A line loss association management method based on digital twinning

Abstract

This invention discloses a line loss correlation management method based on digital twins, relating to the field of power management technology. The method includes: performing electrical data measurement and dual-field coupling reconstruction on a target power grid area to create a power grid twin; transmitting the real-time updated data layer from the power grid twin to the line loss analyzer through data interaction between the power grid twin and the line loss analyzer; executing line loss correlation management logic decisions and inverse data interaction verification; outputting line loss management instructions; and performing line loss management control on the target power grid area according to the line loss management instructions. This method addresses the problems of low automation and accuracy of the management process in power grid fluctuation scenarios in existing technologies, insufficient response efficiency from analysis to execution, and difficulty in adapting to the dynamic operation of the power grid and the precise management needs of power grid line losses. It can significantly improve the efficiency and effectiveness of line loss management and reduce power grid energy consumption.

Description

Technical Field

[0001] This invention relates to the field of power management technology, and specifically to a method for managing line loss correlation based on digital twins. Background Technology

[0002] With the continuous expansion of power grid scale and the increase in electricity demand, line losses not only affect the operating efficiency of the power grid, but also directly relate to energy waste and economic benefits. Currently, with the development of smart grid technology, new solutions are being offered through virtual modeling of the power grid and real-time data interaction.

[0003] However, due to the complexity and dynamic changes of the factors affecting line loss, and the lack of effective quantitative correlation analysis methods, it is difficult to dynamically reproduce the real-time operating status of the power grid, such as impedance changes and three-phase imbalance. This leads to problems such as the disconnect between line loss analysis and actual operating conditions, and inaccurate root cause identification.

[0004] In summary, existing technologies still suffer from low levels of automation and accuracy in the governance process under power grid fluctuation scenarios, and insufficient response efficiency from analysis to execution, making it difficult to adapt to the needs of dynamic power grid operation and precise management of power grid line losses. Summary of the Invention

[0005] This application provides a line loss correlation governance method based on digital twins, which is used to address the technical problems of low automation and accuracy of governance processes in power grid fluctuation scenarios in existing technologies, insufficient response efficiency from analysis to execution, and difficulty in adapting to the needs of dynamic power grid operation and precise management of power grid line losses.

[0006] In view of the above problems, this application provides a method for managing line loss correlation based on digital twins.

[0007] This application provides a line loss correlation management method based on digital twins. The method includes: using measured electrical data and dual-field coupling reconstruction of a target power grid area to create a power grid twin, wherein the dual fields include a dynamic impedance field and an energy flow field; transmitting the real-time updated data layer from the power grid twin to the line loss analyzer through data interaction between the power grid twin and the line loss analyzer, executing line loss correlation management logic decision-making and inverse data interaction verification, and outputting line loss management instructions; wherein the line loss management instructions include targeted work orders and target management plans, and the line loss correlation management logic includes the construction of a time-varying causal network based on voltage-line loss, causal key feature mining, and targeted plan generation and verification; and performing line loss management and control on the target power grid area according to the line loss management instructions.

[0008] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0009] This application provides a line loss correlation management method based on digital twins. By performing electrical data measurement and dual-field coupling reconstruction on a target power grid area, a power grid twin is created. Through data interaction between the power grid twin and a line loss analyzer, the real-time updated data layer in the power grid twin is transmitted to the line loss analyzer. Line loss correlation management logic decisions and inverse data interaction verification are executed, and line loss management instructions are output. Based on these instructions, line loss management and control are implemented on the target power grid area. This method addresses the problems of low automation and accuracy in the management process under power grid fluctuation scenarios in existing technologies, insufficient response efficiency from analysis to execution, and difficulty in adapting to the dynamic operation of the power grid and the precise management needs of power grid line losses. It can significantly improve the efficiency and effectiveness of line loss management and reduce power grid energy consumption. Attached Figure Description

[0010] Figure 1 This application provides a schematic diagram of a line loss correlation management method based on digital twins.

[0011] Figure 2 This application provides a schematic diagram of the construction process of the dynamic impedance field in a line loss correlation management method based on digital twins. Detailed Implementation

[0012] This application provides a line loss correlation governance method based on digital twins to address the problems of low automation and accuracy of governance processes in power grid fluctuation scenarios in existing technologies, insufficient response efficiency from analysis to execution, and difficulty in adapting to the needs of dynamic power grid operation and precise management of power grid line losses.

[0013] Example: Figure 1 , Figure 2 As shown, this application provides a method for managing line loss correlation based on digital twins, the method comprising:

[0014] S1: By performing electrical data measurement and dual-field coupling reconstruction on the target power grid area, a power grid twin is created, in which the dual fields include the dynamic impedance field and the power flow field.

[0015] In this embodiment, electrical data is first measured in the target power grid area to obtain electrical parameters of each key node in the target power grid area, including voltage, current, and power, providing basic data for subsequent twin construction. The scope of the electrical data measurement should cover representative loads, key power grid equipment, and major power transmission lines to ensure the comprehensiveness and representativeness of the data.

[0016] Subsequently, dual-field coupling reconstruction was performed based on the electrical data to construct a digital twin based on the power grid virtual model.

[0017] The dual-field coupling reconstruction process in this application involves constructing two parts: a dynamic impedance field and an electrical flow field. The dynamic impedance field reflects the relationship between current and voltage in the power grid, especially how the impedance characteristics of different nodes in the grid change over time. The electrical flow field represents the flow of electrical energy in the power grid, mainly reflecting the distribution of current and power. The coupling of these two fields can comprehensively describe the electrical behavior of the power grid, forming an accurate power grid model.

