Digital twinborn modeling method and system for distribution network tower

Abstract

The invention provides a digital twinborn modeling method and system for a distribution network tower. The method comprises the following steps: carrying out classified statistical analysis on geometric structure characteristic data of a target tower to determine a parameter distribution proportion of each component; modeling samples are selected from the tower point cloud database to construct a three-dimensional modeling sample set, and it is ensured that component parameter distribution in the sample set is consistent with the statistical proportion; and generating a model construction instruction based on the sample set, driving a digital twin modeling engine to perform tower reconstruction, and collecting performance data in a modeling process to generate a quality evaluation index. According to the method, accurate and efficient reconstruction of the tower model is realized through parameterized sample selection and standardized instruction generation, and a quantitative evaluation basis is provided for modeling quality. According to the method, the construction precision and efficiency of the digital twinborn model of the distribution network tower can be improved, and modeling evaluation and optimization are supported.

Description

Technical Field

[0001] The present invention relates to the technical field of digital modeling of power facilities, and more specifically, to a method and system for digital twin modeling of distribution network towers. Background Art

[0002] Distribution towers are critical infrastructure in power systems, and their operational status is directly related to the stability and security of power supply. With the continuous development of power systems and the advancement of digital transformation, higher requirements are being placed on the management and maintenance of distribution towers. Traditional distribution tower modeling methods rely primarily on manual measurement and two-dimensional drawings. This approach is not only inefficient but also fails to accurately reflect the tower's actual three-dimensional structure and operational status. In recent years, the development of 3D laser scanning and digital twin technologies has provided new ideas and methods for the modeling and management of distribution towers. 3D laser scanning technology can quickly acquire point cloud data of towers, while digital twin technology processes and analyzes this data to construct a digital twin model of the tower, enabling real-time monitoring and intelligent management of the tower. However, existing digital twin modeling methods have some shortcomings when dealing with distribution towers. For example, the modeling process typically requires a large amount of computing resources and has low modeling efficiency, making it difficult to meet the needs of large-scale distribution tower modeling. Furthermore, existing methods have limitations in modeling accuracy and fidelity, failing to accurately reflect the complex structure and subtle defects of towers.

[0003] In the process of implementing the embodiments of the present invention, the inventors discovered that there are at least the following problems or defects in the prior art: When processing the modeling of distribution network towers, the prior art lacks systematic classification and analysis of the geometric structure feature data of the towers, resulting in insufficient accuracy and representativeness in the selection of modeling samples, which in turn affects the accuracy and efficiency of modeling. At the same time, the existing methods fail to fully utilize distributed computing and parallel processing technologies in the modeling process, resulting in low execution efficiency of modeling tasks and difficulty in adapting to the processing requirements of large-scale data. In addition, the prior art lacks a comprehensive and effective evaluation index system for modeling quality evaluation, and is unable to accurately evaluate the performance of the modeling process and the fidelity of the model, thereby limiting the reliability and credibility of digital twin models in practical applications. Summary of the Invention

[0004] The present invention provides a method, system and equipment for digital twin modeling of distribution network towers.

[0005] In a first aspect of the present invention, a method for modeling a digital twin of a distribution network tower is provided, comprising:

[0006] Conduct statistical analysis on the geometric structural characteristic data of the target tower to determine the parameter distribution ratio of various tower components;

[0007] Selecting modeling samples from the tower point cloud database to construct a three-dimensional modeling sample set; wherein the parameter distribution of various tower components in the sample set is consistent with the parameter distribution ratio;

[0008] Based on the three-dimensional modeling sample set, a model construction instruction is generated, the digital twin modeling engine is driven by the instruction to reconstruct the tower, and the performance data of the modeling process is collected to generate a modeling quality evaluation index.

[0009] Furthermore, before performing a modeling operation based on the three-dimensional modeling sample set, the method further includes:

[0010] Calculate the modeling operation area based on the topological distribution density of the target tower and the estimated modeling scale, and determine the number of parallel modeling threads based on the tower modeling priority and the operation area;

[0011] configuring distributed modeling nodes according to the number of parallel modeling threads and the sample capacity of the three-dimensional modeling sample set;

[0012] The generating of the model building instruction based on the three-dimensional modeling sample set includes: loading the three-dimensional modeling samples in parallel through the distributed modeling nodes;

[0013] Each distributed modeling node synchronously generates model building instructions and submits modeling tasks to the digital twin modeling engine.

[0014] Furthermore, generating a model building instruction based on the three-dimensional modeling sample set includes:

[0015] The sample data is formatted according to the IEC61970 standard to generate standardized point cloud data;

[0016] Encapsulate standardized point cloud data into model building instructions that comply with the OPC UA protocol.

[0017] Furthermore, before generating the model building instruction, the method further includes:

[0018] Based on the tower defect feature library, abnormal point cloud filtering is performed on the modeling samples.

[0019] Furthermore, the collection of modeling process performance data to generate modeling quality evaluation indicators includes:

[0020] Record the sending timestamp t of the model building instruction send ;

[0021] Capture the receiving timestamp t of the reconstruction result returned by the digital twin modeling engine recv ;

[0022] Calculate the single modeling delay Δt=t recv -tsend ;

[0023] The modeling delay data of all samples are counted to generate a delay distribution feature matrix.

[0024] Furthermore, the capturing of the receiving timestamp of the returned reconstruction result includes: recording the initial point cloud feedback timestamp t first ; and / or record the complete grid model receiving timestamp t last ;

[0025] The calculation modeling delay includes: calculating the initial response delay Δt first =t first -t send ; and / or calculate the complete modeling period Δt last =t last -t send ;

[0026] The generation of quality evaluation indicators includes: constructing an initial response delay histogram and a complete modeling cycle probability density distribution diagram.

