Power distribution network intelligent dispatching method and system based on digital twinning

Abstract

The application discloses a power distribution network intelligent scheduling method and system based on digital twinning, and belongs to the field of power distribution network scheduling, wherein the method comprises the following steps: acquiring multi-source heterogeneous data of a power distribution network, and performing fusion preprocessing on the multi-source heterogeneous data; constructing a power distribution network digital twinning model based on the data after the fusion preprocessing; performing abnormal diagnosis on the power distribution network based on the power distribution network digital twinning model; performing situation prediction in the power distribution network digital twinning model based on the power distribution network, to obtain a weak resilience node; performing sand table deduction verification based on the weak resilience node, to confirm an executable strategy; issuing the executable strategy to the power distribution network for execution, and performing evolution correction on the power distribution network digital twinning model based on an actual running state. The application solves the technical problems that the existing power distribution network scheduling technology has insufficient risk prediction ability and is prone to causing secondary operation risks.

Description

Technical Field

[0001] This invention relates to the field of distribution network dispatching, and specifically to a distribution network intelligent dispatching method and system based on digital twins. Background Technology

[0002] With the large-scale integration of distributed renewable energy into the distribution network, the electricity load patterns of users are becoming increasingly diverse and complex. The uncertainty and dynamic characteristics of distribution network operation have been significantly enhanced, placing higher demands on the accuracy, foresight, and security of dispatching decisions. Traditional distribution network dispatching modes rely heavily on offline static simulation models and human experience-based decision-making. It is difficult to achieve deep integration of multi-source heterogeneous electrical operation data and external environmental data in the spatiotemporal dimensions. There are problems of disconnect between the physical power grid and the simulation model, and the state synchronization is lagging behind. Not only does the diagnosis of abnormal faults lag behind the development of risks and make it difficult to identify weak links in the grid resilience in advance, but the dispatching strategy also lacks a pre-emptive full-scenario simulation verification process. Directly issuing and executing the strategy can easily lead to secondary operational risks such as overload of adjacent branches and voltage exceeding limits at nodes. At the same time, static simulation models cannot dynamically iterate and evolve with the actual operating characteristics of the power grid. The simulation accuracy continues to decline with equipment aging and grid adjustments, making it difficult to adapt to the dynamic dispatching needs of distribution networks under high-penetration renewable energy. Summary of the Invention

[0003] This application provides a digital twin-based intelligent dispatching method and system for distribution networks, aiming to solve the technical problems of insufficient risk prediction capabilities in existing distribution network dispatching technologies, which easily lead to secondary operational risks. It achieves the technical effect of improving dispatching decision-making efficiency and power grid operation economy while ensuring the safe operation of the distribution network.

[0004] In view of the above problems, this application provides a method and system for intelligent dispatching of distribution networks based on digital twins.

[0005] The first aspect disclosed in this application provides a method for intelligent dispatching of distribution networks based on digital twins, the method comprising: Acquire multi-source heterogeneous data of the distribution network and perform fusion preprocessing on the multi-source heterogeneous data; construct a digital twin model of the distribution network based on the fusion preprocessed data; perform anomaly diagnosis on the distribution network based on the digital twin model of the distribution network; perform situation prediction in the digital twin model of the distribution network based on the distribution network anomalies to obtain weak nodes; perform sand table simulation verification based on the weak nodes of the distribution network to confirm executable strategies; distribute the executable strategies to the distribution network for execution, and evolve and correct the digital twin model of the distribution network based on the actual operating status.

[0006] Another aspect of this application discloses a digital twin-based intelligent dispatching system for distribution networks, which includes: The system comprises the following modules: a fusion module for acquiring multi-source heterogeneous data from the distribution network and performing fusion preprocessing on the data; a construction module for constructing a digital twin model of the distribution network based on the fused and preprocessed data; a diagnosis module for performing anomaly diagnosis on the distribution network based on the digital twin model; a prediction module for performing situation prediction on the digital twin model based on distribution network anomalies to identify weak nodes; a simulation module for performing sand table simulation verification based on the weak nodes to confirm executable strategies; and an execution module for distributing the executable strategies to the distribution network for execution and evolving and correcting the digital twin model based on the actual operating status.