[0018] The construction of a dynamic impedance field involves measured data of voltage and current phasors. Through reverse power flow calculations, the equivalent impedance of each micro-segment in the power grid can be deduced. That is, based on the voltage and current differences between grid nodes, the equivalent impedance of each micro-segment, such as lines, connections, or user access points, can be calculated through reverse deduction. The key to this step is calculating the electrical impedance characteristics of each region of the power grid through measurements of the grid's topology and local data, and then modeling it as a dynamic impedance field.

[0019] When constructing the power flow field, it is first necessary to read the impedance dynamic characteristics of the dynamic impedance field. Combined with transformer outlet voltage and user load data as boundary conditions, the starting and ending points of the power flow are determined. Based on this, by using three-phase unbalanced power flow calculation based on distributed impedance, the current and power distribution in various regions of the power grid can be accurately calculated, and further modeled as a power flow field to reflect the power flow and loss in various regions of the power grid in real time.

[0020] Ultimately, the dynamic impedance field and the power flow field are used as the two basic data layers of the power grid twin. By separating the physical part and the data part of the power grid, the constructed power grid twin not only has the ability to accurately describe the current state of the power grid, but also provides strong data support for subsequent line loss analysis and management.

[0021] Furthermore, such as Figure 2 As shown, electrical data is measured and reconstructed using dual-field coupling for the target power grid area. The construction of the dynamic impedance field, in step S1 of this application, includes:

[0022] Based on the voltage and current phasors in the electrical data, reverse power flow calculation is performed through the voltage gradient to deduce the equivalent impedance of the transformer area and construct the dynamic impedance field. The voltage phasor includes amplitude and phase.

[0023] The calculation and deduction steps include: based on Kirchhoff's laws, using the voltage difference and current between adjacent nodes, the equivalent impedance value of each micro-segment is deduced in reverse, wherein the micro-segment type includes at least discrete micro-segments of the line, contacts, and user access points; using the power grid topology as a twin space, the equivalent impedance value is spatially interpolated and rendered to generate the dynamic impedance field.

[0024] In this embodiment of the application, during the process of constructing the dynamic impedance field, voltage phasors and current phasors are first obtained from the collected electrical data. Voltage phasors include voltage amplitude and voltage phase, and current phasors include current amplitude and current phase. They are important parameters reflecting the electrical state of the power grid.

[0025] In this application, reverse power flow calculations can be performed by combining voltage gradient, i.e., the rate at which voltage changes with spatial location, with current data, and then the equivalent impedance of the distribution area in the power grid can be deduced.

[0026] In one specific implementation, the equivalent impedance value of each micro-segment in the power grid is calculated using known voltage and current data. In this application, a micro-segment refers to a small section of line, connection point, or user access point in the power grid, representing different elements within the grid. The impedance characteristics of each micro-segment can be derived using Kirchhoff's laws, namely the voltage law and the current law.

[0027] In the specific calculation process, the equivalent impedance of each micro-segment is calculated by reverse deduction using the voltage difference and current data between adjacent nodes. Specifically, the voltage difference and current values ​​between adjacent nodes are first obtained. Then, using this difference, the equivalent impedance of each micro-segment is calculated through the ratio of current to voltage. That is, based on the measured single-point data, the impedance information of each micro-segment in the power grid is derived as the equivalent impedance value.

[0028] Next, spatial interpolation and rendering are performed on the determined equivalent impedance value. That is, the calculated discrete impedance data is transformed into a continuous impedance distribution through a mathematical model, so that the impedance characteristics of each region of the power grid can be accurately reflected in the twin.

[0029] Specifically, by using the grid topology as a twin space, the equivalent impedance values ​​of each micro-segment are precisely mapped and interpolated according to their location. Preferably, a rendering method can be used to characterize the changing state of impedance, such as the trend of color gradation, ultimately generating a dynamic impedance field.

[0030] In summary, this not only ensures the continuity of the dynamic impedance field, but also more realistically reproduces the electrical characteristics of different regions of the power grid, providing a more accurate basis for predicting power grid behavior and analyzing line losses.

[0031] Furthermore, electrical data is measured and reconstructed using dual-field coupling for the target power grid area. The construction of the electrical energy flow field, in step S1 of this application, includes:

[0032] The impedance dynamic characteristics based on the dynamic impedance field are read, and the transformer outlet voltage and user load data in the electrical data are read as boundary conditions. Based on the impedance dynamic characteristics and boundary conditions, the field quantity calculation results are determined by three-phase unbalanced power flow calculation based on distributed impedance. The field quantity calculation results include at least current vector and power vector. In the twin space, the field quantity calculation results are converted into current density vector field and line loss power scalar field to generate the power flow field.

[0033] In this embodiment of the application, during the process of constructing the power flow field, it is first necessary to read the impedance dynamic characteristics based on the dynamic impedance field, which reflects the changes in the electrical response of each node in the power grid over time, including but not limited to the relationship between current and voltage, and the characteristics of impedance changing with load, which can determine the response and performance of the power grid under different operating conditions.

[0034] Next, transformer outlet voltage and user load data are read from the electrical data as boundary conditions. The transformer outlet voltage represents the voltage source of the power grid, while the user load data reflects the power demand of the power grid under different load conditions. This data determines the starting and ending points of the power flow in the power grid, and also affects the power distribution in the power flow field.