[0027] Furthermore, it also includes:

[0028] Get the number of characteristic topologies N of the benchmark model std , the benchmark model is an ideal tower model generated under no-load condition;

[0029] Extract the number of feature topologies N output by real-time modeling actual ;

[0030] Calculating model fidelity

[0031] The fidelity data of all samples is counted to generate the model accuracy evaluation index.

[0032] Furthermore, it also includes:

[0033] Obtain the baseline feature number N under QoS guarantee conditions QoS ;

[0034] The number of features N generated by statistical real-time modeling within the service level agreement (SLA) time limit SLA ;

[0035] Calculating the service degradation index

[0036] Generate a service reliability evaluation matrix.

[0037] In a second aspect of the present invention, a distribution network tower digital twin modeling system is provided, comprising:

[0038] Statistical processing unit, used to perform classified statistical analysis on the geometric structural characteristic data of the target tower and determine the parameter distribution ratio of various tower components;

[0039] A modeling data set construction unit is used to select modeling samples from the tower point cloud database to construct a three-dimensional modeling sample set, wherein the parameter distribution of various tower components in the sample set is consistent with the parameter distribution ratio;

[0040] A modeling processing unit is used to generate model construction instructions based on the three-dimensional modeling sample set, drive the digital twin modeling engine to reconstruct the tower through the instructions, and collect modeling process performance data to generate modeling quality evaluation indicators.

[0041] In a third aspect of the present invention, a computing device is provided, comprising: at least one processor, a memory, and an input / output unit; wherein the memory is used to store a computer program, and the processor is used to call the computer program stored in the memory to execute any one of the methods described in the first aspect.

[0042] According to the above-mentioned embodiments of the present invention, there are at least the following beneficial effects: the digital twin modeling method and system of the distribution network tower of the present invention can improve the modeling efficiency and accuracy. By classifying and statistically analyzing the geometric structural feature data of the target tower, the parameter distribution ratio of each type of tower component can be accurately determined, thereby constructing a more representative three-dimensional modeling sample set, providing basic data guarantee for high-quality modeling. At the same time, by utilizing distributed modeling nodes and parallel processing technology, the number of parallel modeling threads can be flexibly configured according to the modeling operation area, modeling priority and sample capacity, so as to achieve efficient parallel processing of modeling tasks, greatly shorten the modeling time, improve the modeling efficiency, and meet the needs of large-scale distribution network tower modeling. In addition, the sample data is formatted according to the IEC61970 standard and encapsulated as a model building instruction that complies with the OPCUA protocol, which can ensure the standardization and normalization of the modeling data, improve the compatibility and stability of the modeling process, and further improve the modeling accuracy.

[0043] The present invention can also comprehensively and accurately evaluate the modeling quality and model performance. By recording the sending timestamp of the model building instruction and the receiving timestamp of the reconstruction result returned by the digital twin modeling engine, calculating the single modeling delay, and statistically analyzing the modeling delay data of all samples to generate a delay distribution feature matrix, it can intuitively reflect the timeliness of the modeling process and provide a basis for optimizing modeling parameters and resource allocation. At the same time, the number of characteristic topologies of the baseline model is obtained and compared with the number of characteristic topologies output by real-time modeling, the model fidelity is calculated, and the fidelity data of all samples are statistically analyzed to generate a model accuracy evaluation index, which can effectively evaluate the degree of restoration of the model to the actual tower structure and ensure the accuracy and reliability of the digital twin model. In addition, the service degradation index is calculated in combination with the service level agreement (SLA) time limit, and a service reliability evaluation matrix is ​​generated, which can further evaluate the stability and reliability of the modeling system and provide strong support for the digital operation and maintenance of the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] The above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily apparent by reading the following detailed description with reference to the accompanying drawings, in which several embodiments of the present invention are shown by way of example and not limitation, in which:

[0045] Figure 1 A schematic diagram of a flow chart of a distribution network tower digital twin modeling method provided by one embodiment of the present invention;

[0046] Figure 2 A schematic diagram of the structure of a distribution network tower digital twin modeling system provided by one embodiment of the present invention;

[0047] Figure 3 The figure schematically shows a structural diagram of a computing device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0048] The principles and spirit of the present invention will be described below with reference to several exemplary embodiments. It should be understood that these embodiments are provided solely to enable those skilled in the art to better understand and implement the present invention, and are not intended to limit the scope of the present invention in any way. Rather, these embodiments are provided to make the present invention more thorough and complete, and to fully convey the scope of the present invention to those skilled in the art.

[0049] Those skilled in the art will appreciate that the embodiments of the present invention may be implemented as a system, apparatus, device, method, or computer program product. Therefore, the present invention may be implemented in the following forms: entirely in hardware, entirely in software (including firmware, resident software, microcode, etc.), or in a combination of hardware and software.

[0050] It should be noted that any number of elements in the drawings is for illustration only and not for limitation, and any naming is only for distinction and does not have any limiting meaning.

[0051] Reference below Figure 1 , Figure 1 This is a flow chart of a method for modeling a digital twin of a distribution network tower provided by one embodiment of the present invention. Figure 1 As shown in FIG, a distribution network tower digital twin modeling method includes:

[0052] S1 performs classification and statistical analysis on the geometric structural characteristic data of the target tower to determine the parameter distribution ratio of various tower components;

[0053] S2 selects modeling samples from the tower point cloud database to construct a three-dimensional modeling sample set; wherein the parameter distribution of various tower components in the sample set is consistent with the parameter distribution ratio;

[0054] S3 generates model construction instructions based on the three-dimensional modeling sample set, drives the digital twin modeling engine to reconstruct the tower through the instructions, and collects modeling process performance data to generate modeling quality evaluation indicators.

