation plan in the updated digital twin model; the output of the decision tree model is the optimal construction path and resource allocation plan.

[0172] The following will explain in detail how to implement a smart management system for power transmission and transformation project construction based on digital twins, especially the intelligent optimization unit of the construction process and its process of selecting the optimal construction path and resource allocation plan using a decision tree model.

[0173] First, the smart management system continuously monitors the real-time status and progress of the construction site through its data acquisition unit, which includes but is not limited to weather conditions, personnel distribution, equipment usage, and material inventory. This data is collected and transmitted to the system in real time for real-time updating of the digital twin model. The digital twin model is a virtual representation of the construction project that can accurately reflect the various conditions and changes on the construction site.

[0174] Then, using the updated digital twin model, the system can simulate multiple construction paths and resource allocation options, predicting the performance of each option under specific construction conditions. These performance indicators include expected cost, construction period, and quality, providing a comprehensive evaluation for each option.

[0175] To select the best solution from multiple options, the system applies a trained decision tree model. The input features of the decision tree model include weather conditions, personnel distribution, equipment usage and material inventory at the construction site, as well as each construction path and resource allocation scheme and its predicted performance in the digital twin model. These input data provide the decision tree with the necessary information for effective learning and decision-making.

[0176] The decision tree model analyzes these input features to evaluate the pros and cons of various construction paths and resource allocation schemes. The output of the model is a clear choice, that is, the construction path and resource allocation scheme predicted to be the best under given construction conditions and available resources.

[0177] The key to implementing this process lies in the accurate collection and processing of real-time data from the construction site, as well as the effective use of digital twin models to simulate and predict the performance of different solutions. In addition, the correct training and application of decision tree models are decisive factors in ensuring the selection of the best solution. In this way, the smart management system can improve construction efficiency, optimize resource allocation, reduce costs, and ensure construction quality, bringing significant improvements and advantages to construction management.

[0178] The following is a reference Python code implementation that uses a decision tree algorithm to select the optimal construction path and resource allocation solution.

[0179] from sklearn.tree import DecisionTreeClassifier

[0180] import pandas as pd

[0181] #Assume that there is already a DataFrame containing the data of the construction site

[0182] #Data columns may include: 'WeatherCondition', 'PersonnelDistribution', 'EquipmentUsage', 'MaterialStock', 'ConstructionPathAndResourceOption', 'PredictedPerformance'

[0183] #Where 'PredictedPerformance' is the predicted performance of the digital twin model for each solution, which can be a numerical value based on some scoring or classification criteria

[0184] #'ConstructionPathAndResourceOption' is a categorical variable representing different construction paths and resource configuration options

[0185] # Loading sample data

[0186] #data=pd.read_csv('construction_site_data.csv')

[0187] #Assume the data has been preprocessed and prepared

[0188] X=data[['WeatherCondition','PersonnelDistribution','EquipmentUsage','MaterialStock','PredictedPerformance']]#Input features

[0189] y=data['ConstructionPathAndResourceOption']#Output target (different construction paths and resource allocation plans)

[0190] # Initialize the decision tree classifier

[0191] decision_tree_classifier=DecisionTreeClassifier(random_state=0)

[0192] #Train model

[0193] decision_tree_classifier.fit(X,y)

[0194] #In actual applications, real-time updated data is needed to predict the optimal construction path and resource allocation plan

[0195] #real_time_data=[...]

[0196] #Use the model to make predictions

[0197] #predicted_option=decision_tree_classifier.predict(real_time_data)

[0198] #Print out the selected optimal construction path and resource allocation plan

[0199] #print("Selected optimal construction path and resource configurationoption:",predicted_option)

[0200] Please note that in order to run the above code, you need to replace the data = pd.read_csv('construction_site_data.csv') and real_time_data = [...] parts in the comments to load the actual construction site data and provide real-time input to the model.

[0201] This code snippet provides a basic framework for using decision trees to optimize construction paths and resource allocation. In practical applications, further preprocessing of the data may be required, such as handling missing values, encoding categorical variables, etc., and the parameters of the decision tree model may need to be adjusted to achieve optimal performance.

[0202] The implementation of this unit will help optimize construction plans, improve resource utilization efficiency, ensure construction safety, and ultimately achieve the successful completion of construction projects.

[0203] The control unit 105 is used to provide a user interface, which allows the user to monitor the construction progress and status in real time, receive abnormal and safety alarms, and make decisions and adjustments based on the model data provided by the three-dimensional digital twin model by displaying the three-dimensional digital twin model.

