A power grid planning project monitoring method and device based on digital twinning and a storage medium

Abstract

The present application provides a power grid planning project monitoring method and device based on digital twinning, and a storage medium. The method comprises: collecting multi-source heterogeneous data of the whole process of the power grid planning project implementation process, wherein the multi-source heterogeneous data at least includes power grid operation data, engineering construction data, environmental meteorological data and equipment state data; performing fusion processing on the heterogeneous multi-source data to obtain fusion feature data; performing geometric and spatial twinning modeling, electrical and operation twinning modeling and construction process time sequence twinning modeling based on the fusion feature data to obtain a power grid planning digital project; and performing simulation deduction based on the power grid planning digital project to determine whether there is a potential risk in the power grid planning project, and issuing an alarm information if there is a potential risk. The present application improves the control efficiency of the power grid planning project implementation and ensures the accurate landing of the power grid planning scheme.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence, digital twins and the integration of power grid planning and construction technologies, specifically to a monitoring method, device and storage medium for power grid planning projects based on digital twins. Background Technology

[0002] Current technologies focus on quality monitoring and equipment status monitoring during the construction phase of power grid infrastructure, or load forecasting and scheme optimization during the planning phase, but research on monitoring technologies that cover the entire process of power grid planning projects is relatively scarce. Furthermore, the need for data collection nodes to share raw data leads to privacy leaks.

[0003] The entire process of power grid planning projects involves multiple professional departments, including planning, feasibility studies, design, infrastructure, dispatching, operation and maintenance, marketing, and materials. The data has typical characteristics of being multi-source, heterogeneous, localized, highly private, and not centrally aggregated.

[0004] Although digital twin technology has a wide range of applications, its implementation in power grid planning projects and the prediction of corresponding risks still require technical adjustments. Summary of the Invention

[0005] In view of one or more technical defects in the prior art, the present invention proposes the following technical solution.

[0006] A monitoring method for power grid planning projects based on digital twins, the method comprising: The data collection process involves collecting multi-source heterogeneous data from the entire process of implementing the power grid planning project. This multi-source heterogeneous data includes at least power grid operation data, engineering construction data, environmental meteorological data, and equipment status data. Fusion step: The heterogeneous multi-source data is fused to obtain fused feature data; The generation step involves performing geometric and spatial twin modeling, electrical and operational twin modeling, and construction process temporal twin modeling based on the fused feature data to obtain a digital project for power grid planning. The monitoring process involves performing simulations based on the power grid planning digital project to determine whether there are any potential risks in the power grid planning project. If potential risks are found, an alarm message is issued.

[0007] Furthermore, in the data acquisition step, k computing nodes are used to acquire data. The acquired data includes structured time-series data, spatiotemporal geographic raster data, and unstructured data. The structured time-series data includes load, current, voltage, project progress, and weather. The spatiotemporal geographic raster data includes GIS data, tower locations, corridors, and terrain. The unstructured data includes project drawings, project images taken by UAVs, project feasibility study reports, and project change orders.

[0008] Furthermore, in the fusion step, a federated learning model is composed of K local computing nodes. Each local computing node only processes local data, while the central node performs model aggregation and outputs fused feature data.

[0009] Furthermore, the loss function for the k-th local computing node is: ; in, Represents the local dataset. express The number of data in the middle, h() Representing a classification model, l() This represents a regression or classification model. This represents the regularity coefficient of L2, typically taken as 0.001. This represents the learnable parameters of the encoder of the federated learning model on the k-th local computing node. for Data in the middle; Global Model of Federated Learning It is obtained by weighting the K local computing nodes: ; in, , ; in, As weight, For the kth The amount of effective data collected For the i-th The amount of effective data collected For the kth The weighting adjustment parameters are used to increase the weight of important departments; The loss function used for the feature alignment calculation is: ; in, This represents the data distribution of computing node k. Represents a computing node l Data distribution Indicates from Mid-sampling feature z, Indicates from Mid-sampling feature z, This represents mapping the feature z to a higher-dimensional space, E[] represents calculating the mathematical expectation, and H represents the reproducing kernel Hilbert space. This indicates the calculation of the squared distance in the regenerating kernel Hilbert space; The total loss function for federated learning is: , This represents the alignment loss weight, typically a value between 0.05 and 0.2.

[0010] Furthermore, in the generation step, let the set of physical entity features be... ,in, For geometric features, For electrical characteristics, For spatial features, Given the time-series characteristics, the digital project model for power grid planning of the virtual twin is as follows: The twin mapping relationship is as follows:

[0011] in, The twin mapping function is implemented using a CNN-LSTM hybrid neural network to achieve multi-feature fusion mapping; This represents the set of learnable parameters for the twin model, including geometric modeling parameters, electrical simulation parameters, and spatiotemporal correlation parameters. This represents the model fitting error, which is controlled through optimization algorithms. To ensure the accuracy of twins.