[0055] It should be noted that the present invention proposes a digital twin modeling method for distribution network towers. The core of this method is to determine the parameter distribution ratios of various tower components through classified statistical analysis of the geometric structural characteristic data of the target tower. The geometric structural characteristic data here refers to data related to the tower structure, such as the shape, size, and spatial position of each tower component. This data can comprehensively reflect the physical form and structural characteristics of the tower. Through classified statistical analysis of this data, the proportion of different types of tower components in the overall tower can be clarified, providing a scientific basis for the selection of subsequent modeling samples. For example, for a common distribution network tower, its components may include the tower body, crossarms, insulators, etc. Through analysis, it can be determined that crossarms account for approximately 30% of the tower components, and insulators account for approximately 20%. Such parameter distribution ratios can help us construct a more targeted modeling sample set, ensuring that the sample set can truly reflect the structural characteristics of the target tower.

[0056] Specifically, modeling samples are selected from a tower point cloud database to construct a 3D modeling sample set, where the parameter distribution of various tower components in the sample set is consistent with the parameter distribution ratio. The tower point cloud database here refers to a database that stores a large amount of tower point cloud data. This point cloud data is acquired through technologies such as laser scanning and accurately represents the tower's 3D spatial information. The modeling samples are a subset of data selected from this database for use in constructing the digital twin model. These samples must represent the overall characteristics of the target tower. For example, if the target tower is a complex high-voltage distribution network tower, the modeling samples should include point cloud data for different parts of the tower body, different types of crossarms, and various components such as insulators. The proportion of these components in the sample set should match the parameter distribution ratio obtained in the previous analysis. For example, if the tower body components account for 40%, then the point cloud data for the tower body should also account for approximately 40% of the modeling sample set. The 3D modeling sample set constructed in this way provides high-quality input data for the digital twin modeling engine, thereby generating a more accurate tower digital twin model.

[0057] Preferably, model building instructions are generated based on the 3D modeling sample set. These instructions drive the digital twin modeling engine to reconstruct the tower, and performance data from the modeling process is collected to generate modeling quality evaluation metrics. In actual operation, the process of generating model building instructions can be further refined. For example, specific algorithms can be used to preprocess the point cloud data in the 3D modeling sample set, such as filtering and noise reduction, to remove outliers and interference information from the data and improve data quality. Then, based on the preprocessed data, model building instructions are generated using geometric modeling algorithms or machine learning algorithms. These instructions contain detailed information such as the shape, size, and position of each tower component, guiding the digital twin modeling engine to reconstruct the tower according to these parameters. During the modeling process, performance data of the modeling engine, such as computation time and memory usage, can be collected in real time. This data will be used to generate modeling quality evaluation metrics. For example, modeling efficiency can be evaluated by calculating the execution time of the modeling task, and model accuracy can be assessed by comparing the similarity between the generated model and the actual tower. Through these detailed steps and parameter settings, the efficiency of the digital twin modeling process and the quality of the model can be ensured, providing strong support for the digital management and maintenance of distribution network towers.

[0058] In some embodiments, before performing a modeling operation based on the three-dimensional modeling sample set, the method further includes:

[0059] Calculate the modeling operation area based on the topological distribution density of the target tower and the estimated modeling scale, and determine the number of parallel modeling threads based on the tower modeling priority and the operation area;

[0060] configuring distributed modeling nodes according to the number of parallel modeling threads and the sample capacity of the three-dimensional modeling sample set;

[0061] The generating of the model building instruction based on the three-dimensional modeling sample set includes: loading the three-dimensional modeling samples in parallel through the distributed modeling nodes;

[0062] Each distributed modeling node synchronously generates model building instructions and submits modeling tasks to the digital twin modeling engine.

[0063] It should be noted that in the digital twin modeling method for distribution network towers of the present invention, in order to improve modeling efficiency and adapt to large-scale modeling needs, a series of preparatory steps are performed before modeling operations based on a 3D modeling sample set. These preparatory steps mainly include calculating the modeling operation area based on the topological distribution density of the target tower and the estimated modeling scale, determining the number of parallel modeling threads based on the tower modeling priority and the operation area, and configuring distributed modeling nodes based on the number of parallel modeling threads and the sample capacity of the 3D modeling sample set. Here, the topological distribution density refers to the number of tower components distributed per unit area or unit volume, which reflects the complexity of the tower structure; the estimated modeling scale is the size of the modeling task estimated in advance based on actual needs, including factors such as the number of towers to be modeled and the tower complexity; the modeling operation area refers to the specific work area divided during the modeling process to rationally allocate modeling tasks; the number of parallel modeling threads refers to the number of threads performing modeling tasks simultaneously; parallel processing can improve modeling efficiency; and distributed modeling nodes refer to multiple computing nodes distributed in the network that work together to complete the modeling task. Through these steps, efficient parallel processing of modeling tasks can be achieved, modeling efficiency can be improved, and the needs of large-scale distribution network tower modeling can be met.

[0064] Specifically, the topological distribution density can be obtained by analyzing the point cloud data of the target tower. For example, it can be approximated by calculating the number of point cloud data points per unit volume of the tower components. Assuming that the number of point cloud data points in a certain area is 1,000, and the volume of the area is 1 cubic meter, the topological distribution density of the area is 1,000 points / cubic meter. The estimated modeling scale can be set based on actual needs. For example, if 100 towers need to be modeled, each with roughly the same complexity, the estimated modeling scale can be set to 100 unit tasks. The modeling work area can be divided based on the topological distribution density and the estimated modeling scale. For example, the entire modeling task can be divided into 10 work areas, each responsible for modeling 10 towers. The number of parallel modeling threads can be set based on the number of work areas and the workload of each work area. For example, allocating 2 parallel modeling threads to each work area will require a total of 20 parallel modeling threads. The configuration of distributed modeling nodes can be adjusted based on the number of parallel modeling threads and the sample size of the 3D modeling sample set. For example, if each modeling thread needs to process 100MB of sample data and each distributed modeling node has 1GB of memory, each distributed modeling node can handle the tasks of 10 modeling threads simultaneously. Through these specific parameter settings and concept definitions, efficient parallel processing of modeling tasks can be ensured, improving modeling efficiency.