[0035] Subsequently, based on the characteristics and boundary conditions of the dynamic impedance field, a three-phase unbalanced power flow calculation method based on distributed impedance is adopted to calculate the current and power distribution of the power grid. Specifically, considering the load imbalance and the differences between phases in the power grid, parameters such as voltage, current, and load are defined for each node in the power grid topology. The line impedance between each node is refined into the impedance difference of each phase, which can reflect the imbalance in the power grid. Next, the load and voltage conditions of each node are set to ensure that they reflect the actual operating state. The impedance value of the line is determined by resistance and reactance. In the calculation process, Kirchhoff's voltage law and current law are used to establish the relationship between voltage and current between nodes, and the voltage and current of each node in the power grid are solved iteratively through this relationship. The impact of unbalanced load on current and power distribution can be accurately analyzed to determine the power flow and loss of each line and node, and the current vector and power vector of each node in the power grid are obtained as the calculation results of the field quantities.

[0036] Among them, the current vector represents the distribution of current in various directions, while the power vector reflects the flow and conversion efficiency of electrical energy.

[0037] Subsequently, in the twin space, the field calculation results are converted into a current density vector field and a line loss power scalar field. The current density vector field shows the distribution density of current in space, while the line loss power scalar field reflects the power loss caused by current flow in the power grid, thus determining the energy flow field.

[0038] In summary, the power flow field in the digital twin reflects the real-time operating status of the power grid, providing comprehensive data support for subsequent power grid dispatching, line loss management, and optimization decisions.

[0039] Furthermore, step S1 of this application includes: obtaining the power grid topology of the target power grid area, using the power grid topology as the base layer, the dynamic impedance field as the first data layer, and the power flow field as the second data layer to construct the power grid twin; wherein, the power grid twin is dynamically updated and has corrections based on dynamic household-transformer relationships.

[0040] In this embodiment of the application, during the process of constructing a power grid twin, the power grid topology of the target power grid area is obtained, that is, the various nodes in the power grid, such as generators, transformers, contacts, user access points, etc., and their connection relationships, in order to construct the spatial structure of the power grid to reflect the actual connection and operation of the power grid, and to provide a basis for subsequent analysis and optimization.

[0041] Next, the power grid topology is used as the base layer, or foundational support layer, for further modeling and analysis. Based on this, the dynamic impedance field, as the first data layer, and the power flow field, as the second data layer, together constitute the core content of the power grid twin. The dynamic impedance field contains the impedance characteristics of each node in the power grid, which changes dynamically with variations in load and operating conditions. The power flow field demonstrates the distribution of electrical energy within the power grid, reflecting changes in current and power. By superimposing these two data layers onto the power grid topology, a three-dimensional virtual model encompassing the power grid's spatial structure and electrical characteristics—the power grid twin—is formed.

[0042] In the specific implementation process, the real-time collected electrical data is synchronized to the power grid topology of the base layer, serving as a virtual synchronization of the operation of the power grid physical entity, and the operation status of data dimensions is synchronized and analyzed at each data layer.

[0043] Specifically, since the operating status of the power grid changes over time, the power grid twin needs to receive electrical data from the power grid monitoring system in real time and update it dynamically, including but not limited to real-time electrical parameters such as voltage, current, and power.

[0044] Furthermore, this application preferably includes a correction based on dynamic user-transformer relationships in the power grid twin. User loads in the power grid vary over time; for example, user electricity demand may fluctuate significantly at different times or in different seasons. Therefore, the power grid twin needs to be corrected accordingly based on these dynamic changes. By correcting the dynamic user-transformer relationships, it can be ensured that the power flow field and dynamic impedance field in the twin accurately reflect the actual load changes in various areas of the power grid, thereby improving the accuracy and practicality of the power grid twin.

[0045] In summary, the power grid twin, by combining power grid topology, dynamic impedance field, and power flow field, forms a comprehensive virtual power grid model, ensuring more precise and efficient management and optimization of the power grid.

[0046] S2: Through data interaction between the power grid twin and the line loss analyzer, the real-time updated data layer in the power grid twin is transmitted to the line loss analyzer, the line loss associated governance logic decision and reverse data interaction verification are executed, and the line loss governance instructions are output.

[0047] The line loss management instruction includes targeted work orders and target management plans, and the line loss associated management logic includes the construction of a time-varying causal network based on voltage-line loss, the mining of key causal features, and the generation and verification of targeted plans.

[0048] S3: Perform line loss management and control on the target power grid area according to the line loss management instruction.

[0049] In this embodiment, data interaction occurs between the power grid twin and the line loss analyzer. Specifically, real-time collected electrical data is synchronized to the power grid topology of the power grid twin, and further, a directional conversion of data dimensions is performed between the first and second data layers. Subsequently, the real-time updated data layer in the power grid twin is transmitted to the line loss analyzer for further line loss analysis. The timely and accurate information transmission between the power grid twin and the line loss analyzer ensures the effectiveness of line loss mitigation.

[0050] Subsequently, in the line loss analyzer, a time-varying causal network based on the relationship between voltage and line loss is first constructed based on data transmitted from the power grid twin. Specifically, based on the dynamic relationship between voltage and line loss, by calculating the changes in line loss under different voltage conditions, potential problem points and key factors affecting line loss in the power grid are identified. This not only reflects the time-varying relationship between voltage and line loss in the power grid, but also reveals the causal influence between different parts of the power grid, in order to identify the root cause of line loss problems.