[0007] One or more technical solutions provided in this application have at least the following technical effects or advantages: By fusing and preprocessing multi-source heterogeneous data from the distribution network to build a standardized data foundation, a dynamic digital twin model that is virtually and physically bound to the physical power grid is constructed. Based on the twin model, full-condition anomaly diagnosis, situation prediction and location of weak nodes are carried out sequentially, and sand table simulation is used to verify and screen executable scheduling strategies. In addition, the twin model is modified and corrected in a closed loop based on actual operating data. This effectively solves the above-mentioned technical problems and achieves the technical effect of improving the efficiency of scheduling decisions and the economic efficiency of power grid operation while ensuring the safe operation of the distribution network.

[0008] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, specific embodiments of this application are given below. Attached Figure Description

[0009] Figure 1 A flowchart illustrating a digital twin-based intelligent dispatching method for power distribution networks is provided for embodiments of this application. Figure 2 A schematic diagram of the structure of a distribution network intelligent dispatching system based on digital twins is provided for the embodiments of this application.

[0010] Figure labeling: Fusion module 11, Construction module 12, Diagnosis module 13, Prediction module 14, Inference module 15, Execution module 16. Detailed Implementation

[0011] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structure, features and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0012] The overall concept of the technical solution provided in this application is as follows: This application provides a method and system for intelligent dispatching of distribution networks based on digital twins. The method involves acquiring multi-source heterogeneous data from the distribution network and performing fusion preprocessing on this data; constructing a digital twin model of the distribution network based on the fused preprocessed data; performing anomaly diagnosis on the distribution network based on the digital twin model; predicting the situation in the digital twin model based on distribution network anomalies to identify weak nodes; conducting sand table simulations to verify the weak nodes and confirm executable strategies; distributing the executable strategies to the distribution network for execution; and evolving and correcting the digital twin model based on actual operating conditions. This approach aims to improve dispatching decision-making efficiency and grid operation economy while ensuring the safe operation of the distribution network.

[0013] After introducing the basic principles of this application, various non-limiting embodiments of this application will be described in detail below with reference to the accompanying drawings.

[0014] Example 1, as Figure 1 As shown in the figure, this application provides a distribution network intelligent dispatching method based on digital twins, the method including: S100: Acquire multi-source heterogeneous data of the power distribution network and perform fusion preprocessing on the multi-source heterogeneous data.

[0015] Specifically, the first step is to conduct full-domain data collection across the entire distribution network scenario, simultaneously acquiring operational measurement data records and external environmental data records. The operational measurement data records cover real-time and historical electrical operation data collected by field devices such as SCADA systems, feeder terminals, distribution transformer terminals, and smart meters, including node voltage, branch current, active and reactive power, switch on / off status, distributed power output, and load power. The external environmental data records include non-electrical data such as meteorological temperature and humidity, wind speed, rainfall, geospatial coordinates, equipment ledger parameters, and user electricity consumption behavior. The above-mentioned data, which are scattered from different sources and have significantly different forms, together constitute multi-source heterogeneous data. Multi-source means that the data comes from multiple independent business systems and acquisition devices such as measurement terminals, meteorological systems, geographic information systems, and marketing systems, and the subjects and channels that generate the data are different. Heterogeneous means that the data have significant differences in storage format, sampling frequency, time base, and spatial coordinate system. It includes both structured time-series numerical data and semi-structured communication messages and unstructured equipment ledger information, which cannot be directly used for the construction and simulation calculation of digital twin models.

