BIM-based power grid digital twin modeling method, device and storage medium

Abstract

The application discloses a BIM-based power grid digital twin modeling method and device and a storage medium, relates to the technical field of electronic data processing, and comprises the following steps: based on power grid project parameters, a BIM initialization model conforming to a specification is automatically generated from a parameterized BIM component library through a constraint satisfaction algorithm; multi-modal perception network is used to collect multi-source data with a unified space-time reference; a joint optimization objective function fusing geometry, semantics and physical laws is constructed, and a model optimal adjustment amount is obtained by solving the joint optimization objective function fusing geometry, semantics and physical laws by using the multi-source data, so as to generate a high-precision power grid BIM model; and real-time and historical data are injected into a hybrid simulation engine integrating data driving and a physical mechanism model, so as to drive the BIM model to generate a dynamic digital twin with real-time mapping and prediction capabilities. The application deeply fuses parameterized BIM of typical design and multi-source real-time data of a construction site, and constructs a power grid infrastructure digital twin capable of dynamic evolution and accurate mapping of a physical entity.

Description

Technical Field

[0001] This invention relates to the field of electronic data processing technology, specifically to a BIM-based digital twin modeling method, device, and storage medium for power grids. Background Technology

[0002] With the deepening of smart grid construction, the scale and complexity of power grid infrastructure are increasing, placing higher demands on the lean management of its entire lifecycle, including planning, construction, and operation. Digital twin technology, as a core means of achieving deep integration of cyber-physical systems, provides a new paradigm for the digital and intelligent transformation of power grid infrastructure. Currently, in the field of power grid engineering, Building Information Modeling (BIM) technology has been widely applied to 3D visualization design and clash detection, initially realizing the digitalization of the design phase.

[0003] However, existing BIM-based digital management methods still have significant limitations. First, traditional BIM models are mostly static 3D displays, and their generation heavily relies on manual modeling, which is inefficient and prone to errors, and cannot be quickly and automatically generated and adjusted according to the project's core parameters. Second, the model is disconnected from the physical state of the entity. Geometric deviations, structural deformations, and environmental changes during construction mainly rely on manual inspection and recording, making it difficult to perceive and map them to the digital model in real time through effective technical means, resulting in the model failing to accurately reflect the actual site conditions. Existing systems lack built-in simulation and extrapolation capabilities; the model is static and cannot perform predictive analysis of structural health and equipment status trends.

[0004] For example, Chinese Patent Publication No. CN119623093A discloses a method and apparatus for constructing a digital twin model of a power distribution network. The method includes: constructing a 3D real-world model of the power distribution network's ground environment in Unreal Engine using oblique photogrammetry; constructing a 3D BIM model of the underground cables and their channels within the 3D real-world model using parametric modeling technology; and performing simulation processing on the 3D BIM model using Unreal Engine to generate a digital twin model of the power distribution network. This application addresses the problem in existing technologies that cannot intuitively display the 3D spatial relationship between cables and their surrounding environment.

[0005] For example, Chinese patent CN118114480B discloses a multi-system compatible, fully perceptive, self-optimizing digital twin model for power grid projects and its data alignment method. The transformation and alignment of the BIM model and GIS data is based on linear affine transformation, with the underlying algorithm constructed and multi-parameterized local nonlinear optimization matching performed to obtain the main alignment algorithm framework. Simultaneously, it integrates with the power grid GIS (or other power grid data systems) based on self-developed data fusion technology, constructing consistent, collaborative, and comprehensive perception of power construction data for the entire lifecycle of power construction projects. This invention, based on the power construction BIM model and the power grid GIS system, has undergone secondary development, proposing an efficient geometric transformation and alignment method for BIM models and GIS data, achieving higher accuracy and efficiency in data alignment. Summary of the Invention

[0006] The technical problem to be solved by the present invention is to address the shortcomings of the existing technology by providing a BIM-based digital twin modeling method, equipment and storage medium for power grids.

[0007] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0008] The BIM-based digital twin modeling method for power grids includes the following steps:

[0009] Step S1: Based on the parameter vector of the power grid infrastructure project, a BIM initialization model that conforms to the pre-stored design specification constraint set is automatically matched and combined from the pre-built parametric BIM component library through the constraint satisfaction problem solving algorithm.