[0012] Furthermore, the geometric and spatial twin modeling employs BIM+GIS integrated modeling technology to construct a three-dimensional geometric model of the project. The geometric feature mapping formula is as follows: ; in, Indicates the first Three-dimensional geometric coordinate parameters of a device / structure Indicates the first The geometric weight coefficient of each piece of equipment / structure is determined by its importance. This represents a geospatial coordinate mapping function that integrates latitude, longitude, elevation, and topographic data. Based on power flow calculation and short-circuit calculation theories, an electrical and operational twin model is performed to simulate the power grid operation state before and after the commissioning of the planned project. The electrical characteristic mapping formula is as follows: ; in, These represent the node voltage and branch current phasors, respectively. These represent the branch impedance and admittance matrices, respectively. Active and reactive power respectively It represents the power factor and matches real-time data on actual grid operation. The formula for time-series twin modeling of the construction process is: ; in, express and The construction twin status is constantly monitored. This represents the incremental progress of construction per unit of time. This represents the construction progress rate function.

[0013] Furthermore, the potential risks include construction delays, power grid flow exceeding limits, and extreme weather risks.

[0014] Furthermore, the prediction method for construction schedule delay risk is as follows: input the delay data of construction manpower, materials, and weather into the power grid planning digital project model, the power grid planning digital project model simulates the impact of schedule delay on the total project duration, subsequent acceptance, and power grid commissioning, and outputs the scope of delay and remedial measures suggestions; The prediction method for power grid power flow over-limit risk is as follows: use the power grid planning digital project model to simulate the new energy fluctuation and load surge conditions after the planned project is put into operation, calculate the power flow distribution in real time, warn of line overload and voltage over-limit risks, and output the grid optimization scheme; The method for predicting extreme weather risks is as follows: meteorological data such as heavy rain, strong winds, and high temperatures are input into the power grid planning digital project model. The power grid planning digital project model simulates construction safety hazards and equipment operation failures, delineates risk areas in advance, and formulates prevention and control plans.

[0015] Furthermore, the deviation values ​​between the power grid planning project and the actual construction project are calculated, and the contribution of the deviation is calculated to determine the causes of the deviation between the power grid planning project and the actual construction project, so as to carry out construction rectification. The deviation value is calculated as follows: ; in, Indicates the deviation value. This indicates geometric layout deviation, calculated based on the differences in location and dimensions between the planned project and the actual construction. This indicates the deviation in electrical performance, calculated based on the difference between the design parameters of the planned project and the actual operating electrical parameters. This indicates the schedule deviation, calculated based on the difference between the planned project duration and the actual construction progress.

[0016] use , , Divide by respectively The deviation contribution rates of geometric layout, electrical performance, and schedule are obtained. The largest deviation contribution rate among the three is selected to determine the main reasons for the deviation between the power grid planning project and the actual construction project, and construction rectification is carried out.

[0017] This invention also proposes a power grid planning project monitoring device based on digital twins, the device comprising: The data acquisition unit collects multi-source heterogeneous data throughout the entire process of implementing the power grid planning project. The multi-source heterogeneous data includes at least power grid operation data, engineering construction data, environmental meteorological data, and equipment status data. Fusion unit: performs fusion processing on the heterogeneous multi-source data to obtain fused feature data; The generation unit performs geometric and spatial twin modeling, electrical and operational twin modeling, and construction process temporal twin modeling based on the fused feature data to obtain a digital project for power grid planning; The monitoring unit performs simulations based on the power grid planning digital project to determine whether there are potential risks in the power grid planning project. If there are potential risks, it issues an alarm message.

[0018] Furthermore, in the acquisition unit, k computing nodes are used to acquire data. The acquired data includes structured time-series data, spatiotemporal geographic raster data, and unstructured data. The structured time-series data includes load, current, voltage, project progress, and weather. The spatiotemporal geographic raster data includes GIS data, tower locations, corridors, and terrain. The unstructured data includes project drawings, project images taken by UAVs, project feasibility study reports, and project change orders.

[0019] Furthermore, in the fusion unit, a federated learning model is composed of K local computing nodes. Each local computing node only processes local data, while the central node performs model aggregation and outputs fused feature data.

[0020] Furthermore, the loss function for the k-th local computing node is: ; in, Represents the local dataset. express The number of data in the middle, h() Representing a classification model, l() This represents a regression or classification model. This represents the regularity coefficient of L2, typically taken as 0.001. This represents the learnable parameters of the encoder of the federated learning model on the k-th local computing node. for Data in the middle; Global Model of Federated Learning It is obtained by weighting the K local computing nodes: ; in, , ; in, As weight, For the kth The amount of effective data collected For the i-th The amount of effective data collected For the kth The weighting adjustment parameters are used to increase the weight of important departments; The loss function used for the feature alignment calculation is: ; in, This represents the data distribution of computing node k. Represents a computing node l Data distribution Indicates from Mid-sampling feature z, Indicates from Mid-sampling feature z, This represents mapping the feature z to a higher-dimensional space, E[] represents calculating the mathematical expectation, and H represents the reproducing kernel Hilbert space. This indicates the calculation of the squared distance in the regenerating kernel Hilbert space; The total loss function for federated learning is: , This represents the alignment loss weight, typically a value between 0.05 and 0.2.