[0065] Preferably, when generating model building instructions based on a set of 3D modeling samples, the 3D modeling samples can be loaded in parallel via distributed modeling nodes. Specifically, a multi-threaded or multi-process approach can be employed to allow each distributed modeling node to load multiple modeling samples simultaneously, improving data loading efficiency. For example, assuming each modeling sample is 50MB in size and each distributed modeling node has a memory bandwidth of 100MB / s, each distributed modeling node can load a modeling sample in 0.5 seconds. Through parallel loading, multiple distributed modeling nodes can complete the loading task of a large number of modeling samples in a short period of time. After loading is complete, each distributed modeling node synchronously generates model building instructions and submits the modeling task to the digital twin modeling engine. During this process, a synchronization mechanism can be employed to ensure that the instruction generation and submission operations of all modeling nodes are coordinated and consistent. For example, technologies such as message queues or distributed locks can be used to ensure that each modeling node submits the generated instructions to the modeling engine in a timely and accurate manner. In this way, modeling efficiency can be further improved, ensuring the rapid completion of large-scale distribution network tower modeling tasks.

[0066] In some embodiments, generating a model building instruction based on the three-dimensional modeling sample set includes:

[0067] The sample data is formatted according to the IEC61970 standard to generate standardized point cloud data;

[0068] Encapsulate standardized point cloud data into model building instructions that comply with the OPC UA protocol.

[0069] It should be noted that the present invention adopts a method of data format specification and protocol encapsulation in the process of generating model building instructions. Specifically, the sample data is formatted and processed in accordance with the IEC61970 standard to generate standardized point cloud data, and then these standardized point cloud data are encapsulated into model building instructions that comply with the OPCUA protocol. IEC61970 is an international standard for regulating the exchange and sharing of power system data, which ensures data compatibility and interoperability between different systems. OPCUA (Open Platform Communications Unified Architecture) is an industrial communication protocol for achieving secure and reliable data exchange between different devices and systems. In this way, it can be ensured that the generated model building instructions can be correctly parsed and executed by the digital twin modeling engine, thereby improving the accuracy and efficiency of modeling. This method of data processing and protocol encapsulation can effectively solve the compatibility issues between different data sources and ensure the smooth progress of the modeling process.

[0070] Specifically, the IEC61970 standard defines a series of data models and interface specifications for describing information such as equipment, network topology, and parameters of power systems. In the present invention, when formatting sample data, the geometric structural feature data of the tower is converted into a format that complies with the IEC61970 standard. For example, information such as the tower's component type, size, and location is converted into a standard data model, such as the device class and attributes in the CIM (Common Information Model). The OPCUA protocol defines the data transmission format and communication mechanism, including data encoding, encryption, and authentication. When encapsulating model building instructions, the standardized point cloud data is encapsulated in the format of the OPCUA protocol to generate instructions that can be recognized and processed by the digital twin modeling engine. For example, each model building instruction may contain information such as the unique identifier of the tower component, geometric shape parameters, and spatial position coordinates. This information is organized and transmitted using the specific data types and structures of the OPCUA protocol. In this way, the standardization and normalization of modeling instructions can be ensured, improving the compatibility and reliability of the modeling system.

[0071] Preferably, the data processing and packaging steps can be further refined during the generation of model building instructions. For example, when formatting sample data, specific algorithms can be used to preprocess the point cloud data, such as filtering, noise reduction, and feature extraction. These preprocessing steps can remove outliers and interference from the data, improving data quality. When packaging the model building instructions, appropriate data encoding methods and transmission parameters can be selected based on the specific requirements of the modeling task. For example, for complex tower structures, more sophisticated data encoding methods can be used to ensure that the modeling instructions contain sufficient detailed information. Furthermore, instruction transmission parameters, such as transmission rate and packet size, can be adjusted based on the modeling engine's performance and network conditions to optimize instruction transmission efficiency. Furthermore, metadata information, such as the modeling task priority and timestamp, can be added when packaging the instructions to facilitate task scheduling and management when the modeling engine processes the instructions. Through these refined steps and parameter settings, the quality of the model building instructions and the performance of the modeling system can be further improved.

[0072] In some embodiments, before generating the model building instruction, the method further includes:

[0073] Based on the tower defect feature library, abnormal point cloud filtering is performed on the modeling samples.

[0074] It should be noted that the present invention introduces an abnormal point cloud filtering processing step based on the tower defect feature library before generating the model building instructions. The tower defect feature library is a data set containing common tower defect features. These defect features may include abnormal conditions such as cracks, deformation, corrosion, etc. of tower components. By comparing with this feature library, abnormal point cloud data in the modeling sample can be identified and filtered out, thereby improving the accuracy and reliability of modeling. Abnormal point cloud data refers to point cloud data that is inconsistent with the normal tower structure due to measurement errors, environmental interference or actual defects of the tower. After filtering out these abnormal point cloud data, it can ensure that the data used for modeling more accurately reflects the actual structure of the tower, avoiding modeling errors caused by abnormal data.