[0010] Step S2: Collect multi-source sensing data with a unified timestamp and spatial coordinate reference through a network of multimodal sensing devices deployed at the project site;

[0011] Step S3: Construct a joint optimization objective function. Using the multi-source sensing data as input, by solving the minimization problem of the joint optimization objective function, obtain the optimal adjustment amount of the BIM initialization model parameter vector, thereby generating a power grid BIM model consistent with the physical entity.

[0012] Step S4: Inject the real-time acquired multi-source sensing data and historical data into the hybrid simulation engine. The hybrid simulation engine integrates a data-driven prediction model and a physical mechanism model to drive the power grid BIM model and generate a dynamic digital twin of the power grid.

[0013] Furthermore, step S1 specifically includes the following steps:

[0014] Each component in the pre-built parametric BIM component library is defined as a variable whose value range is all the optional parametric specifications of that component;

[0015] The design specifications, electrical safety distances, spatial collision rules, and process logic in the design scheme are defined as constraints.

[0016] The model generation process is transformed into an optimization problem, the objective function of which is to minimize the total degree of violation of the constraints by the selected combination of components, the degree of violation being calculated by weighted sum of the weights of each constraint;

[0017] The optimization problem is solved using a genetic algorithm to obtain the optimal combination of components that satisfies all constraints and has the lowest degree of violation, thereby generating the BIM initialization model.

[0018] Furthermore, in step S2, the multi-source sensing data includes: geometric state point cloud data acquired by a laser scanning device, physical attribute time-series data acquired by an IoT sensor, and environmental data.

[0019] Furthermore, after acquiring the geometric state point cloud data, the method also includes point cloud preprocessing and semantic segmentation steps:

[0020] Statistical filtering algorithms are used to denoise and downsample the original point cloud.

[0021] The preprocessed point cloud is input into a pre-trained 3D deep learning semantic segmentation network. The semantic segmentation network assigns a semantic label to each point in the point cloud. The semantic label corresponds to the component category in the BIM model, thereby outputting a semantic point cloud with component category information.

[0022] Furthermore, in step S3, the joint optimization objective function is a weighted sum of geometric fitting terms, semantic consistency terms, and physical law terms;

[0023] The geometric fitting term is used to calculate the sum of squares of the point-to-surface distances between the surface of the power grid BIM model and the semantic point cloud data; the semantic consistency term is used to calculate the bidirectional Chamfer distance between the point cloud cluster of each semantic label and the spatial volume of the corresponding component in the BIM model; and the physical law term is used to calculate the mean square error between the IoT sensor measurement values ​​and the calculated values ​​based on the physical mechanism model.

[0024] Furthermore, the step of dynamically binding the physical attribute time-series data with the component attributes in the power grid BIM model includes:

[0025] In the power grid BIM model, a set of dynamic attributes is defined for each component object;

[0026] Establish a mapping table between IoT sensors and the dynamic attributes of components on the digital twin platform;

[0027] When data flows in, the sensor data stream is assigned to the corresponding component dynamic attributes in real time according to the mapping table, thereby completing the real-time update of the component status.

[0028] Furthermore, in step S4, the operation method of the hybrid simulation engine includes:

[0029] Using a long short-term memory network model, the historical sequence of dynamic attribute values ​​of components is used as input, and the output data drives the predicted values.

[0030] Using the current state of the power grid BIM model, the current values ​​of component attributes, and environmental data as boundary conditions, the physical mechanism model is invoked for calculation to obtain the calculated value of the physical mechanism model;

[0031] The Kalman filter algorithm is used to fuse the data-driven prediction value and the physical mechanism model calculation value to obtain the final predicted state value, which serves as the future inference state of the digital twin.

[0032] Furthermore, in step S4, the digital twin receives control commands through a predefined interactive interface and maps them to the physical entity actuator to form a closed-loop control.

[0033] The interactive interface provides an application programming interface for receiving control command vectors from external application systems. After the control command vectors are verified by the rule set in the digital twin, they are converted into control signals that can be recognized by the physical actuator through the Internet of Things gateway, thereby realizing reverse control of the physical entity.

[0034] A storage medium, characterized in that the storage medium stores instructions, which, when read by a computer, cause the computer to execute a BIM-based digital twin modeling method for power grids.

[0035] An electronic device, characterized in that it includes a processor and a storage medium, wherein the processor executes instructions in the storage medium.

[0036] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0037] 1. This invention, through a parametric component library and constraint satisfaction problem-solving algorithm, can automatically match and generate a BIM initialization model that conforms to the specifications based on key project parameters, thereby improving the efficiency and accuracy of model construction.