[0016] After completing the full-domain data collection, the first step is to perform spatiotemporal alignment and format unification processing. Spatiotemporal alignment addresses the issues of inconsistent sampling frequencies and timestamp discrepancies among different devices by using linear interpolation and resampling to unify all time-series data onto the same time scale. This ensures that electrical and environmental quantities at the same moment have consistent time labels. Simultaneously, network topology and device location data from different systems using different spatial coordinate systems are matched to a unified spatial system through geographic information coordinate mapping. This achieves a one-to-one correspondence between the spatial locations of electrical nodes and physical devices, eliminating spatiotemporal misalignment issues. Format unification involves converting heterogeneous data such as binary protocol messages, CSV tables, and JSON format messages output by different systems into a standardized structured dataset that conforms to the input specifications of the digital twin model through communication protocol parsing, field standardization mapping, and unstructured data structure conversion. This completely breaks down data format barriers between different systems.

[0017] Further bad data identification and removal are carried out. This involves identifying and processing abnormal and invalid data caused by electromagnetic interference, communication link interruption, and data acquisition terminal failure in the distribution network through a multi-dimensional verification mechanism. Specifically, threshold verification is used to compare the rated operating parameters of the equipment to determine whether the values ​​have exceeded the limit. Topology logic verification is used to verify the logical consistency between switch status and power flow distribution. Temporal continuity verification is used to screen out abrupt data that does not conform to the operating rules. Finally, state estimation residual verification is combined to complete the final confirmation. Identified bad data is directly removed. At the same time, missing data of a small number of key nodes is filled by neighbor interpolation and fitting of historical data from the same period. Finally, spatiotemporally consistent, complete and accurate fused preprocessed data is output.

[0018] S200: Construct a digital twin model of the distribution network based on the preprocessed data.

[0019] Specifically, firstly, relying on the pre-processed equipment ledger, network topology, geospatial information, and historical operation datasets, a full-element virtual entity system of the distribution network is built. First, the spatial location and electrical connection structure of all physical entities, such as distribution lines, distribution transformers, sectionalizing switches, distributed power sources, and user loads, are accurately restored according to the actual wiring relationships of the physical distribution network. Then, corresponding electrical characteristic parameters and simulation calculation logic are matched to each virtual entity, covering line impedance parameters, transformer turns ratio and loss characteristics, switch opening and closing logic, distributed power source output response characteristics, and load time-varying fluctuation patterns. Simultaneously, a basic algorithm engine for power flow calculation and state solving is embedded, thus forming an initial digital twin model. This initial digital twin model is a virtual mirror base of the distribution network built solely based on static parameters and historical data, and has not yet formed a dynamic linkage with the real-time operation status of the physical distribution network, possessing only offline structural representation and basic simulation calculation capabilities.

[0020] After the initial model is built, a two-way data interaction link between the physical and digital twins is established using a standardized communication protocol adapted to the power industry scenario. This communication protocol is a unified standard for data transmission between the physical distribution network field acquisition terminal and the digital twin platform. It typically adopts adaptation schemes such as IEC61850 power communication protocol and MQTT IoT transmission protocol to ensure low-latency, high-reliability real-time transmission of multi-source measurement data, while supporting the subsequent issuance and transmission of downlink dispatch commands. Based on this stable data link, the measurement data transmitted back from the physical distribution network in real time is bound and mapped to the corresponding virtual entities in the initial digital twin model. The binding and mapping refers to establishing a unique identification correspondence between each physical device and the virtual object in the twin model. The operating data such as node voltage, branch current, active and reactive power, and switch status collected in real time on the physical side are synchronously assigned to the operating attributes of the corresponding entities in the virtual model, driving the operating status of the virtual model to be updated synchronously with the actual operation of the physical distribution network. This achieves precise synchronization between the operating status of the physical entities and the status of the virtual model. After the binding and mapping is completed, the initial model is upgraded to a digital twin model of the distribution network that can dynamically reproduce the operation of the physical power grid and support online simulation and deduction.

[0021] S300: Perform anomaly diagnosis on the distribution network based on the aforementioned digital twin model of the distribution network.