[0075] Specifically, the tower defect signature library is a pre-established database that stores characteristic information for various tower defects. This characteristic information can include parameters such as defect shape, size, and location. For example, a crack defect may have specific length, width, and orientation characteristics; a corrosion defect may manifest as irregular missing points or density variations in surface point cloud data. When filtering out anomalous point cloud data, the point cloud data in the modeling sample is matched against features in the defect signature library. If the similarity between the sample point cloud data and a feature in the defect signature library exceeds a set threshold, for example, a similarity exceeding 80%, the point cloud data is considered anomalous and filtered out. This method effectively removes noise and interference from the modeling sample, improving the quality of the modeled data. Furthermore, anomalous point cloud filtering is a data processing process that uses specific algorithms or rules to identify and remove data points that do not conform to normal tower structural characteristics. This process can be based on the geometric or statistical characteristics of the point cloud data, or by comparing it with known defect signatures.

[0076] Preferably, when filtering out abnormal point clouds, a multi-step filtering algorithm can be employed to improve filtering accuracy. First, statistical filtering methods, such as mean filtering or median filtering, can be used to remove obvious noise points. For example, for each point cloud data point, the mean or median of the point cloud data within a certain range surrounding it is calculated. If a data point deviates from this mean or median by more than a set threshold, such as more than 10%, it is considered a noise point and filtered out. Next, geometric feature-based filtering methods, such as curvature filtering or normal filtering, can be used to identify and filter out potentially defective point cloud data. For example, the curvature or normal direction of each point cloud data point is calculated. If a point's curvature or normal direction differs significantly from the average curvature or normal direction of the surrounding points, such as more than 20%, it is considered a possible defective point cloud data and compared with the tower defect signature database to determine whether it needs to be filtered out. This multi-step filtering algorithm can more accurately identify and remove abnormal point cloud data, thereby improving the quality and accuracy of modeling data.

[0077] In some embodiments, the collecting modeling process performance data to generate modeling quality evaluation indicators includes:

[0078] Record the sending timestamp t of the model building instruction send ;

[0079] Capture the receiving timestamp t of the reconstruction result returned by the digital twin modeling engine recv ;

[0080] Calculate the single modeling delay Δt=t recv -t send ;

[0081] The modeling delay data of all samples are counted to generate a delay distribution feature matrix.

[0082] It should be noted that when collecting the performance data of the modeling process and generating the modeling quality evaluation index, the present invention calculates the single modeling delay by recording the sending timestamp of the model building instruction and the receiving timestamp of the reconstruction result returned by the digital twin modeling engine, and generates a delay distribution feature matrix by counting the modeling delay data of all samples. The timestamp here refers to a mark at a specific time point, which is used to record the specific time when the event occurs. By recording the timestamps of sending instructions and receiving results, the time required for the modeling process, that is, the modeling delay, can be accurately calculated. Modeling delay is an important indicator for measuring modeling efficiency, and the delay distribution feature matrix can reflect the delay distribution of different modeling samples, providing a basis for optimizing the modeling process and resource allocation.

[0083] Specifically, the sending timestamp of a model building instruction refers to the time recorded when the instruction was sent from the client or distributed modeling node during the modeling process. For example, when a modeling task is assigned to a distributed modeling node, the node records a timestamp at the moment the modeling instruction was sent, typically in the format of year-month-day hour:minute:second.millisecond. Similarly, the receiving timestamp of the reconstruction result returned by the digital twin modeling engine refers to the time recorded when the modeling engine completes the modeling task and returns the result to the client or modeling node. The difference between these two timestamps is the single modeling latency, which reflects the total time required from the start to completion of the modeling task. For example, if the sending timestamp is 2025-04-12 10:00:00.000 and the receiving timestamp is 2025-04-12 10:00:05.000, the single modeling latency is 5 seconds. By statistically analyzing the modeling latency data of all modeling samples, a latency distribution feature matrix can be generated. This matrix contains the modeling latency information for each sample, which can be used to analyze the execution efficiency and stability of the modeling task.

[0084] Preferably, when generating the modeling quality evaluation index, the processing and analysis of the delay data can be further refined. For example, the modeling delay data of all modeling samples can be classified and counted to generate a delay distribution histogram. The horizontal axis of the histogram represents the interval of the modeling delay, and the vertical axis represents the number of samples falling within the interval. By observing the histogram, you can intuitively understand the distribution of the modeling delay and determine whether there are abnormal long delay samples. In addition, you can also calculate the statistical parameters of the delay data, such as the average delay, median delay, standard deviation, etc., to more comprehensively evaluate the modeling efficiency. For example, the average delay can reflect the overall execution speed of the modeling task, and the standard deviation can measure the fluctuation of the delay. Through these refined processing steps and analysis methods, the modeling quality can be evaluated more accurately, providing strong support for optimizing the modeling process.

[0085] In some embodiments, capturing the receiving timestamp of the returned reconstruction result includes: recording the initial point cloud feedback timestamp t first ; and / or record the complete grid model receiving timestamp t last ;

[0086] The calculation modeling delay includes: calculating the initial response delay Δt first =t first -t send ; and / or calculate the complete modeling period Δt last =t last -t send ;

[0087] The generation of quality evaluation indicators includes: constructing an initial response delay histogram and a complete modeling cycle probability density distribution diagram.

[0088] It should be noted that when capturing the reception timestamp of the reconstruction result returned by the digital twin modeling engine, the present invention not only records the reception timestamp of the complete grid model, but also records the initial point cloud feedback timestamp. In this way, the initial response delay and the complete modeling cycle can be calculated separately, so as to more comprehensively evaluate the performance of the modeling process. The initial response delay refers to the time interval from sending the modeling instruction to receiving the initial point cloud feedback, which reflects the response speed of the modeling engine to the task; the complete modeling cycle is the time interval from sending the modeling instruction to receiving the complete grid model, which represents the total time consumption of the entire modeling process. Based on these data, the initial response delay histogram and the complete modeling cycle probability density distribution diagram are generated, which can intuitively demonstrate the efficiency and stability of the modeling process and provide data support for optimizing modeling parameters and resource allocation.