[0021] Furthermore, in the generation unit, let the set of physical entity features be... ,in, For geometric features, For electrical characteristics, For spatial features, Given the time-series characteristics, the digital project model for power grid planning of the virtual twin is as follows: The twin mapping relationship is as follows:

[0022] in, The twin mapping function is implemented using a CNN-LSTM hybrid neural network to achieve multi-feature fusion mapping; This represents the set of learnable parameters for the twin model, including geometric modeling parameters, electrical simulation parameters, and spatiotemporal correlation parameters. This represents the model fitting error, which is controlled through optimization algorithms. To ensure the accuracy of twins.

[0023] Furthermore, the geometric and spatial twin modeling employs BIM+GIS integrated modeling technology to construct a three-dimensional geometric model of the project. The geometric feature mapping formula is as follows: ; in, Indicates the first Three-dimensional geometric coordinate parameters of a device / structure Indicates the first The geometric weight coefficient of each piece of equipment / structure is determined by its importance. This represents a geospatial coordinate mapping function that integrates latitude, longitude, elevation, and topographic data. Based on power flow calculation and short-circuit calculation theories, an electrical and operational twin model is performed to simulate the power grid operation state before and after the commissioning of the planned project. The electrical characteristic mapping formula is as follows: ; in, These represent the node voltage and branch current phasors, respectively. These represent the branch impedance and admittance matrices, respectively. Active and reactive power respectively It represents the power factor and matches real-time data on actual grid operation. The formula for time-series twin modeling of the construction process is: ; in, express and The construction twin status is constantly monitored. This represents the incremental progress of construction per unit of time. This represents the construction progress rate function.

[0024] Furthermore, the potential risks include construction delays, power grid flow exceeding limits, and extreme weather risks.

[0025] Furthermore, the prediction method for construction schedule delay risk is as follows: input the delay data of construction manpower, materials, and weather into the power grid planning digital project model, the power grid planning digital project model simulates the impact of schedule delay on the total project duration, subsequent acceptance, and power grid commissioning, and outputs the scope of delay and remedial measures suggestions; The prediction method for power grid power flow over-limit risk is as follows: use the power grid planning digital project model to simulate the new energy fluctuation and load surge conditions after the planned project is put into operation, calculate the power flow distribution in real time, warn of line overload and voltage over-limit risks, and output the grid optimization scheme; The method for predicting extreme weather risks is as follows: meteorological data such as heavy rain, strong winds, and high temperatures are input into the power grid planning digital project model. The power grid planning digital project model simulates construction safety hazards and equipment operation failures, delineates risk areas in advance, and formulates prevention and control plans.

[0026] Furthermore, the deviation values ​​between the power grid planning project and the actual construction project are calculated, and the contribution of the deviation is calculated to determine the causes of the deviation between the power grid planning project and the actual construction project, so as to carry out construction rectification. The deviation value is calculated as follows: ; in, Indicates the deviation value. This indicates geometric layout deviation, calculated based on the differences in location and dimensions between the planned project and the actual construction. This indicates the deviation in electrical performance, calculated based on the difference between the design parameters of the planned project and the actual operating electrical parameters. This indicates the schedule deviation, calculated based on the difference between the planned project duration and the actual construction progress.

[0027] use , , Divide by respectively The deviation contribution rates of geometric layout, electrical performance, and schedule are obtained. The largest deviation contribution rate among the three is selected to determine the main reasons for the deviation between the power grid planning project and the actual construction project, and construction rectification is carried out.

[0028] Furthermore, the present invention also proposes a computer-readable storage medium storing computer program code, which, when executed by a computer, performs any of the methods described above.

[0029] The technical effect of this invention is as follows: This invention provides a method, device, and storage medium for monitoring power grid planning projects based on digital twins. The method includes: a data acquisition step S101, acquiring multi-source heterogeneous data throughout the entire implementation process of a power grid planning project, wherein the multi-source heterogeneous data includes at least power grid operation data, engineering construction data, environmental meteorological data, and equipment status data; a fusion step S102, fusing the heterogeneous multi-source data to obtain fused feature data; a generation step S103, performing geometric and spatial twin modeling, electrical and operational twin modeling, and construction process temporal twin modeling based on the fused feature data to obtain a digital power grid planning project; and a monitoring step S104, performing simulation based on the digital power grid planning project to determine whether there are potential risks in the power grid planning project, and issuing alarm information if potential risks exist. This invention addresses the shortcomings of existing technologies that focus on single aspects of power grid infrastructure construction, such as quality monitoring and equipment status monitoring, or load forecasting and scheme optimization during the planning phase. There is a lack of research on end-to-end monitoring technologies for power grid planning projects, and the need for data acquisition nodes to share raw data leads to privacy leaks. This invention proposes a federated learning approach to fuse heterogeneous multi-source data into fused feature data. Then, digital twin technology is used to perform geometric and spatial twin modeling, electrical and operational twin modeling, and construction process temporal twin modeling based on this fused feature data to obtain a digital power grid planning project. This digital project can then be used for simulation to determine potential risks. If potential risks are identified, alarm information is issued. This enables real-time collection and dynamic display of key indicators at each stage of the power grid planning project, ensuring full traceability of the project status, improving the management efficiency of power grid planning project implementation, and guaranteeing the accurate implementation of power grid planning schemes. Attached Figure Description

[0030] Other features, objects, and advantages of this application will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings.