[0038] 2. This invention establishes a dynamic binding mechanism between sensor data and model components, realizing real-time synchronous mapping from the physical entity state to the digital twin, transforming the model from a static snapshot into a dynamic living entity, thus providing a foundation for real-time monitoring and diagnosis.

[0039] 3. This invention proposes a multi-objective joint optimization algorithm that integrates geometry, semantics and physical laws. It uses multi-source sensing data such as laser point cloud and IoT data to dynamically reverse correct the initial model, ensuring that the digital twin is consistent with the physical entity in terms of geometry, attributes and state in all weather conditions and with high precision.

[0040] 4. By defining a standard interaction interface, the analysis results and control commands received by the digital twin can be sent to the physical entity actuator after security verification, thereby forming a complete closed loop of "perception-analysis-decision-control". Attached Figure Description

[0041] Other features, objects, and advantages of the invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0042] Figure 1 This is a schematic diagram of the method flow according to an embodiment of the present invention;

[0043] Figure 2 This is a schematic diagram of the digital twin modeling system architecture according to an embodiment of the present invention;

[0044] Figure 3 This is a schematic diagram of the hybrid simulation engine workflow according to an embodiment of the present invention. Detailed Implementation

[0045] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0046] like Figure 1 As shown, the BIM-based digital twin modeling method for power grids includes the following steps:

[0047] Step S1: Based on the parameter vector of the power grid infrastructure project, a BIM initialization model that conforms to the pre-stored design specification constraint set is automatically matched and combined from the pre-built parametric BIM component library through the constraint satisfaction problem solving algorithm.

[0048] Step S2: Collect multi-source sensing data with a unified timestamp and spatial coordinate reference through a network of multimodal sensing devices deployed at the project site;

[0049] Step S3: Construct a joint optimization objective function. Using the multi-source sensing data as input, by solving the minimization problem of the joint optimization objective function, obtain the optimal adjustment amount of the BIM initialization model parameter vector, thereby generating a power grid BIM model consistent with the physical entity.

[0050] Step S4: Inject the real-time acquired multi-source sensing data and historical data into the hybrid simulation engine. The hybrid simulation engine integrates a data-driven prediction model and a physical mechanism model to drive the power grid BIM model and generate a dynamic digital twin of the power grid.

[0051] Step S1 specifically includes the following steps:

[0052] Each component in the pre-built parametric BIM component library is defined as a variable whose value range is all the optional parametric specifications of that component;

[0053] The design specifications, electrical safety distances, spatial collision rules, and process logic in the design scheme are defined as constraints.

[0054] The model generation process is transformed into an optimization problem, the objective function of which is to minimize the total degree of violation of the constraints by the selected combination of components, the degree of violation being calculated by weighted sum of the weights of each constraint;

[0055] The optimization problem is solved using a genetic algorithm to obtain the optimal combination of components that satisfies all constraints and has the lowest degree of violation, thereby generating the BIM initialization model.

[0056] Suppose a parametric BIM component library contains multiple components, each corresponding to a set of parametric specifications; define a set of constraints, including design specifications, electrical safety distances, and spatial collision rules; transform model generation into an optimization problem, with the objective function being:

[0057]

[0058] in, Indicates the constraint index. This indicates the total number of constraints. This represents the weight of the r-th constraint. This represents a project parameter vector, which represents the set of key parameters for a power grid infrastructure project. The component specification vector represents the set of parametric specifications for the j-th BIM component. Let r represent the r-th constraint. This indicates a violation of the degree function.

[0059] The formula for calculating the degree of violation is as follows:

[0060]

[0061] in, The constraint function represents the parametric specification of the r-th constraint and the j-th BIM element;

[0062] in, Each constraint is defined independently based on its specific logic and rules, as shown in Table 1.

[0063] Table 1

[0064]

[0065] Among them, electrical safety distance constraints: the minimum air gap between conductors and the structure must be greater than the value specified by the State Grid standard; spatial collision constraints: there must be no spatial interference between equipment or between equipment and building structure; process and layout constraints: the area of ​​the main transformer room must be able to accommodate the selected main transformer size and reserve maintenance passages; performance constraints: the total capacity of the main transformers in the substation must be greater than the maximum expected load planned for the project.

[0066] The optimal component combination is obtained by using a genetic algorithm to generate the BIM initialization model.