[0051] Furthermore, by mining key causal features, the features and key paths that have the greatest impact on line loss can be identified from the causal network. This application includes, but is not limited to, key causal paths, root cause nodes, and causal contribution. It can generate targeted governance plans for each region or link of the power grid and formulate personalized governance strategies for different line loss problems.

[0052] In this application, the line loss mitigation instructions include targeted work orders and target mitigation plans. Targeted work orders are repair instructions for specific problem points, while target mitigation plans are line loss optimization schemes formulated for the entire power grid area based on the results of causal network analysis and feature mining. These plans may include measures such as adjusting certain nodes in the power grid, optimizing equipment configuration, or redistributing loads.

[0053] Finally, by executing the aforementioned line loss management instructions, line loss management and control are implemented in the target power grid area. Specifically, based on targeted work orders, maintenance personnel can promptly and specifically address line loss issues in the power grid, while the target management plan provides optimal maintenance strategies or overall strategies for long-term optimization of power grid efficiency and reduction of line losses. This can effectively alleviate power grid line loss problems, thereby improving the operational efficiency of the power grid.

[0054] Therefore, the data interaction between the power grid twin and the line loss analyzer, the construction of time-varying causal networks and the mining of causal characteristics, as well as the generation of targeted work orders and target governance plans, constitute a closed-loop line loss governance and control system, which helps to achieve precise management and optimization of the power grid.

[0055] Furthermore, prior to data interaction between the power grid twin and the line loss analyzer, the construction of the line loss analyzer, in step S2 of this application, includes:

[0056] The system uses a logical architecture consisting of a first causal construction node, a second feature inference node, and a third target location node. Through supervised logical training until convergence, a line loss analyzer is determined. The power grid twin and the line loss analyzer are embedded plugins deployed in the power grid centralized control platform. Specifically, the first causal construction node constructs a time-varying causal network, the second feature inference node mines key causal features (including at least key causal paths, root cause nodes, causal contribution, and marginal causal effects), and the third target location node makes targeted governance decisions.

[0057] In this embodiment, the line loss analyzer is constructed using a three-stage logical architecture, including a first causal construction node, a second feature inference node, and a third target location node, each performing different functions to ensure the efficiency and accuracy of the line loss analyzer in analyzing and managing power grid line losses.

[0058] First, the first causal construction node is responsible for constructing the time-varying causal network. Specifically, based on real-time updated data provided by the power grid twin, it constructs the time-varying causal relationship between voltage and line loss. The construction of the causal network analyzes the dynamic changes between voltage and line loss data to uncover key factors affecting line loss in the target power grid area, revealing how voltage fluctuations affect line loss under different operating conditions of the power grid, and the causal chains between fluctuation changes.

[0059] Preferably, the construction of time-varying causal networks can not only reflect the instantaneous behavior of the power grid during operation, but also capture the changing trend of line losses during long-term operation.

[0060] Next, the second feature inference node performs the mining of key causal features. By analyzing the relationship between voltage and line loss in the causal network, it infers the key causal features affecting line loss, including at least: key causal paths, i.e., voltage change paths affecting line loss; root cause nodes, i.e., the initial causes leading to line loss problems; causal contribution, i.e., the degree of influence of each node on line loss; and marginal causal effects, i.e., the response effect to small disturbances in the system. These key causal features provide a deep understanding of the relationship between each node and line loss during power grid operation, thereby identifying key factors affecting power grid efficiency and losses.

[0061] Finally, the third target location node is responsible for executing targeted governance decisions. Specifically, based on the causal features mined from the aforementioned nodes, targeted governance decisions are generated. Optionally, areas or links in the target power grid region that require key governance are identified, and corresponding governance strategies are generated. Examples of strategy dimensions include: optimization and adjustment of specific power grid nodes, equipment upgrades, or load redistribution. The formulation of targeted governance decisions considers the causal contribution and influence path of each node, thereby achieving precise control of power grid line losses.

[0062] Optionally, the line loss analyzer is constructed using a logic-supervised training to convergence approach. That is, using the aforementioned logical architecture and line loss analysis logic as a baseline, its analysis and decision-making processes are continuously optimized through data training until convergence conditions are met, such as the identification and decision-making meeting a preset accuracy, thus obtaining the completed line loss analyzer.

[0063] In the implementation of this application, the power grid twin and the line loss analyzer work together as embedded plug-ins deployed in the power grid centralized control platform. Through real-time data interaction and feedback, a complete intelligent power grid management system is formed. This not only improves the accuracy of line loss analysis but also ensures the efficiency and intelligence of power grid optimization management.

[0064] Furthermore, the first causal construction node performs the construction of the time-varying causal network, and step S2 of this application includes:

[0065] The first and second data layers, which are updated in real time in the power grid twin, are transmitted to the line loss analyzer; the first causal construction node is driven to identify line loss anomalies and calculate the time-varying causal strength between the voltage of each node and the system line loss by adopting a time-varying convergence crossover method with a sliding time window as a constraint; a time-varying causal network based on voltage-line loss is constructed with voltage monitoring points as nodes and time-varying causal strength as time-varying edge weights.

[0066] In this embodiment, the first data layer (dynamic impedance field data layer) and the second data layer (power flow field data layer) in the power grid twin, which are updated in real time, are transmitted to the line loss analyzer. This enables accurate line loss analysis and mitigation decisions.

[0067] Subsequently, the first causal construction node in the line loss analyzer is driven to construct a time-varying causal network. The main task here is to identify line loss anomalies in the power grid and analyze the relationship between these anomalies and the voltages of various nodes in the grid. By calculating the time-varying causal strength between voltage and line loss in the power grid, the impact of voltage fluctuations on line loss at different time points and under different grid conditions is identified.