[0022] Specifically, the measurement data transmitted back in real time from the physical distribution network is first synchronously input into the digital twin model of the distribution network. State estimation calculations are then performed in the built-in electrical calculation engine of the model. State estimation refers to using the accurate network topology and rated electrical parameters of the equipment built into the digital twin model of the distribution network as the calculation benchmark, combined with the measurement data collected in real time from multiple nodes, and using statistical algorithms such as weighted least squares to correct and complete the original measurement data that has acquisition noise and missing local points. This results in a complete, consistent, and high-precision operating profile covering all nodes and branches of the entire network, obtaining the global operating status including the voltage amplitude and phase angle of each node, the active and reactive power of each branch, the load rate of lines and transformers, and the status of the switch topology.

[0023] Subsequently, based on the acquisition of the global operating status, the operating safety boundary indicators are continuously monitored and compared in real time. The operating safety boundary indicators refer to a set of critical thresholds set according to the rated parameters of the distribution network equipment, national standards for power supply quality, and safe operation procedures. These indicators cover the upper and lower limits of allowable deviation of node voltage, the maximum load rate limit of lines and transformers, the allowable threshold of three-phase imbalance, and the power quality indicators of distributed power supply grid connection points. They serve as a quantitative benchmark for determining whether the operating status of the distribution network is within a safe and controllable range. During the diagnosis process, the various operating parameters output by the state estimation are matched and verified one by one with the corresponding safety boundary indicators. At the same time, the topological correlation of the digital twin model is combined to perform power flow transmission analysis. This not only identifies obvious anomalies where parameters directly exceed the limits, but also identifies hidden anomalies close to the safety boundary and anomalies with topological logical contradictions through trend inference. Finally, the accurate location and type determination of distribution network anomalies are completed. S400: Based on distribution network anomalies, situation prediction is performed in the digital twin model of the distribution network to obtain nodes with weak resilience.

[0024] Specifically, based on the identified distribution network anomaly types, locations, and current severity, and combined with the load time-series change patterns in the fused preprocessed data and the distributed power output fluctuation trends derived from meteorological forecasts, situation prediction is carried out in the electrical simulation engine built into the distribution network digital twin model. Situation prediction refers to the dynamic calculation process of extrapolating the distribution network operation status for multiple consecutive periods in the future along the time axis, starting from the current real-time operating section as the initial state. Unlike traditional static safety verification, it can completely reproduce the transmission process and diffusion trend of abnormal states as load and power output change.

[0025] During the simulation, the model will solve the power flow distribution of the distribution network at each time period. The power flow distribution is a set of core electrical characteristics that characterize the operating conditions of the distribution network. It covers parameters such as the voltage amplitude and phase angle of all nodes in the network, the active and reactive power of all branches, and the equipment load rate. The simulation process will continuously superimpose the cumulative effect of the current anomaly. For example, the heavy load anomaly of a certain branch will continue to worsen as the load rises during the evening peak, and will raise the load level of adjacent branches through the network topology association, thus fully restoring the time path and spatial impact range of the anomaly evolution.

[0026] After obtaining the power flow distribution projection results for the entire future time period, preset fault scenarios are injected into the digital twin model in batches. The preset fault scenarios are a set of typical faults pre-set by combining the historical fault statistics of the distribution network, equipment risk level and external environmental risk. They cover single-point faults and cascading fault types such as single line disconnection, distribution transformer overload tripping, distributed power generation grid connection point disconnection, and multi-equipment shutdown caused by extreme weather. Various faults are applied to the power flow distribution sections at different time periods to simulate the dynamic response process of the distribution network after the fault is triggered, including the voltage drop range, the scale of power loss load, and the power supply restoration path after topology reconfiguration.

[0027] Finally, based on the simulation results of multiple scenarios, a quantitative assessment of resilience was carried out to screen out the weak nodes. The weak nodes refer to the key nodes or branches that are the first to break through the operational safety boundary, have the weakest power supply recovery capability after the fault, and have the highest impact on the reliability of the entire power supply network under the dual effects of abnormal evolution and fault impact. The assessment dimensions cover multiple indicators such as the proportion of power loss load after the fault, the voltage over-limit amplitude, the fault range, and the power supply recovery time. The final location was completed by weighted ranking of multiple scenarios.