[0031] Figure 1 This is a flowchart of a power grid planning project monitoring method based on digital twin according to an embodiment of the present invention.

[0032] Figure 2 This is a structural diagram of a power grid planning project monitoring device based on a digital twin according to an embodiment of the present invention. Detailed Implementation

[0033] The present application will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and not intended to limit it. Furthermore, it should be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings.

[0034] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0035] Figure 1 This invention illustrates a power grid planning project monitoring method based on digital twins, the method comprising: In step S101, multi-source heterogeneous data of the entire process of power grid planning project implementation are collected. The multi-source heterogeneous data includes at least power grid operation data, engineering construction data, environmental meteorological data and equipment status data. Fusion step S102: The heterogeneous multi-source data is fused to obtain fused feature data; In step S103, geometric and spatial twin modeling, electrical and operational twin modeling, and construction process temporal twin modeling are performed based on the fused feature data to obtain a digital project for power grid planning. In monitoring step S104, a simulation is performed based on the power grid planning digital project to determine whether there are potential risks in the power grid planning project. If there are potential risks, an alarm message is issued.

[0036] This invention addresses the shortcomings of existing technologies, which often focus on single aspects like quality monitoring and equipment status monitoring during the construction phase of power grid infrastructure, or load forecasting and scheme optimization during the planning phase. There is a lack of research on integrated monitoring technologies for the entire power grid planning project process, and the need for data acquisition nodes to share raw data leads to privacy leaks. This invention proposes a federated learning approach to fuse heterogeneous multi-source data into fused feature data. Then, digital twin technology is used to perform geometric and spatial twin modeling, electrical and operational twin modeling, and construction process temporal twin modeling based on this fused feature data to obtain a digital power grid planning project. This digital project can then be used for simulation to determine potential risks. If potential risks are identified, alarm messages are issued. This enables real-time collection and dynamic display of key indicators at each stage of the power grid planning project, ensuring full traceability of the project status, improving the management efficiency of power grid planning project implementation, and guaranteeing the accurate implementation of power grid planning schemes. This is one of the key inventive concepts of this invention.

[0037] In one embodiment, multi-source heterogeneous data can be collected from various types of data using deployed IoT sensors, drones, high-definition cameras, power grid data acquisition terminals, engineering management terminals, and other devices. Specifically, in the acquisition step S101, k computing nodes are used to collect data, including structured time-series data, spatiotemporal geographic raster data, and unstructured data. The structured time-series data includes load, current, voltage, project progress, and weather; the spatiotemporal geographic raster data includes GIS data, tower locations, corridors, and terrain; and the unstructured data includes project drawings, drone-captured project images, project feasibility study reports, and project change orders. This lays the foundation for the unified access, governance, and sharing of multi-source heterogeneous data such as power grid topology data, geographic environment data, engineering construction data, equipment operation data, meteorological environment data, and market transaction data.

[0038] In one embodiment, the entire process of a power grid planning project involves multiple professional departments such as planning, feasibility study, design, infrastructure, dispatching, operation and maintenance, marketing, and materials. The data is typically multi-source, heterogeneous, localized, highly privacy-sensitive, and cannot be centrally aggregated. Accurately fusing this multi-source heterogeneous data is a technical challenge. This invention creatively proposes that in the fusion step S102, a federated learning model composed of K local computing nodes is used. Each local computing node processes only local data, while the central node performs model aggregation and outputs fused feature data. This ensures that the entire process of monitoring data fusion for the power grid planning project is achieved without data leaving its domain or privacy being compromised. This is another important inventive concept of this invention.

[0039] In one embodiment, the loss function of the k-th local computing node is: ; in, Represents the local dataset. express The number of data in the middle, h() Representing a classification model, l() This represents a regression or classification model. This represents the regularity coefficient of L2, typically taken as 0.001. This represents the learnable parameters of the encoder of the federated learning model on the k-th local computing node. for Data in the middle; This invention provides a creative global model for federated learning. The calculation method has been improved, specifically... The calculation method, in the calculation The decision considered increasing the weight of the scheduling and construction departments, i.e., setting... , which is the kth The weight adjustment parameters are therefore obtained by weighting the model from K local computing nodes: ; in, , ; in, As weight, For the kth The amount of effective data collected For the i-th The amount of effective data collected For the kth The weight adjustment parameters are adjusted to increase the weight of important departments, forming a federated feature fusion algorithm for monitoring power grid planning projects. This makes the federated learning method consistent with the actual situation of power grid planning projects, which is another important inventive concept of this invention.

[0040] In this invention, in order to ensure the effect of data fusion, feature alignment calculation across computing nodes is added on the basis of traditional FedAvg to solve the problem of distribution offset of multi-source heterogeneous data in the power grid.