[0068] In a preferred embodiment of this application, in order to accurately capture time-varying causal relationships in the power grid, a time-varying convergence crossover method is adopted, and a processing method based on multiple iterations and convergence is used to ensure the accuracy and stability of causal strength.

[0069] Specifically, under the constraint of a sliding time window, the relationship between the voltage of each node in the power grid and the system line loss is calculated. Optionally, the sliding time window can be dynamically moved according to a certain time step during the calculation process. Based on the voltage and line loss data within each time window, the time-varying causal strength between the voltage of each node and the system line loss is calculated. This method can capture instantaneous changes in power grid operation and infer the specific impact on line loss based on the iterative trend of these instantaneous changes.

[0070] Next, voltage monitoring points are used as nodes, and time-varying causal strength is used as the time-varying edge weights between these nodes. In this application, voltage monitoring points represent key locations in the target power grid area where actual voltage is measured, while time-varying causal strength represents the strength of the causal relationship between voltage fluctuations and line losses between a voltage monitoring point and other power grid nodes. In the construction of the time-varying causal network in this application, the determined causal strength is used as edge weights to connect voltage monitoring points with other nodes, thereby forming a time-varying causal network based on voltage and line losses. Simultaneously, the causal strength changes as the power grid operates.

[0071] In summary, the constructed time-varying causal network not only demonstrates the causal strength of each voltage monitoring point in the power grid, but also reveals the dynamic relationship between voltage fluctuations and line losses in the power grid. This provides accurate data support for subsequent line loss management and enables timely detection of line loss anomalies during power grid operation, achieving precise line loss monitoring and optimized management.

[0072] Furthermore, the second feature inference node performs causal key feature mining, and step S2 of this application includes:

[0073] The second feature inference node is driven to calculate the causal outflow centrality for each node in the time-varying causal network, wherein the sum of the causal strengths of the nodes pointing to other nodes under the guidance of line loss is used as the calculation method; and a screening based on the causal outflow centrality is performed with a preset threshold to determine the key causal path and root cause node.

[0074] In this embodiment, the main task of the second feature reasoning node is to calculate key causal features based on each node in the time-varying causal network, so as to further analyze the causal relationships in the power grid, especially the dynamic relationship between line loss and each node of the power grid.

[0075] Specifically, regarding the analysis of causal outflow centrality, in this application, causal outflow centrality is used to quantify the importance of nodes in a network, and its importance in the network is determined by assessing the influence of a node on other nodes.

[0076] In this method, the quantification of causal outflow centrality focuses on the impact of nodes on system line losses. Therefore, in the calculation method of this application, the calculation of causal outflow centrality is based on the sum of causal strengths from a node to other nodes under the guidance of line loss. Specifically, for each node in a time-varying causal network, causal outflow centrality reflects the sum of causal strengths from that node to other nodes. This characteristic indicates that if a node has strong causal outflow centrality, it has a greater impact on the line losses of other nodes in the power grid and may be a key node causing line loss anomalies.

[0077] Specifically, the first step is to determine the causal strength of each node in the power grid, which is the strength of the causal relationship between that node and other nodes regarding voltage changes and line losses. This reflects the degree of direct correlation between voltage fluctuations and line loss changes at a node. Then, the causal strengths are accumulated in each direction to obtain the influence of each node on other nodes, i.e., the causal outflow centrality.

[0078] Among them, nodes with high causal outflow centrality usually mean that changes in voltage or load at that node have a significant impact on the overall line loss of the power grid, and therefore require special attention and optimization.

[0079] Next, a filtering operation is performed based on causal outflow centrality. By setting a preset threshold, nodes whose causal outflow centrality exceeds the threshold are filtered out; these are typically key nodes in the power grid that have a significant impact on line losses. Optionally, the preset threshold can be customized based on line loss monitoring requirements. Ensuring that the filtered nodes are those that significantly impact line losses under current regulatory requirements effectively reduces redundant information and focuses on key factors affecting line losses.

[0080] Finally, by selecting nodes with high causal outflow centrality, critical causal paths and root cause nodes are analyzed and identified. In this application, a critical causal path refers to the causal chain between voltage fluctuations and line losses in a time-varying causal network, which illustrates the path of line loss impact from one node to another. Root cause nodes are the initial nodes that cause abnormal power grid line losses, i.e., the source nodes.

[0081] In summary, by identifying key causal paths and root cause nodes, we can accurately identify the areas and links in the power grid that most need governance, providing a scientific basis for subsequent governance decisions.

[0082] In summary, by calculating causal outflow centrality, performing screening operations, and identifying key causal paths and root cause nodes, important feature support is provided for the precise management of power grid line losses. This support can directly guide subsequent targeted management decisions, ensuring the efficiency and relevance of line loss management.

[0083] Furthermore, step S2 of this application includes:

[0084] Define causal contribution and quantify marginal causal effect; wherein, the causal contribution of a node within a preset time period is defined by the integral of the causal intensity within the preset time period;

[0085] The quantification method of marginal causal effect is as follows: in the power grid twin, by applying a unit perturbation to the first node, intervention simulation is performed to determine the bus loss change, wherein the first node is any node in the time-varying causal network; the bus loss change is normalized as the marginal causal effect of the first node.

[0086] In the embodiments of this application, causal contribution and marginal causal effect are key indicators for quantifying the impact of each node on line loss in the power grid, respectively reflecting the long-term impact of a node on the power grid and the short-term impact of a single node on the power grid line loss.