[0089] Specifically, the initial point cloud feedback timestamp refers to the time during the modeling process when the digital twin modeling engine first returns point cloud data after receiving the modeling command. This typically indicates that the modeling engine has begun processing the task and feedback of preliminary results. For example, if the modeling engine returns preliminary point cloud data within 1 second of receiving the command, the initial point cloud feedback timestamp can be recorded as 2025-04-12 10:00:01.000. The complete mesh model reception timestamp refers to the time when the modeling engine completes the entire modeling task and returns the complete mesh model. For example, if the complete mesh model is returned 5 seconds after sending the command, the complete mesh model reception timestamp is 2025-04-12 10:00:05.000. By calculating the initial response latency (i.e., the difference between the complete mesh model reception timestamp and the initial point cloud feedback timestamp) and the complete modeling cycle (i.e., the difference between the complete mesh model reception timestamp and the modeling command sending timestamp), two important performance metrics can be obtained. For example, if the modeling instruction is sent at 2025-04-12 10:00:00.000, the initial response latency is 1 second and the complete modeling cycle is 5 seconds. These metrics can help evaluate the response speed and processing capabilities of the modeling engine.

[0090] Preferably, when generating the initial response delay histogram and the complete modeling cycle probability density distribution plot, the data processing and visualization steps can be further refined. For example, the initial response delay data and complete modeling cycle data for all modeling samples can be grouped and statistically analyzed. For initial response delay, the samples can be grouped into intervals such as 0-1 seconds, 1-2 seconds, and 2-3 seconds. The number of samples in each interval can be counted to generate an initial response delay histogram. Similarly, for complete modeling cycles, the samples can be grouped into intervals such as 0-5 seconds, 5-10 seconds, and 10-15 seconds to generate a complete modeling cycle probability density distribution plot. These histograms and distribution plots allow for intuitive observation of the response speed and completion time distribution of the modeling task. For example, if the initial response delay of most samples is concentrated in the 0-1 second interval, it indicates that the modeling engine has a fast response speed; if the probability density distribution plot of the complete modeling cycle shows that most samples are concentrated in the 5-10 second interval, it indicates that the completion time of the modeling task is relatively stable. In addition, statistical parameters such as the mean, median, and standard deviation can be calculated for this data to more comprehensively evaluate modeling performance. Through these refined processing steps and visualization methods, the efficiency and stability of the modeling process can be analyzed more accurately, providing strong support for optimizing modeling parameters and resource allocation.

[0091] In some embodiments, further comprising:

[0092] Get the number of characteristic topologies N of the benchmark model std , the benchmark model is an ideal tower model generated under no-load condition;

[0093] Extract the number of feature topologies N output by real-time modeling actual ;

[0094] Calculating model fidelity

[0095] The fidelity data of all samples is counted to generate the model accuracy evaluation index.

[0096] It should be noted that the present invention not only focuses on modeling efficiency during the modeling process, but also pays attention to the accuracy and fidelity of the model. By obtaining the number of characteristic topologies of the baseline model and comparing it with the number of characteristic topologies output by real-time modeling, the model fidelity is calculated, and then a model accuracy evaluation index is generated. The baseline model here refers to an ideal tower model generated under no-load conditions, which reflects the geometric structure and topological relationship of the tower under ideal conditions. The number of characteristic topologies refers to the number of key structural features in the model, such as the connection points of the tower, the intersection points of the components, etc. These features can reflect the complexity and integrity of the tower structure. By calculating the model fidelity, the degree of similarity between the real-time modeling output and the baseline model can be quantitatively evaluated, thereby ensuring the accuracy and reliability of the digital twin model.

[0097] Specifically, the characteristic topology number of the benchmark model refers to the number of key structural features in the ideal tower model under no-load conditions. For example, a typical tower model may contain multiple connection points, intersections between crossarms and the tower body, and so on. The number of these characteristic points can be determined through geometric analysis of the benchmark model. Assume that a benchmark model has 100 characteristic topology points, representing key connections and intersections in the tower structure. The characteristic topology number of the real-time modeling output refers to the number of key structural features of the same type identified in the model generated from the point cloud data during the actual modeling process. For example, if 95 characteristic topology points are identified in the real-time modeling output, the model fidelity can be calculated by calculating the ratio between these two numbers. Model fidelity can be calculated as the ratio of the characteristic topology number of the real-time modeling output to the characteristic topology number of the benchmark model. For example, 95 / 100 = 0.95, indicating a model fidelity of 95%. This method can quantitatively assess modeling accuracy and quality.

[0098] Preferably, when generating model accuracy evaluation metrics, the extraction of characteristic topological quantities and the calculation of fidelity can be further refined. For example, when extracting characteristic topological quantities, specific algorithms, such as geometric feature-based recognition algorithms or machine learning algorithms, can be employed to accurately identify key structural features in the model. For complex tower structures, different feature extraction thresholds can be set to ensure accurate identification of feature points. When calculating model fidelity, in addition to simple proportional calculations, weighting factors can be introduced to assign different weights based on the importance of characteristic topological points. For example, key connection points within the tower body can be given higher weights, while less important intersections can be given lower weights. This approach can more accurately reflect the fidelity of the model. Furthermore, the fidelity data of all samples can be statistically analyzed to generate model accuracy evaluation metrics, such as the mean and standard deviation of fidelity, to comprehensively assess the quality of the modeling process. Through these refined processing steps and analysis methods, the accuracy and reliability of the model can be more accurately assessed, providing strong support for optimizing the modeling process.