[0041] The loss function used for the feature alignment calculation is: ; in, This represents the data distribution of computing node k. Represents a computing node l Data distribution Indicates from Mid-sampling feature z, Indicates from Mid-sampling feature z, This represents mapping the feature z to a higher-dimensional space, E[] represents calculating the mathematical expectation, and H represents the reproducing kernel Hilbert space. This indicates that the squared distance is calculated in the regenerating kernel Hilbert space; the total loss function of federated learning is: , This represents the alignment loss weight, typically a value between 0.05 and 0.2.

[0042] This invention adds feature alignment calculation, thereby ensuring that the features in the fused feature data obtained from the processed heterogeneous multi-source data are aligned, thus improving the accuracy of the model generated by the subsequent digital twin, and further improving the accuracy of monitoring. This is another important inventive concept of this invention.

[0043] In one embodiment, in the generation step S103, let the set of physical entity features be... ,in, For geometric features, For electrical characteristics, For spatial features, Given the time-series characteristics, the digital project model for power grid planning of the virtual twin is as follows: The twin mapping relationship is as follows:

[0044] in, The twin mapping function is implemented using a CNN-LSTM hybrid neural network to achieve multi-feature fusion mapping; This represents the set of learnable parameters for the twin model, including geometric modeling parameters, electrical simulation parameters, and spatiotemporal correlation parameters. This represents the model fitting error, which is controlled through optimization algorithms. To ensure the accuracy of twins.

[0045] In this invention, the fused feature data obtained through federated learning is used to construct a data-driven model for digital twins, achieving a precise mapping from physical entities to virtual models. Specifically, in the digital twin process, CNN is used to extract spatiotemporal geographic raster data and unstructured data from the fused feature data, while LSTM is used to extract data with temporal features from the fused feature data. This ensures the accuracy of data extraction and guarantees that the data is aligned, which is another important inventive concept of this invention.

[0046] In one embodiment, the geometric and spatial twin modeling employs BIM+GIS fusion modeling technology to construct a three-dimensional geometric model of the project, and the geometric feature mapping formula is: ; in, Indicates the first Three-dimensional geometric coordinate parameters of a device / structure Indicates the first The geometric weight coefficient of each piece of equipment / structure is determined by its importance. This represents a geospatial coordinate mapping function that integrates latitude, longitude, elevation, and topographic data. Based on power flow calculation and short-circuit calculation theories, an electrical and operational twin model is performed to simulate the power grid operation state before and after the commissioning of the planned project. The electrical characteristic mapping formula is as follows: ; in, These represent the node voltage and branch current phasors, respectively. These represent the branch impedance and admittance matrices, respectively. Active and reactive power respectively It represents the power factor and matches real-time data on actual grid operation. The formula for time-series twin modeling of the construction process is: ; in, express and The construction twin status is constantly monitored. This represents the incremental progress of construction per unit of time. This represents the construction progress rate function. It is influenced by factors such as manpower, materials, environment, and weather, and is driven by real-time data from the on-site Internet of Things (IoT).

[0047] This invention proposes specific formulas for geometric and spatial twin modeling, electrical and operational twin modeling, and construction process temporal twin modeling, thereby enabling the use of the established twin models for corresponding risk calculations. This is another important inventive concept of this invention.

[0048] In one embodiment, the potential risks include construction delay risk, power grid flow overload risk, and extreme weather risk.

[0049] In one embodiment, the prediction method for construction schedule delay risk is as follows: input the delay data of construction manpower, materials, and weather into the power grid planning digital project model, the power grid planning digital project model simulates the impact of schedule delay on the total project duration, subsequent acceptance, and power grid commissioning, and outputs the scope of delay and remedial measures suggestions; The prediction method for power grid power flow over-limit risk is as follows: use the power grid planning digital project model to simulate the new energy fluctuation and load surge conditions after the planned project is put into operation, calculate the power flow distribution in real time, warn of line overload and voltage over-limit risks, and output the grid optimization scheme; The method for predicting extreme weather risks is as follows: meteorological data such as heavy rain, strong winds, and high temperatures are input into the power grid planning digital project model. The power grid planning digital project model simulates construction safety hazards and equipment operation failures, delineates risk areas in advance, and formulates prevention and control plans.

[0050] In this invention, geometric and spatial twin modeling, electrical and operational twin modeling, and construction process temporal twin modeling are performed in the digital project model of power grid planning. By inputting relevant parameters, the corresponding risks can be calculated, and corresponding risk prevention and control schemes can be generated for user reference. This is another important inventive concept of this invention.

[0051] In one embodiment, the deviation value between the power grid planning project and the actual construction project is calculated, and the deviation contribution is calculated to determine the cause of the deviation between the power grid planning project and the actual construction project, so as to carry out construction rectification. The deviation value is calculated as follows: ; in, Indicates the deviation value. This indicates geometric layout deviation, calculated based on the differences in location and dimensions between the planned project and the actual construction. This indicates the deviation in electrical performance, calculated based on the difference between the design parameters of the planned project and the actual operating electrical parameters. This indicates the schedule deviation, calculated based on the difference between the planned project duration and the actual construction progress.