[0087] The causal contribution rate proposed in this application refers to the integral of causal strength over a preset time period. Specifically, the causal contribution rate measures the degree to which a node in the power grid contributes to line losses, reflecting the cumulative impact of voltage fluctuations and current changes at that node on overall line losses over a certain period. One possible calculation method is to integrate the causal strength of the node.

[0088] Since causal strength represents the strength of the causal relationship between a node and other nodes regarding voltage and line loss, it reflects the direct impact of node changes on line loss. By integrating the causal strength over a preset time period, the resulting causal contribution can measure the sustained impact of a node on line loss.

[0089] For example, voltage fluctuations at node A over a certain period of time may cause current changes in multiple surrounding nodes, thereby affecting the line loss of the entire power grid. By integrating this effect, the causal contribution of that node can be obtained.

[0090] The synchronous, marginal causal effect quantification method involves applying unit perturbations to each node in the power grid twin to conduct intervention simulations, thereby observing the impact of changes in that node on the overall line loss of the power grid. The specific implementation steps are as follows:

[0091] First, in the power grid twin, for any node, namely the first node, a unit perturbation is applied to that node. In this application, the perturbation refers to a very small change in the node's voltage, load, or other electrical parameters, preferably a controllable disturbance with minimal impact on other parts of the power grid. Specifically, a unit perturbation refers to a one-unit change in the voltage or load of the first node.

[0092] Subsequently, after applying a perturbation, an intervention simulation is conducted to simulate the impact of the perturbation on other nodes in the power grid system and the overall line loss. Specifically, a simulation is performed in the base layer of the power grid topology in the power grid twin. Through data layer transformation analysis, the overall response after the perturbation, especially the change in line loss, can be obtained.

[0093] Furthermore, the change in line loss is determined; that is, through simulation, the change in line loss of the power grid system under a unit perturbation is calculated, which reflects how much the line loss in the power grid changes due to the perturbation to the first node. This quantified change value serves as the basis for assessing the impact of nodes on line loss.

[0094] Subsequently, the calculated change in line loss is normalized by comparing it to a reference value, such as the system's maximum or average line loss, to eliminate differences in power grid system size and operating conditions. This normalization process yields a standardized and easily comparable causal effect index.

[0095] Ultimately, the normalized change in line loss represents the marginal causal effect of the first node, measuring its marginal contribution to line loss variation in the power grid and reflecting its sensitivity and influence under small disturbances. A large marginal causal effect of a node indicates significant control over power grid line losses; even minor changes can lead to substantial fluctuations in line losses.

[0096] Similarly, the above steps are performed for each node to determine the marginal causal effect of each node.

[0097] By calculating causal contribution and marginal causal effect, we can comprehensively assess the contribution and impact of each node in the power grid on line loss. This helps power grid managers accurately identify key nodes affecting line loss, providing a scientific basis for subsequent line loss optimization and management strategies, and supporting efficient management and optimized scheduling of the power grid.

[0098] Furthermore, the third target location node performs targeted governance decisions, and step S2 of this application includes:

[0099] The third target location node reads the time-varying causal network and key causal features; using a preset line loss threshold as a judgment condition, it automatically triggers a targeted work order based on the time-varying causal network; as the targeted work order is generated, using key causal features as matching criteria and the governance strategy library as a data pool, it performs governance measures matching and governance priority ranking for each line loss anomaly point to generate a first governance plan; it performs twin simulation and plan optimization on the first governance plan to generate a target governance plan.

[0100] In this embodiment, the third target location node first reads the time-varying causal network and key causal features, captures the dynamic relationship between each node of the power grid and the overall system, and includes information such as key causal paths, root cause nodes, causal contribution, and marginal causal effects. Subsequently, the third target location node identifies the line loss anomalies that need to be focused on through causal analysis, providing decision support for subsequent governance measures.

[0101] Next, a targeted work order is automatically triggered based on a time-varying causal network, using a preset line loss threshold as the judgment condition. In this application, the preset line loss threshold refers to the normal line loss range set according to the power grid's operating standards and historical data. When the line loss of a certain part of the power grid exceeds this threshold, it indicates that there is an anomaly or potential fault in that area. At this time, the target location node will automatically generate a targeted work order, indicating the specific area or node that needs to be handled, and assigning the remediation task.

[0102] In one alternative embodiment, the generation of targeted work orders is automatically triggered based on anomalies detected in the causal network, without the need for manual intervention, thereby improving the response speed and automation level of line loss management.

[0103] As targeted work orders are generated, the third target location node will use causal key features as matching criteria and the governance strategy library as a data pool to match governance measures and prioritize governance for each line loss anomaly point.

[0104] Among these, key causal features are a quantitative description of the influence relationships between nodes in the power grid. For example, a node with a large causal contribution may have a significant impact on the line losses of the entire power grid, and therefore should be prioritized for remediation. The remediation strategy library contains various strategies for addressing different line loss problems, such as equipment maintenance, load adjustment, and line reconfiguration, and can match suitable remediation measures based on causal features. By matching the causal features of anomalies, the third target point location node can determine the priority and implementation order of each remediation measure.

[0105] After matching and prioritizing the mitigation measures, a first mitigation plan was generated, which includes mitigation measures, priority ranking, and implementation plans for each abnormal line loss point in the power grid. This first mitigation plan serves as the initial mitigation strategy, and further optimization is performed using a power grid twin, conducting simulation calculations.