[0099] In some embodiments, further comprising:

[0100] Obtain the baseline feature number N under QoS guarantee conditions QoS ;

[0101] The number of features N generated by statistical real-time modeling within the service level agreement (SLA) time limit SLA ;

[0102] Calculating the service degradation index

[0103] Generate a service reliability evaluation matrix.

[0104] It should be noted that the present invention further introduces the concepts of service degradation index and service reliability evaluation matrix when evaluating modeling quality. By obtaining the baseline feature number under QoS (quality of service) guarantee conditions and counting the feature number generated by real-time modeling within the service level agreement (SLA) time limit, the service degradation index is calculated to generate a service reliability evaluation matrix. The baseline feature number under QoS guarantee conditions here refers to the minimum number of features that the model should have while meeting the established service quality requirements. The service level agreement (SLA) time limit refers to the time limit that the modeling service promises to complete the task. Through these indicators, the performance of the modeling service within the specified time can be evaluated to ensure the stability and reliability of the modeling system.

[0105] Specifically, the baseline feature count under QoS guarantees refers to the number of key features that a model must possess to meet service quality requirements. For example, assuming that a tower model must contain at least 80 key feature points to meet QoS requirements, these 80 feature points constitute the baseline feature count. The feature count generated within the service level agreement (SLA) timeframe for real-time modeling refers to the number of feature points actually generated by the modeling system within the SLA timeframe. For example, if the SLA stipulates that a modeling task must be completed within 10 minutes, and the system generates 75 feature points within 10 minutes, the actual number of generated features is 75. The service degradation index can be calculated by comparing the actual number of generated features with the baseline number. For example, if the actual number of generated features is lower than a certain percentage of the baseline number (e.g., lower than 90%), service is considered degraded. The service degradation index can be expressed as the ratio of the actual number of generated features to the baseline number. The service reliability evaluation matrix aggregates the service degradation indices of multiple modeling tasks into a single matrix used to assess the reliability of the entire modeling system. For example, each element in the matrix can represent the service degradation index of a modeling task. By analyzing the data distribution in the matrix, the performance of the modeling system under different conditions can be evaluated.

[0106] Preferably, when generating a service reliability evaluation matrix, the calculation of the service degradation index and the matrix construction process can be further refined. For example, when calculating the service degradation index, the time factor can be introduced, that is, different weights can be given according to the early or late completion time of the modeling task. If a task is completed within half of the time specified in the SLA, a higher weight can be given even if the number of features is slightly lower than the benchmark number. When constructing the service reliability evaluation matrix, it can be classified according to different modeling task types, priorities or time sequences to more clearly analyze the performance of the modeling system. For example, the matrix can be divided into two parts: high-priority tasks and low-priority tasks, and the service degradation index can be calculated separately. In addition, historical data can be combined to perform trend analysis on the service reliability evaluation matrix to predict the future performance of the modeling system. Through these refined processing steps and analysis methods, the stability and reliability of the modeling system can be more comprehensively evaluated, providing strong support for optimizing modeling services.

[0107] The above-described embodiments of the present invention have the following beneficial effects: The present invention can improve the accuracy and efficiency of digital twin modeling of distribution network towers. By classifying and statistically analyzing the geometric characteristics of towers and constructing a parameterized sample set, the modeling data can be ensured to be consistent with the actual component distribution ratio, thereby improving model accuracy. The use of a distributed parallel computing architecture can dynamically allocate resources based on the operating area and modeling scale, effectively improving the processing speed of large-scale tower group modeling. The standardized data processing process based on IEC61970 and OPCUA protocols can enhance system compatibility, and the abnormal point cloud filtering function can ensure the quality of input data.

[0108] In terms of modeling quality assessment, this invention can establish a multi-dimensional quantitative evaluation system. By recording modeling instruction timestamps and calculating response delays, a delay distribution feature matrix can be generated, providing data support for system performance optimization. Introducing indicators such as model fidelity calculation and service degradation index can simultaneously evaluate modeling accuracy and service reliability, forming a comprehensive evaluation mechanism that includes timeliness, accuracy, and stability. This closed-loop feedback mechanism can provide a scientific basis for the continuous optimization of digital twin systems.

[0109] like Figure 2 As shown, some embodiments provide a distribution network tower digital twin modeling system, the system comprising:

[0110] The statistical processing unit 201 is used to perform classified statistical analysis on the geometric structural characteristic data of the target tower to determine the parameter distribution ratio of various tower components;

[0111] A modeling data set construction unit 202 is configured to select modeling samples from the tower point cloud database to construct a three-dimensional modeling sample set, wherein the parameter distribution of various tower components in the sample set is consistent with the parameter distribution ratio;

[0112] The modeling processing unit 203 is used to generate model construction instructions based on the three-dimensional modeling sample set, drive the digital twin modeling engine to reconstruct the tower through the instructions, and collect modeling process performance data to generate modeling quality evaluation indicators.

[0113] It is understandable that the modules recorded in the distribution network tower digital twin modeling system are the same as those in the reference Figure 1 The steps in the distribution network tower digital twin modeling method described above correspond to each other. Therefore, the operations, features, and beneficial effects described above for the distribution network tower digital twin modeling method are also applicable to the distribution network tower digital twin modeling system and the modules contained therein, and will not be repeated here.

[0114] Reference below Figure 3, which shows a schematic diagram of the structure of a computing device 300 suitable for implementing some embodiments of the present invention. The computing device in some embodiments of the present invention may include, but is not limited to, mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), in-vehicle terminals (e.g., in-vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. Figure 3 The terminal device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0115] like Figure 3 As shown, computing device 300 may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 301, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 302 or a program loaded from a storage device 308 into a random access memory (RAM) 303. Various programs and data required for the operation of computing device 300 are also stored in RAM 303. Processing device 301, ROM 302, and RAM 303 are connected to each other via a bus 304. An input / output (I / O) interface 305 is also connected to bus 304.