[0052] use , , Divide by respectively The invention obtains the contribution rates of deviations in geometric layout, electrical performance, and schedule. The largest contribution rate among these three factors is selected to determine the main cause of the deviation between the power grid planning project and the actual construction project, enabling corrective action during construction. This invention's key inventive concept lies in its ability to pinpoint the source of deviations through deviation contribution analysis, thus facilitating project rectification.

[0053] Figure 2 This invention illustrates a power grid planning project monitoring device based on digital twins, the device comprising: The data acquisition unit 201 acquires multi-source heterogeneous data throughout the entire process of implementing the power grid planning project. The multi-source heterogeneous data includes at least power grid operation data, engineering construction data, environmental meteorological data, and equipment status data. Fusion unit 202: performs fusion processing on the heterogeneous multi-source data to obtain fused feature data; The generation unit 203 performs geometric and spatial twin modeling, electrical and operational twin modeling, and construction process temporal twin modeling based on the fused feature data to obtain a digital project for power grid planning; The monitoring unit 204 performs simulations based on the power grid planning digital project to determine whether there are potential risks in the power grid planning project. If there are potential risks, it issues an alarm message.

[0054] This invention addresses the shortcomings of existing technologies, which often focus on single aspects like quality monitoring and equipment status monitoring during the construction phase of power grid infrastructure, or load forecasting and scheme optimization during the planning phase. There is a lack of research on integrated monitoring technologies for the entire power grid planning project process, and the need for data acquisition nodes to share raw data leads to privacy leaks. This invention proposes a federated learning approach to fuse heterogeneous multi-source data into fused feature data. Then, digital twin technology is used to perform geometric and spatial twin modeling, electrical and operational twin modeling, and construction process temporal twin modeling based on this fused feature data to obtain a digital power grid planning project. This digital project can then be used for simulation to determine potential risks. If potential risks are identified, alarm messages are issued. This enables real-time collection and dynamic display of key indicators at each stage of the power grid planning project, ensuring full traceability of the project status, improving the management efficiency of power grid planning project implementation, and guaranteeing the accurate implementation of power grid planning schemes. This is one of the key inventive concepts of this invention.

[0055] In one embodiment, multi-source heterogeneous data can be collected from various types of data using deployed IoT sensors, drones, high-definition cameras, power grid data acquisition terminals, engineering management terminals, and other devices. Specifically, in the acquisition unit 201, k computing nodes are used to collect data, including structured time-series data, spatiotemporal geographic raster data, and unstructured data. The structured time-series data includes load, current, voltage, project progress, and weather; the spatiotemporal geographic raster data includes GIS data, tower locations, corridors, and terrain; and the unstructured data includes project drawings, drone-captured project images, project feasibility study reports, and project change orders. This lays the foundation for the unified access, governance, and sharing of multi-source heterogeneous data such as power grid topology data, geographic environment data, engineering construction data, equipment operation data, meteorological environment data, and market transaction data.

[0056] In one embodiment, the entire process of a power grid planning project involves multiple professional departments such as planning, feasibility study, design, infrastructure, dispatching, operation and maintenance, marketing, and materials. The data is typically multi-source, heterogeneous, localized, highly privacy-sensitive, and cannot be centrally aggregated. Accurately fusing this multi-source heterogeneous data is a technical challenge. This invention creatively proposes using a federated learning model composed of K local computing nodes in the fusion unit 202. Each local computing node processes only local data, while the central node performs model aggregation and outputs fused feature data. This ensures that the entire process of monitoring data fusion for the power grid planning project is achieved without data leaving its domain or privacy being compromised. This is another important inventive concept of this invention.

[0057] In one embodiment, the loss function of the k-th local computing node is: ; in, Represents the local dataset. express The number of data in the middle, h() Representing a classification model, l() This represents a regression or classification model. This represents the regularity coefficient of L2, typically taken as 0.001. This represents the learnable parameters of the encoder of the federated learning model on the k-th local computing node. for Data in the middle; This invention provides a creative global model for federated learning. The calculation method has been improved, specifically... The calculation method, in the calculation The decision considered increasing the weight of the scheduling and construction departments, i.e., setting... , which is the kth The weight adjustment parameters are therefore obtained by weighting the model from K local computing nodes: ; in, , ; in, As weight, For the kth The amount of effective data collected For the i-th The amount of effective data collected For the kth The weight adjustment parameters are adjusted to increase the weight of important departments, forming a federated feature fusion algorithm for monitoring power grid planning projects. This makes the federated learning method consistent with the actual situation of power grid planning projects, which is another important inventive concept of this invention.

[0058] In this invention, in order to ensure the effect of data fusion, feature alignment calculation across computing nodes is added on the basis of traditional FedAvg to solve the problem of distribution offset of multi-source heterogeneous data in the power grid.

[0059] The loss function used for the feature alignment calculation is: ; in, This represents the data distribution of computing node k. Represents a computing node l Data distribution Indicates from Mid-sampling feature z, Indicates from Mid-sampling feature z, This represents mapping the feature z to a higher-dimensional space, E[] represents calculating the mathematical expectation, and H represents the reproducing kernel Hilbert space. This indicates that the squared distance is calculated in the regenerating kernel Hilbert space; the total loss function of federated learning is: , This represents the alignment loss weight, typically a value between 0.05 and 0.2.