[0067] In step S2, the multi-source sensing data includes: geometric state point cloud data acquired by a laser scanning device, physical attribute time-series data acquired by an IoT sensor, and environmental data.

[0068] After acquiring the geometric state point cloud data, the process also includes point cloud preprocessing and semantic segmentation steps:

[0069] Statistical filtering algorithms are used to denoise and downsample the original point cloud.

[0070] The preprocessed point cloud is input into a pre-trained 3D deep learning semantic segmentation network. The semantic segmentation network assigns a semantic label to each point in the point cloud. The semantic label corresponds to the component category in the BIM model, thereby outputting a semantic point cloud with component category information.

[0071] Denoising: Statistical filtering algorithm is used to remove outliers;

[0072] Downsampling: Voxel grid filtering is used to reduce point cloud density.

[0073] Semantic segmentation is performed using a pre-trained PointNet++ network, outputting point clouds with semantic labels.

[0074] In step S3, the joint optimization objective function is a weighted sum of geometric fitting terms, semantic consistency terms, and physical law terms;

[0075] The geometric fitting term is used to calculate the sum of squares of the point-to-surface distances between the surface of the power grid BIM model and the semantic point cloud data; the semantic consistency term is used to calculate the bidirectional Chamfer distance between the point cloud cluster of each semantic label and the spatial volume of the corresponding component in the BIM model; and the physical law term is used to calculate the mean square error between the IoT sensor measurement values ​​and the calculated values ​​based on the physical mechanism model.

[0076] The formula for calculating the joint optimization objective function is as follows:

[0077]

[0078] in, This represents the joint optimization objective function. Represents the geometric fit term. Indicates semantic consistency terms. Represents physical laws. , and These represent the weight coefficients of the geometric fitting term, semantic consistency term, and physical law term, respectively.

[0079] The formula for calculating the geometric fit term is as follows:

[0080]

[0081] in, Represents points in a point cloud. Point At the nearest point on the model surface, Point The normal vector at that point, This represents the total number of points in the point cloud. The norm of a vector.

[0082] The formula for calculating the semantic consistency term is:

[0083]

[0084] in, Indicates semantic tag category, Indicates the total number of semantic tag categories. This represents the component in the BIM model corresponding to the semantic label L. This represents the set of point clouds with semantic label L. This represents the three-dimensional volume occupied by component C. The surface of component C is represented. Represents the Euclidean distance function. This represents a point on the surface of component C in the BIM model. This represents the total number of points in the point cloud set belonging to the semantic label L. This represents the total number of points sampled from the surface of component C in the BIM model.

[0085] The formula for calculating the physical law term is:

[0086]

[0087] in, Indicates the sensor index. Indicates the total number of sensors. This represents the confidence weight of the k-th sensor. This represents the measurement value of the k-th sensor. This represents the calculated value from the physical mechanism model.

[0088] In power grid infrastructure construction, the finite element method (FEM) is a common physical mechanism model, typically used to calculate the stress, strain, and displacement of structures under load. For example, it is used to analyze the deformation of structures under strong winds.

[0089] The objective function is minimized using gradient descent, iteratively adjusting the parameter vector of the BIM initialization model until the joint optimization objective function is achieved. The model converges to its minimum value, at which point the parameter vector of the model is the optimal adjustment amount.

[0090] The steps for dynamically binding the physical attribute time-series data with the component attributes in the power grid BIM model include:

[0091] In the power grid BIM model, a set of dynamic attributes is defined for each component object;

[0092] Establish a mapping table between IoT sensors and the dynamic attributes of components on the digital twin platform;

[0093] When data flows in, the sensor data stream is assigned to the corresponding component dynamic attributes in real time according to the mapping table, thereby completing the real-time update of the component status.

[0094] In step S4, the working method of the hybrid simulation engine includes:

[0095] Using a long short-term memory network model, the historical sequence of dynamic attribute values ​​of components is used as input, and the output data drives the predicted values.

[0096] Using the current state of the power grid BIM model, the current values ​​of component attributes, and environmental data as boundary conditions, the physical mechanism model is invoked for calculation to obtain the calculated value of the physical mechanism model;

[0097] The Kalman filter algorithm is used to fuse the data-driven prediction value and the physical mechanism model calculation value to obtain the final predicted state value, which serves as the future inference state of the digital twin.