[0204] The control unit 105 plays a pivotal role in the intelligent management system for power transmission and transformation project construction. It provides an interactive interface for users, through which users can not only monitor the construction progress and status in real time, but also receive abnormal and safety alarms, and make decisions and adjustments based on the data provided by the three-dimensional digital twin model. The key to implementing the control unit 105 is to ensure that it can effectively integrate and present information from various parts of the system, while providing an intuitive and easy-to-use operating interface to support various construction management tasks.

[0205] The key steps to realize the control unit include:

[0206] 1. Interface design: The user interface of the control unit should be designed to be intuitive and easy to use, and can clearly display the 3D digital twin model of the construction site, including real-time construction progress, resource allocation, environmental status and other information. The interface should also provide navigation and control functions so that users can easily browse different views and perform different operations, such as zooming in and rotating the model, and viewing detailed information of a specific area.

[0207] 2. Data integration and presentation: The control unit needs to be able to receive and process in real time the sensor data from the data acquisition unit 102, the analysis results of the data processing and anomaly detection unit 103, and the optimization suggestions of the construction process intelligent optimization unit 104. These data should be integrated and presented on the user interface in an easy-to-understand form, such as through charts, indicator lights, or notification messages.

[0208] 3. Abnormality and alarm mechanism: The control unit should have real-time monitoring and alarm functions, and be able to immediately notify the user when abnormal conditions or safety hazards are detected. This includes receiving abnormal analysis results from the data processing and abnormality detection unit 103, and issuing alarms to the user on the interface through sound, pop-up windows, etc.

[0209] 4. Decision support: The control unit should also provide decision support functions to help users make construction plans and adjustments based on the model data provided by the 3D digital twin model. This may include providing simulation tools that allow users to preview construction progress and resource usage under different strategies, as well as providing optimization suggestions and operational recommendations.

[0210] 5. Interactive operation: Users should be able to make real-time adjustments to construction plans and resource configurations through the control unit interface, including modifying construction paths, reallocating resources, etc. The system should be able to respond to these operations instantly and update the 3D digital twin model to reflect these changes.

[0211] Technical implementation details include:

[0212] Front-end technologies: Developing control unit interfaces can leverage modern web technologies such as HTML5, CSS3, and JavaScript frameworks such as React or Vue.js to create responsive and interactive user interfaces.

[0213] Backend integration: The control unit needs to be tightly integrated with the system's backend services to ensure that data can be received and processed in real time. This may involve using WebSocket or polling mechanisms to obtain the latest data, and using technologies such as RESTful API or GraphQL to communicate with the backend services.

[0214] 3D model display: Displaying 3D digital twin models may require the use of specialized 3D graphics libraries such as Three.js or Unity3D, which provide rich functions for rendering and manipulating 3D graphics.

[0215] Through the above detailed description, it should be possible to understand and implement the control unit 105 to provide a powerful and intuitive user interaction interface for the intelligent management system for power transmission and transformation project construction, thereby supporting efficient and flexible construction management.

[0216] Although the present application is disclosed as above in the form of a preferred embodiment, it is not intended to limit the present application. Any technical personnel in this field may make possible changes and modifications without departing from the spirit and scope of the present application. Therefore, the scope of protection of the present application shall be based on the scope defined by the claims of the present application.

Claims

1. A digital twin-based intelligent management system for power transmission and transformation project construction, characterized in that: include: A digital twin model construction unit, used to construct a three-dimensional digital twin model of the power transmission and transformation project, wherein the three-dimensional digital twin model is used to map and simulate the status of the construction site in real time, and provide accurate basic data and visual reference; A data acquisition unit is configured at the construction site to collect sensor data during the construction process and map and compare the sensor data with the corresponding parts in the three-dimensional digital twin model to verify the construction progress and quality; The data processing and anomaly detection unit is used to analyze and process the real-time collected data to identify construction anomalies and safety hazards, and use the digital twin model to locate and analyze anomalies; An intelligent optimization unit for the construction process, which is used to optimize the construction path and resource allocation in real time, and simulate and verify the real-time optimization on the 3D digital twin model; A control unit is used to provide a user interface, which allows the user to monitor the construction progress and status in real time, receive abnormal and safety alarms, and make decisions and adjustments based on the model data provided by the three-dimensional digital twin model by displaying the three-dimensional digital twin model.