[0060] This invention adds feature alignment calculation, thereby ensuring that the features in the fused feature data obtained from the processed heterogeneous multi-source data are aligned, thus improving the accuracy of the model generated by the subsequent digital twin, and further improving the accuracy of monitoring. This is another important inventive concept of this invention.

[0061] In one embodiment, in the generation unit 203, the set of physical entity features is set as follows: ,in, For geometric features, For electrical characteristics, For spatial features, Given the time-series characteristics, the digital project model for power grid planning of the virtual twin is as follows: The twin mapping relationship is as follows:

[0062] in, The twin mapping function is implemented using a CNN-LSTM hybrid neural network to achieve multi-feature fusion mapping; This represents the set of learnable parameters for the twin model, including geometric modeling parameters, electrical simulation parameters, and spatiotemporal correlation parameters. This represents the model fitting error, which is controlled through optimization algorithms. To ensure the accuracy of twins.

[0063] In this invention, the fused feature data obtained through federated learning is used to construct a data-driven model for digital twins, achieving a precise mapping from physical entities to virtual models. Specifically, in the digital twin process, CNN is used to extract spatiotemporal geographic raster data and unstructured data from the fused feature data, while LSTM is used to extract data with temporal features from the fused feature data. This ensures the accuracy of data extraction and guarantees that the data is aligned, which is another important inventive concept of this invention.

[0064] In one embodiment, the geometric and spatial twin modeling employs BIM+GIS fusion modeling technology to construct a three-dimensional geometric model of the project, and the geometric feature mapping formula is: ; in, Indicates the first Three-dimensional geometric coordinate parameters of a device / structure Indicates the first The geometric weight coefficient of each piece of equipment / structure is determined by its importance. This represents a geospatial coordinate mapping function that integrates latitude, longitude, elevation, and topographic data. Based on power flow calculation and short-circuit calculation theories, an electrical and operational twin model is performed to simulate the power grid operation state before and after the commissioning of the planned project. The electrical characteristic mapping formula is as follows: ; in, These represent the node voltage and branch current phasors, respectively. These represent the branch impedance and admittance matrices, respectively. Active and reactive power respectively It represents the power factor and matches real-time data on actual grid operation. The formula for time-series twin modeling of the construction process is: ; in, express and The construction twin status is constantly monitored. This represents the incremental progress of construction per unit of time. This represents the construction progress rate function. It is influenced by factors such as manpower, materials, environment, and weather, and is driven by real-time data from the on-site Internet of Things (IoT).

[0065] This invention proposes specific formulas for geometric and spatial twin modeling, electrical and operational twin modeling, and construction process temporal twin modeling, thereby enabling the use of the established twin models to perform corresponding risk calculations. This is another important inventive concept of this invention.

[0066] In one embodiment, the potential risks include construction delay risk, power grid flow overload risk, and extreme weather risk.

[0067] In one embodiment, the prediction method for construction schedule delay risk is as follows: input the delay data of construction manpower, materials, and weather into the power grid planning digital project model, the power grid planning digital project model simulates the impact of schedule delay on the total project duration, subsequent acceptance, and power grid commissioning, and outputs the scope of delay and remedial measures suggestions; The prediction method for power grid power flow over-limit risk is as follows: use the power grid planning digital project model to simulate the new energy fluctuation and load surge conditions after the planned project is put into operation, calculate the power flow distribution in real time, warn of line overload and voltage over-limit risks, and output the grid optimization scheme; The method for predicting extreme weather risks is as follows: meteorological data such as heavy rain, strong winds, and high temperatures are input into the power grid planning digital project model. The power grid planning digital project model simulates construction safety hazards and equipment operation failures, delineates risk areas in advance, and formulates prevention and control plans.

[0068] In this invention, geometric and spatial twin modeling, electrical and operational twin modeling, and construction process temporal twin modeling are performed in the digital project model of power grid planning. By inputting relevant parameters, the corresponding risks can be calculated, and corresponding risk prevention and control schemes can be generated for user reference. This is another important inventive concept of this invention.

[0069] In one embodiment, the deviation value between the power grid planning project and the actual construction project is calculated, and the deviation contribution is calculated to determine the cause of the deviation between the power grid planning project and the actual construction project, so as to carry out construction rectification. The deviation value is calculated as follows: ; in, Indicates the deviation value. This indicates geometric layout deviation, calculated based on the differences in location and dimensions between the planned project and the actual construction. This indicates the deviation in electrical performance, calculated based on the difference between the design parameters of the planned project and the actual operating electrical parameters. This indicates the schedule deviation, calculated based on the difference between the planned project duration and the actual construction progress.

[0070] use , , Divide by respectively The invention obtains the contribution rates of deviations in geometric layout, electrical performance, and schedule. The largest contribution rate among these three factors is selected to determine the main cause of the deviation between the power grid planning project and the actual construction project, enabling corrective action during construction. This invention's key inventive concept lies in its ability to pinpoint the source of deviations through deviation contribution analysis, thus facilitating project rectification.