[0098] In step S4, the digital twin receives control commands through a predefined interactive interface and maps them to the physical entity actuator to form a closed-loop control.

[0099] The interactive interface provides an application programming interface for receiving control command vectors from external application systems. After the control command vectors are verified by the rule set in the digital twin, they are converted into control signals that can be recognized by the physical actuator through the Internet of Things gateway, thereby realizing reverse control of the physical entity.

[0100] The specific definition of the digital twin is:

[0101]

[0102] in, This represents the complete form of the dynamic digital twin of the power grid at any time t within its lifecycle. This represents the power grid BIM model after joint optimization at time t. This represents the real-time physical state and operational attributes of all components in the digital twin at time t, driven by real-time sensor data. This represents a database that records all historical state data from the initial time to the current time t. This represents a predefined set of rules stored within the digital twin, which specifies the behavioral logic and constraints of the digital twin. This indicates a hybrid simulation engine, which integrates both data-driven and physical mechanism-based models into its simulation computation core. The interaction interface represents the channel through which a digital twin interacts bidirectionally with the outside world.

[0103] The formula for the digital twin defines a living, holographic, and computable digital twin:

[0104] in, and It accurately describes the static and dynamic states at time t; It records the complete life course; It is about understanding the rules of the world; It endows individuals with the ability to analyze, think critically, and predict the future; This enables it to interact with and control the physical world.

[0105] like Figure 2As shown, the overall system architecture of this invention is divided into four layers: a perception layer, a data layer, a model layer, and an application layer. The perception layer is used for data acquisition via laser scanners, etc. The data layer, located in the middle, includes a data receiving module, a point cloud preprocessing module, and a data fusion module. The model layer, located at the center of the architecture, includes a parametric BIM component library, an initialization model generator, a joint optimization objective function solver, and a hybrid simulation engine. This layer receives data processed by the data layer and outputs a dynamic digital twin of the power grid. The application layer, located at the top layer, can include 3D visualization, displaying the BIM model, simulation analysis (such as charts displaying predicted curves), and decision support (such as dashboards and alarm prompts). Physical entities interact with the perception and application layers to complete the transmission of commands and perceived data.

[0106] like Figure 3 The diagram illustrates the core workflow of the hybrid simulation engine. The engine simultaneously receives historical data archives, the current state of the digital twin, and real-time environmental data as input. It generates predicted values ​​for the future state through two parallel channels: data-driven prediction and physical mechanism calculation. These two prediction results are then weighted and fused at the fusion center using a Kalman filter algorithm to obtain a more accurate and reliable optimal future state prediction. This result is used to update the current state of the digital twin and is also archived in the historical database to provide data support for the next round of predictions, thus forming a self-iteratory, continuously optimizing closed-loop system.

[0107] The core process specifically includes:

[0108] The engine receives three types of input data: historical data, current status, and environmental data;

[0109] The left channel uses an LSTM model for data-driven prediction to generate predicted values.

[0110] The right channel uses a physical mechanism model to generate calculated values;

[0111] The two prediction results are fused using a Kalman filter formula;

[0112] The generated optimal prediction is used to update both the twin state and archived historical data.

[0113] Any combination of one or more computer-readable media may be used. A computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. A computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples (a non-exhaustive list) of computer-readable storage media include: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0114] The examples described herein are merely preferred embodiments of the invention and are not intended to limit the concept and scope of the invention. Any modifications and improvements made by those skilled in the art to the technical solutions of the invention without departing from the design concept of the invention should fall within the protection scope of the invention.