[0028] S500: Based on the aforementioned weak nodes, a sand table simulation is performed to verify and confirm the executable strategy.

[0029] Specifically, the optimization objectives are multi-dimensional, including operational safety, operational economy, and renewable energy consumption. Based on the risk causes and impact range of weak nodes, a set of candidate scheduling strategies is generated by matching the adjustable resources of the distribution network. The set of candidate scheduling strategies is a collection of various scheduling adjustment schemes generated in a targeted manner for the risks of weak nodes. It covers a variety of control measures such as distribution network reconfiguration, load transfer path adjustment, active and reactive power output optimization of distributed power sources, switching of reactive power compensation devices, and demand-side response load regulation. Different strategies correspond to different resource combinations and adjustment intensities, which can cover a variety of risk response paths.

[0030] Subsequently, the candidate scheduling strategy set is input one by one into the digital twin model of the distribution network for sand table simulation verification. Sand table simulation verification refers to reproducing the complete process of strategy execution in a purely virtual simulation environment based on the topology, electrical parameters and operating characteristics of the digital twin model that are completely consistent with the physical power grid. There is no need to carry out trial operation in the real physical power grid. It can verify the effect of the strategy and identify potential risks without loss. During the simulation, the current real-time operating section and the situation prediction data of the future period are loaded as the initial operating conditions. Then, the switching actions, output regulation, load adjustment and other operations corresponding to the strategy are simulated. At the same time, the full-dimensional operation indicators such as distribution network power flow redistribution, node voltage change, equipment load rate fluctuation and renewable energy consumption level after the strategy is implemented are calculated.

[0031] After the simulation is completed, the simulation results are verified against the preset safety constraints item by item. The preset safety constraints are hard thresholds set in advance based on the distribution network safety operation regulations, equipment rated parameters, and power supply quality standards. They cover multiple aspects such as the allowable deviation range of node voltage, the maximum load rate limit of lines and transformers, the three-phase imbalance requirement, the relay protection coordination condition, and the power supply reliability index. If any indicator in the virtual simulation result of a candidate strategy exceeds the safety constraints, such as causing secondary risks such as overload of adjacent branches or voltage exceeding the limit of the terminal node after the strategy is implemented, the candidate strategy is directly eliminated. The strategy parameters are adjusted or a new control scheme is added based on the risk points, and then the simulation is re-introduced, forming an iterative process of generation-simulation-verification-optimization. If the virtual simulation results of the candidate strategy all meet the preset safety constraints and achieve the expected results in terms of operational economy and renewable energy consumption level, it is confirmed as an executable strategy. If multiple strategies meet the constraints, the scheme with the best comprehensive benefits is selected as the final executable strategy through multi-objective weighted scoring.

[0032] S600: The executable strategy is sent to the distribution network for execution, and the digital twin model of the distribution network is evolved and corrected based on the actual operating status.

[0033] Specifically, the executable strategy, which has undergone full-dimensional security verification, is first decomposed according to the instruction specifications of the field equipment. Through a standardized communication protocol link that is bound and mapped to the physical distribution network, the decomposed control instructions are accurately sent to various controllable execution devices in the distribution network, such as feeder automation terminals, distributed power control systems, reactive power compensation devices, and demand-side response terminals. The devices are driven to complete corresponding operations such as switch opening and closing, output adjustment, and load regulation in sequence, so as to realize the implementation of the scheduling strategy in the physical distribution network.