[0071] One embodiment of the present invention provides a computer storage medium storing a computer program. When the computer program on the computer storage medium is executed by a processor, the above-described method is implemented. The computer storage medium may be a hard disk, DVD, CD, flash memory, or other storage device.

[0072] For ease of description, the above apparatus is described by dividing it into various functional units. Of course, in implementing this application, the functions of each unit can be implemented in one or more software and / or hardware.

[0073] As can be seen from the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented by means of software plus necessary general-purpose hardware platforms. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the apparatus described in various embodiments or some parts of the embodiments of this application.

[0074] Finally, it should be noted that the above embodiments are for illustration only and not for limiting the technical solutions of the present invention. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the present invention without departing from the spirit and scope of the present invention. Any modifications or partial substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A monitoring method for power grid planning projects based on digital twins, characterized in that, The method includes: The data collection process involves collecting multi-source heterogeneous data from the entire process of implementing the power grid planning project. This multi-source heterogeneous data includes at least power grid operation data, engineering construction data, environmental meteorological data, and equipment status data. Fusion step: The heterogeneous multi-source data is fused to obtain fused feature data; The generation step involves performing geometric and spatial twin modeling, electrical and operational twin modeling, and construction process temporal twin modeling based on the fused feature data to obtain a digital project for power grid planning. The monitoring process involves performing simulations based on the power grid planning digital project to determine whether there are any potential risks in the power grid planning project. If potential risks are found, an alarm message is issued.

2. The method according to claim 1, characterized in that, In the data acquisition step, k computing nodes are used to acquire data. The acquired data includes structured time-series data, spatiotemporal geographic raster data, and unstructured data. The structured time-series data includes load, current, voltage, project progress, and weather. The spatiotemporal geographic raster data includes GIS data, tower locations, corridors, and terrain. The unstructured data includes project drawings, project images taken by UAVs, project feasibility study reports, and project change orders.

3. The method according to claim 2, characterized in that, In the fusion step, a federated learning model is composed of K local computing nodes. Each local computing node only processes local data, while the central node performs model aggregation and outputs fused feature data.

4. The method according to claim 3, characterized in that, In the generation step, let the set of physical entity features be... ,in, For geometric features, For electrical characteristics, For spatial features, Given the time-series characteristics, the digital project model for power grid planning of the virtual twin is as follows: The twin mapping relationship is as follows: ； in, The twin mapping function is implemented using a CNN-LSTM hybrid neural network to achieve multi-feature fusion mapping; This represents the set of learnable parameters for the twin model, including geometric modeling parameters, electrical simulation parameters, and spatiotemporal correlation parameters. This represents the model fitting error, which is controlled through optimization algorithms. To ensure the accuracy of twins.

5. The method according to claim 4, characterized in that, The potential risks include construction delays, power grid flow exceeding limits, and extreme weather.

6. A power grid planning project monitoring device based on digital twins, characterized in that, The device includes: The data acquisition unit collects multi-source heterogeneous data throughout the entire process of implementing the power grid planning project. The multi-source heterogeneous data includes at least power grid operation data, engineering construction data, environmental meteorological data, and equipment status data. Fusion unit: performs fusion processing on the heterogeneous multi-source data to obtain fused feature data; The generation unit performs geometric and spatial twin modeling, electrical and operational twin modeling, and construction process temporal twin modeling based on the fused feature data to obtain a digital project for power grid planning; The monitoring unit performs simulations based on the power grid planning digital project to determine whether there are potential risks in the power grid planning project. If there are potential risks, it issues an alarm message.

7. The apparatus according to claim 6, characterized in that, In the acquisition unit, k computing nodes are used to acquire data. The acquired data includes structured time-series data, spatiotemporal geographic raster data, and unstructured data. The structured time-series data includes load, current, voltage, project progress, and weather. The spatiotemporal geographic raster data includes GIS data, tower locations, corridors, and terrain. The unstructured data includes project drawings, project images taken by UAVs, project feasibility study reports, and project change orders.

8. The apparatus according to claim 7, characterized in that, In the fusion unit, a federated learning model is composed of K local computing nodes. Each local computing node only processes local data, while the central node performs model aggregation and outputs fused feature data.

9. The apparatus according to claim 8, characterized in that, In the generation unit, let the set of physical entity features be... ,in, For geometric features, For electrical characteristics, For spatial features, Given the time-series characteristics, the digital project model for power grid planning of the virtual twin is as follows: The twin mapping relationship is as follows: ； in, The twin mapping function is implemented using a CNN-LSTM hybrid neural network to achieve multi-feature fusion mapping; This represents the set of learnable parameters for the twin model, including geometric modeling parameters, electrical simulation parameters, and spatiotemporal correlation parameters. This represents the model fitting error, which is controlled through optimization algorithms. To ensure the accuracy of twins.

10. A computer storage medium, characterized in that, The computer storage medium stores a computer program, which, when executed by a processor, implements the method described in any one of claims 1-5.

Images