[0106] Specifically, in the twin simulation phase, the various governance measures in the first governance plan will be simulated in a virtual power grid model. The actual effects on power grid operation and line loss reduction will be evaluated through data layer responses. Simulations can predict the impact of different governance measures on the power grid; for example, whether the maintenance of a certain piece of equipment can effectively reduce line losses, or whether the reconfiguration of a certain line can improve power grid stability. The simulation results provide data support for optimizing the governance plan.

[0107] Guided by this, the first governance plan was iteratively optimized and simulated until the best solution was obtained.

[0108] Finally, after completing the twin simulation and contingency plan optimization, a target governance plan is generated, which includes the most effective governance measures and the optimal implementation sequence. This plan serves as a guide for actual governance, ensuring that line loss problems in the power grid are effectively resolved while maintaining the grid's stability and efficient operation.

[0109] In summary, the third target location node automatically triggers a targeted work order by reading the time-varying causal network and key causal features, combined with a preset line loss threshold. It then matches and prioritizes governance measures based on causal features and a governance strategy library, ultimately generating a first governance plan. After twin simulation and optimization, a target governance plan is formed.

[0110] The preferred embodiment of this application ensures the accuracy, automation, and efficiency of line loss management, thereby improving the operating efficiency and reliability of the power grid.

[0111] Furthermore, step S2 of this application includes:

[0112] Based on the bidirectional interaction channel between the power grid twin and the line loss analyzer, the first governance plan is transmitted to the base layer of the power grid twin, and data layer trend analysis and multi-round iterative optimization under the power grid topology simulation are performed to generate the target governance plan; based on the targeted work order and the target governance plan, a line loss governance instruction is generated.

[0113] In this embodiment, the first governance plan is transmitted to the base layer of the power grid twin through a bidirectional interactive channel between the power grid twin and the line loss analyzer. The base layer of the power grid twin is the basic structure of the power grid model, containing the topology information of the power grid and basic electrical parameters in the power grid, such as nodes, lines, loads, and equipment.

[0114] Upon receiving the first governance plan, the power grid twin performs a power grid topology simulation, simulating and predicting the power grid state based on the existing power grid topology and the governance measures proposed in the first governance plan.

[0115] Specifically, virtual simulation is used to model the impact of various measures in the first governance plan on power grid operation. By rehearsing the power grid topology and analyzing the responses at the first and second data layers, the specific impacts of different governance schemes on power grid stability, operational efficiency, and line losses can be assessed.

[0116] In this process, the power grid twin adjusts the power grid topology to predict the power grid performance after the implementation of governance measures. Specifically, it involves trend analysis of the power grid topology and data layer, that is, analyzing how the power grid data layer changes with the implementation of governance measures during the simulation process.

[0117] Subsequently, the governance scheme was iteratively optimized through multiple rounds based on the evaluation results. Specifically, in each iteration, the power grid topology and governance scheme were adjusted according to the simulation results of the previous round to maximize the effect of line loss mitigation.

[0118] Optionally, in each round of optimization, the power grid twin reassesses the operational status of each node and line in the power grid, adjusting the implementation sequence and priority of mitigation measures to achieve the best line loss mitigation effect. Through the above iterative optimization steps, the mitigation plan is continuously adjusted and improved to ensure the most effective solution.

[0119] Ultimately, the optimal line loss mitigation solution, selected after multiple rounds of iterative optimization, was chosen as the target mitigation plan, aiming to achieve optimal line loss control and improve power grid operating efficiency. This plan includes detailed mitigation measures and implementation sequence, ensuring that power grid line loss issues can be resolved efficiently and accurately.

[0120] Subsequently, based on the targeted work orders and the target governance plan, line loss mitigation instructions are generated. The targeted work orders indicate the key power grid areas or nodes requiring focused mitigation, while the target governance plan provides specific measures and priorities. Combined, the generated line loss mitigation instructions include specific execution steps, required resources, and a timeline, guiding the actual line loss mitigation work and ensuring that mitigation measures can be implemented quickly and effectively.

[0121] In summary, the line loss management process achieves a closed loop from data analysis and simulation optimization to actual implementation, ensuring accurate management and efficient control of power grid line losses.

[0122] This application provides a line loss correlation management method based on digital twins, which has the following technical effects:

[0123] 1. A grid twin integrating dynamic impedance field and power flow field is constructed. Based on measured electrical data and dual-field coupling reconstruction, the grid operating state is accurately reproduced, providing a high-fidelity digital mapping foundation for line loss management and avoiding the problem of traditional models being out of touch with actual operating conditions. Specifically, the dynamic impedance field is used to deduce the equivalent impedance of the transformer area through reverse power flow calculation, while the power flow field is used to calculate three-phase unbalanced power flow by combining impedance characteristics and boundary conditions. The dual fields work together to capture the dynamic changes in grid line losses, improving the comprehensiveness and accuracy of line loss correlation analysis.

[0124] 2. The line loss analyzer is based on a logical architecture of time-varying causal network construction, key feature mining, and targeted decision-making. It calculates time-varying causal strength through a sliding time window, accurately identifying critical paths and root cause nodes of line loss anomalies, avoiding indiscriminate remediation. It quantifies causal contribution and marginal causal effects, clarifying the impact weight of each node on line loss through node perturbation simulation and bus line loss change analysis, providing a quantitative basis for targeted remediation and improving the pertinence of remediation measures. Based on the time-varying causal network, it automatically triggers targeted work orders, matches measures with a remediation strategy library, prioritizes them, and generates target remediation plans through multiple rounds of optimization via twin simulation, achieving automated, precise, and efficient line loss remediation.