Claims

1. A BIM-based digital twin modeling method for power grids, characterized in that, Includes the following steps: Step S1: Based on the parameter vector of the power grid infrastructure project, a BIM initialization model that conforms to the pre-stored design specification constraint set is automatically matched and combined from the pre-built parametric BIM component library through the constraint satisfaction problem solving algorithm. Step S2: Collect multi-source sensing data with a unified timestamp and spatial coordinate reference through a network of multimodal sensing devices deployed at the project site; Step S3: Construct a joint optimization objective function. Using the multi-source sensing data as input, by solving the minimization problem of the joint optimization objective function, obtain the optimal adjustment amount of the BIM initialization model parameter vector, thereby generating a power grid BIM model consistent with the physical entity. Step S4: Inject the real-time acquired multi-source sensing data and historical data into the hybrid simulation engine. The hybrid simulation engine integrates a data-driven prediction model and a physical mechanism model to drive the power grid BIM model and generate a dynamic digital twin of the power grid. In step S1, the parametric BIM component library contains multiple components, each corresponding to a set of parametric specifications; a set of constraints is defined, including design specifications, electrical safety distances, spatial collision rules, etc.; the model generation is transformed into an optimization problem, with the objective function being: in, Indicates the constraint index. This indicates the total number of constraints. This represents the weight of the r-th constraint. This represents a project parameter vector, which represents the set of key parameters for a power grid infrastructure project. The component specification vector represents the set of parametric specifications for the j-th BIM component. This represents the r-th constraint. Indicates the degree of violation function; The formula for calculating the degree of violation function is as follows: in, The constraint function represents the parametric specification of the r-th constraint and the j-th BIM element; In step S3, the joint optimization objective function is a weighted sum of geometric fitting terms, semantic consistency terms, and physical law terms; The geometric fitting term is used to calculate the sum of squares of the point-to-surface distances between the surface of the power grid BIM model and the semantic point cloud data; the semantic consistency term is used to calculate the bidirectional Chamfer distance between the point cloud cluster of each semantic tag and the spatial volume of the corresponding component in the power grid BIM model; and the physical law term is used to calculate the mean square error between the IoT sensor measurement values ​​and the calculated values ​​based on the physical mechanism model. In step S4, the working method of the hybrid simulation engine includes: Using a long short-term memory network model, the historical sequence of dynamic attribute values ​​of components is used as input, and the output data drives the predicted values. Using the current state of the power grid BIM model, the current values ​​of component attributes, and environmental data as boundary conditions, the physical mechanism model is invoked for calculation to obtain the calculated value of the physical mechanism model; The Kalman filter algorithm is used to fuse the data-driven prediction value and the physical mechanism model calculation value to obtain the final predicted state value, which serves as the future projection state of the power grid dynamic digital twin. In step S4, the power grid dynamic digital twin receives control commands through a predefined interactive interface and maps them to physical entity actuators to form closed-loop control. The interactive interface provides an application programming interface for receiving control command vectors from external application systems. After the control command vectors are verified by the rule set in the power grid dynamic digital twin, they are converted into control signals that can be recognized by the physical entity actuators through the Internet of Things gateway, thereby realizing reverse control of the physical entities.

2. The method according to claim 1, characterized in that, Step S1 specifically includes the following steps: Each component in the pre-built parametric BIM component library is defined as a variable whose value range is all the optional parametric specifications of that component; The design specifications, electrical safety distances, spatial collision rules, and process logic in the design scheme are defined as constraints. The model generation process is transformed into an optimization problem, the objective function of which is to minimize the degree of violation of the constraints by the selected combination of components, the degree of violation being calculated by weighted sum of the weights of each constraint. The optimization problem is solved using a genetic algorithm to obtain the optimal combination of components that satisfies all constraints and has the lowest degree of violation, thereby generating the BIM initialization model.

3. The method according to claim 2, characterized in that, In step S2, the multi-source sensing data includes: geometric state point cloud data acquired by a laser scanning device, physical attribute time-series data acquired by an IoT sensor, and environmental data.

4. The method according to claim 3, characterized in that, After acquiring the geometric state point cloud data, the process also includes point cloud preprocessing and semantic segmentation steps: Statistical filtering algorithms are used to denoise and downsample the original point cloud. The preprocessed point cloud is input into a pre-trained 3D deep learning semantic segmentation network. The 3D deep learning semantic segmentation network assigns a semantic label to each point in the point cloud. The semantic label corresponds to the component category in the power grid BIM model, thereby outputting a semantic point cloud with component category information.

5. The method according to claim 4, characterized in that, The steps for dynamically binding the physical attribute time-series data with the component attributes in the power grid BIM model include: In the power grid BIM model, a set of dynamic attributes is defined for each component object; Establish a mapping table between IoT sensors and the dynamic attributes of components on the power grid dynamic digital twin platform; When data flows in, the sensor data stream is assigned to the corresponding component dynamic attributes in real time according to the mapping table, thereby completing the real-time update of the component status.

6. A storage medium, characterized in that, The storage medium stores instructions, which, when read by a computer, cause the computer to execute the BIM-based digital twin modeling method for power grids as described in any one of claims 1-5.

7. An electronic device, characterized in that, It includes a processor and the storage medium of claim 6, wherein the processor executes instructions in the storage medium.

Images