2. The intelligent management system for power transmission and transformation project construction according to claim 1 is characterized in that: The digital twin model construction unit uses a graph neural network combined with an adaptive grid division algorithm to construct a three-dimensional digital twin model of the power transmission and transformation project; The input of the graph neural network is the structural data of the power transmission and transformation project; the structural data is represented in the form of a graph, where nodes represent entities in the power system and edges represent the connection relationship between entities; the output of the graph neural network is a graph representation, which includes the importance score of the node and the strength of the connection between the nodes; the graph neural network uses a multi-layer graph neural network to process the input of the graph neural network; At each layer, the feature information of the node is updated according to the information of its neighboring nodes; after multiple rounds of iterations, the node's features fully integrate the information of the surrounding environment, thus obtaining a graph representation that fully reflects the structural characteristics of the power transmission and transformation project; The input of the adaptive meshing algorithm includes the graph representation output by the graph neural network and the original data of the three-dimensional model of the power transmission and transformation project; the output of the adaptive meshing algorithm is an optimized three-dimensional mesh model; the adaptive meshing algorithm first analyzes the graph representation, and determines the structurally complex areas or functionally critical areas based on the importance scores of the nodes in the graph representation and the strength of the edges between the nodes; dynamically adjusts the density of the three-dimensional model network, including increasing the mesh density for the determined structurally complex areas or functionally critical areas; and uses geometric modeling technology to optimize the mesh of the three-dimensional model to obtain an optimized three-dimensional mesh model.

3. The intelligent management system for power transmission and transformation project construction according to claim 2 is characterized in that: The graph neural network uses an aggregation function to synthesize all neighbor node information of node v in the graph neural network. The aggregation function AGGREGATE avg (v) This is achieved by calculating the average value of the neighbor node feature vectors using the following formula 1: Among them, N(v) is the set of neighbor nodes of node v, h u is the feature vector of neighbor node u; Using the output of the aggregation function as input, the feature vector of node v is updated through linear transformation and nonlinear activation function ReLU using the following formula 2: h′ v =ReLU(W·AGGREGATE avg (v)+b) (2) Among them, h′ v is the updated feature vector of node v; W and b are the weight matrix and bias vector in the graph neural network layer, respectively.

4. The intelligent management system for power transmission and transformation project construction according to claim 2 is characterized in that: The graph neural network is implemented using a three-layer graph convolutional neural network, where the feature dimensions of the output of each layer of the graph convolutional neural network are 64, 64, and 32 respectively.

5. The intelligent management system for power transmission and transformation project construction according to claim 1 is characterized in that: The data processing and anomaly detection unit uses the following formula 3 to identify the safety hazard level H of the construction site: risk : Among them, P ef，i represents the probability that the i-th data point in the construction site is in an abnormal state; I ds，j represents the quantitative value of the j-th construction status indicator provided by the digital twin model; w i YesP ef，i The weight of u j isI ds，j The weight of ; n is the total number of data points in the construction site; m is the total number of construction status indicators provided by the digital twin model; α and β are adjustment coefficients.

6. The intelligent management system for power transmission and transformation project construction according to claim 5 is characterized in that: The probability that the i-th data point in the construction site belongs to an abnormal state is calculated by the following steps: Preprocessing the sensor data related to the i-th data point in the construction site collected from the data acquisition unit to obtain preprocessed data; wherein the preprocessing includes data cleaning, standardization and feature extraction; The preprocessed data is input into a trained support vector machine to obtain the probability that the i-th data point in the construction site is in an abnormal state.

7. The intelligent management system for power transmission and transformation project construction according to claim 1 is characterized in that: The construction process intelligent optimization unit is specifically used for: Monitor the status and progress of the construction site and update the digital twin model in real time through sensor data collected by the data acquisition unit; Simulate multiple construction paths and resource allocation schemes in the updated digital twin model to obtain the expected performance of each construction path and resource allocation scheme under specific construction conditions, including expected cost, time and quality indicators; According to the performance, the trained decision tree model is applied to select the optimal construction path and resource allocation scheme from a variety of construction paths and resource allocation schemes.

8. The intelligent management system for power transmission and transformation project construction according to claim 7 is characterized in that: The input of the decision tree model includes weather conditions at the construction site, personnel distribution, equipment usage status, material inventory, each construction path and resource allocation plan, and the predicted performance of each construction path and resource allocation plan in the updated digital twin model; the output of the decision tree model is the optimal construction path and resource allocation plan.

Images