[0125] Through the foregoing detailed description of a line loss correlation management method based on digital twins, those skilled in the art can clearly understand the line loss correlation management method based on digital twins in this embodiment. As for the apparatus disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and relevant parts can be referred to the method section description.

[0126] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for managing line loss correlation based on digital twins, characterized in that, The method includes: By performing electrical data measurements and dual-field coupling reconstruction on the target power grid area, a power grid twin is constructed. The dual fields include a dynamic impedance field and an electrical energy flow field. The dynamic impedance field serves as the first data layer, and the electrical energy flow field serves as the second data layer. Through data interaction between the power grid twin and the line loss analyzer, the real-time updated data layer in the power grid twin is transmitted to the line loss analyzer to execute line loss associated governance logic decision-making and reverse data interaction verification, and output line loss governance instructions; The line loss management instructions include targeted work orders and target management plans; According to the line loss management instructions, line loss management and control are carried out in the target power grid area; The logical architecture consists of a first causal construction node, a second feature reasoning node, and a third target localization node. The line loss analyzer is determined by training to convergence through logical supervision. The power grid twin and the line loss analyzer are embedded plug-ins deployed in the power grid centralized control platform. The first causal construction node performs the construction of the time-varying causal network, including: The first and second data layers, which are updated in real time in the power grid twin, are transmitted to the first causal construction node. By identifying abnormal line loss points, a time-varying convergence crossover method is adopted, with a sliding time window as a constraint, to calculate the time-varying causal strength between the voltage of each voltage monitoring point and the system line loss. A time-varying causal network based on voltage-line loss is constructed, with voltage monitoring points as nodes and time-varying causal strength as time-varying edge weights. The second feature inference node performs causal key feature mining, wherein the causal key features include at least key causal paths, root cause nodes, causal contribution, and marginal causal effects, including: The second feature inference node is driven to calculate the causal outflow centrality for each node in the time-varying causal network, wherein the sum of the causal strengths of the nodes pointing to the other nodes under the guidance of line loss is used as the calculation method. Using a preset threshold, a screening based on the causal outflow centrality is performed to determine key causal paths and root cause nodes; Define causal contribution and quantify marginal causal effect; wherein, the causal contribution of a node within a preset time period is defined by the integral of the causal intensity within the preset time period; The quantification method for marginal causal effect is as follows: In the power grid twin, by applying a unit perturbation to the first node, intervention simulation is performed to determine the bus loss change, wherein the first node is any node in the time-varying causal network; The bus loss variation is normalized and taken as the marginal causal effect of the first node; The third target location node performs targeted governance decisions.

2. The line loss correlation management method based on digital twin as described in claim 1, characterized in that, The construction of the dynamic impedance field includes: electrical data measurement and dual-field coupling reconstruction of the target power grid area. Based on the voltage and current phasors in the electrical data, reverse power flow calculation is performed through the voltage gradient to deduce the equivalent impedance of the transformer area and construct the dynamic impedance field. The voltage phasor includes amplitude and phase. The calculation and deduction steps include: According to Kirchhoff's laws, the equivalent impedance value of each micro-segment is deduced in reverse by using the voltage difference and current between adjacent nodes. The micro-segment types include at least discrete micro-segments of the line, contacts, and user access points. Using the power grid topology as a twin space, spatial interpolation and rendering are performed on the equivalent impedance value to generate the dynamic impedance field.

3. The method for managing line loss correlation based on digital twins as described in claim 2, characterized in that, The electrical data of the target power grid area is measured and reconstructed using dual-field coupling. The construction of the electrical energy flow field includes: Read the impedance dynamic characteristics based on the dynamic impedance field, and read the transformer outlet voltage and user load data in the electrical data as boundary conditions; Based on the aforementioned impedance dynamic characteristics and boundary conditions, the field quantity calculation results are determined using three-phase unbalanced power flow calculation based on distributed impedance. The field quantity calculation results include at least the current vector and the power vector. In the twin space, the calculated field results are converted into a current density vector field and a line loss power scalar field to generate the electrical energy flow field.

4. The method for managing line loss correlation based on digital twins as described in claim 3, characterized in that, Obtain the power grid topology of the target power grid area, and use the power grid topology as the base layer, combined with the dynamic impedance field and the power flow field, to construct the power grid twin; The power grid twin is dynamically updated and includes corrections based on dynamic household-transformer relationships.

5. The method for managing line loss correlation based on digital twins as described in claim 1, characterized in that, The third target location node performs targeted governance decisions, including: The third target location node reads the time-varying causal network and key causal features; Using a preset line loss threshold as a judgment condition, a targeted work order is automatically triggered based on the time-varying causal network; As the targeted work order is generated, the causal key features are used as the matching criteria, and the governance strategy library is used as the data pool to match governance measures and prioritize governance for each line loss anomaly point, thereby generating the first governance plan. Perform twin simulation and optimization on the first governance plan to generate the target governance plan.

6. The method for managing line loss correlation based on digital twins as described in claim 5, characterized in that, Based on the bidirectional interaction channel between the power grid twin and the line loss analyzer, the first governance plan is transmitted to the base layer of the power grid twin, and data layer trend analysis and multi-round iterative optimization under the power grid topology simulation are performed to generate the target governance plan. Based on the targeted work order and the target governance plan, a line loss management instruction is generated.

Images