[0116] Typically, the following devices may be connected to the I / O interface 305: input devices 306 including, for example, a touch screen, a touchpad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; output devices 307 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; storage devices 308 including, for example, a magnetic tape, a hard disk, etc.; and communication devices 309. The communication devices 309 may allow the computing device 300 to communicate with other devices wirelessly or by wire to exchange data. Although Figure 3 The computing device 300 is shown with various devices, but it should be understood that it is not required to implement or have all of the devices shown. More or fewer devices may be implemented or have instead. Figure 3 Each block shown in the figure may represent one device, or may represent multiple devices as needed.

[0117] Furthermore, the storage medium of the embodiment of the present application stores program instructions that can implement all the above methods, wherein the program instructions can be stored in the above storage medium in the form of a software product, including a number of instructions for causing a computer device (which can be a personal computer, server, or network device, etc.) or a processor to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, or a terminal device such as a computer, a server, a mobile phone, or a tablet.

[0118] The above descriptions are merely some preferred embodiments of the present invention and an illustration of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in the embodiments of the present invention is not limited to the technical solutions formed by a specific combination of the above-mentioned technical features, but should also encompass other technical solutions formed by any combination of the above-mentioned technical features or their equivalents without departing from the above-mentioned inventive concept. For example, a technical solution formed by mutually replacing the above-mentioned features with (but not limited to) technical features having similar functions disclosed in the embodiments of the present invention.

Claims

1. A distribution network tower digital twin modeling method, characterized in that: include: Conduct statistical analysis on the geometric structural characteristic data of the target tower to determine the parameter distribution ratio of various tower components; Selecting modeling samples from the tower point cloud database to construct a three-dimensional modeling sample set; wherein the parameter distribution of various tower components in the sample set is consistent with the parameter distribution ratio; Based on the three-dimensional modeling sample set, a model construction instruction is generated, the digital twin modeling engine is driven by the instruction to reconstruct the tower, and the performance data of the modeling process is collected to generate a modeling quality evaluation index.

2. The modeling method according to claim 1, characterized in that Before performing a modeling operation based on the three-dimensional modeling sample set, the method further includes: Calculate the modeling operation area based on the topological distribution density of the target tower and the estimated modeling scale, and determine the number of parallel modeling threads based on the tower modeling priority and the operation area; configuring distributed modeling nodes according to the number of parallel modeling threads and the sample capacity of the three-dimensional modeling sample set; The generating of the model building instruction based on the three-dimensional modeling sample set includes: loading the three-dimensional modeling samples in parallel through the distributed modeling nodes; Each distributed modeling node synchronously generates model building instructions and submits modeling tasks to the digital twin modeling engine.

3. The modeling method according to claim 1, characterized in that Generating a model building instruction based on the three-dimensional modeling sample set includes: The sample data is formatted according to the IEC61970 standard to generate standardized point cloud data; Encapsulate standardized point cloud data into model building instructions that comply with the OPC UA protocol.

4. The modeling method according to claim 1, characterized in that Before generating the model building instruction, the method further includes: Based on the tower defect feature library, abnormal point cloud filtering is performed on the modeling samples.

5. The modeling method according to any one of claims 1 to 4, characterized in that: The collecting of modeling process performance data to generate modeling quality evaluation indicators includes: Record the sending timestamp t of the model building instruction send ; Capture the receiving timestamp t of the reconstruction result returned by the digital twin modeling engine recv ; Calculate the single modeling delay Δt=t recv -t send ; The modeling delay data of all samples are counted to generate a delay distribution feature matrix.

6. The modeling method according to claim 5, characterized in that: The receiving timestamp of the captured and returned reconstruction result includes: recording the initial point cloud feedback timestamp t first ; and / or record the complete grid model receiving timestamp t last ; The calculation modeling delay includes: calculating the initial response delay Δt first =t first -t send ; and / or calculate the complete modeling period Δt last =t last -t send ; The generation of quality evaluation indicators includes: constructing an initial response delay histogram and a complete modeling cycle probability density distribution diagram.

7. The modeling method according to any one of claims 1 to 4, characterized in that: Also includes: Get the number of characteristic topologies N of the benchmark model std , the benchmark model is an ideal tower model generated under no-load condition; Extract the number of feature topologies N output by real-time modeling actual ; Calculating model fidelity The fidelity data of all samples is counted to generate the model accuracy evaluation index.

8. The modeling method according to claim 7, characterized in that: Also includes: Obtain the baseline feature number N under QoS guarantee conditions QoS ; The number of features N generated by statistical real-time modeling within the service level agreement (SLA) time limit SLA ; Calculating the service degradation index Generate a service reliability evaluation matrix.

9. A digital twin modeling system for distribution network towers, characterized in that: include: Statistical processing unit, used to perform classified statistical analysis on the geometric structural characteristic data of the target tower and determine the parameter distribution ratio of various tower components; A modeling data set construction unit is used to select modeling samples from the tower point cloud database to construct a three-dimensional modeling sample set, wherein the parameter distribution of various tower components in the sample set is consistent with the parameter distribution ratio; A modeling processing unit is used to generate model construction instructions based on the three-dimensional modeling sample set, drive the digital twin modeling engine to reconstruct the tower through the instructions, and collect modeling process performance data to generate modeling quality evaluation indicators.

10. A computing device, characterized in that It includes a processor and a memory, the memory is used to store a computer program; when the computer program is loaded by the processor, the processor executes the distribution network tower digital twin modeling method according to any one of claims 1 to 8.

Images