[0034] Throughout the entire strategy execution process, the field measurement terminal continuously collects the actual operating status of the distribution network after the strategy is implemented, including full-scale operating data such as the voltage amplitude of all network nodes, active and reactive power of branches, real-time load rate of equipment, and switch topology location. The data is then synchronously transmitted back to the distribution network digital twin model. Within the model, the real-time collected actual operating status is compared and verified point by point with the virtual simulation results generated in the previous sand table simulation phase. The virtual simulation results here refer to the set of expected operating indicators simulated and output by the executable strategy in the twin model during the sand table verification phase. These indicators cover all dimensions of expected data such as power flow distribution, voltage level, and load status after the strategy is implemented, and serve as a benchmark for measuring the effectiveness of strategy execution and the accuracy of model simulation.

[0035] Subsequently, the comparison process uses a pre-set threshold as the quantitative standard for deviation judgment. The pre-set threshold is a deviation tolerance range defined by combining the safe operation regulations of the distribution network, the safety margin of the equipment, and the simulation accuracy requirements. It includes several sub-indicators such as node voltage deviation rate, branch power calculation error, and equipment load rate deviation amplitude. If the difference between the actual operating state and the virtual simulation result is all within the pre-set threshold range, it indicates that the strategy execution effect is consistent with the simulation expectation and the operating state is safe and controllable. At this time, the system will include the actual operating data into the model's iterative dataset and fine-tune and optimize the simulation parameters such as equipment electrical parameters, load time-varying characteristics, and equipment response delay in the digital twin model. This completes the normalized evolution correction of the model. Evolution correction refers to the digital twin model relying on continuously accessed real operating data to dynamically iterate its own simulation calculation logic and entity parameters, gradually eliminating the inherent errors between the model's static parameters and the real characteristics of the physical power grid. This achieves a dynamic process in which the model accuracy continuously improves with the accumulation of operating data and evolves synchronously with the characteristics of the physical power grid.

[0036] If the comparison reveals that the difference between the actual operating state and the virtual simulation results exceeds a preset threshold, such as due to uncontrollable factors like sudden load fluctuations, equipment response delays, or sudden changes in the external environment causing significant deviations in power flow distribution and voltage levels from simulation expectations, an emergency correction mechanism is immediately triggered. This emergency correction mechanism is a rapid response logic for operational deviation scenarios. It uses the current actual operating section as the initial condition and quickly conducts secondary simulation and safety verification in the digital twin model, generating targeted compensatory correction strategies, such as supplementing reactive power compensation device switching, fine-tuning distributed power output, and optimizing load transfer paths. After rapid safety verification, the correction strategies are immediately sent to the physical distribution network for execution, quickly pulling the operating state back to a safe and controllable range. At the same time, the system will conduct in-depth analysis of the causes of the deviation based on the deviation data and make targeted corrections to the corresponding equipment response parameters and environmental impact weights in the twin model, completing precise iterative optimization of the model.

[0037] Furthermore, in the method provided in this application embodiment, multi-source heterogeneous data of the distribution network is acquired, and the multi-source heterogeneous data is fused and preprocessed, including: collecting operation measurement data records of the distribution network and external environment data records, performing spatiotemporal alignment and format unification, and identifying and removing bad data to complete the fusion preprocessing.

[0038] Specifically, firstly, various data acquisition devices deployed at the distribution network site, such as SCADA systems, feeder terminals, distribution transformer monitoring terminals, and smart meters, continuously collect operational measurement data records of the distribution network. This includes comprehensive electrical operation sequence data covering all dimensions such as voltage amplitude at each node, active and reactive power of branches, distribution transformer load rate, switch on / off status, real-time output of distributed power sources, and user load power. Simultaneously, it connects to external business platforms such as meteorological service systems, geographic information systems, equipment ledger management systems, and electricity marketing systems to obtain external environmental data records, including meteorological parameters such as temperature, wind speed, and rainfall, as well as equipment spatial coordinates and network... Geographic topology information such as wiring relationships, equipment rated parameters, commissioning years and other ledger data, and non-electrical data such as user electricity consumption patterns—these scattered and diverse data constitute multi-source heterogeneous data. Multi-source refers to data generated from different entities and channels, such as field acquisition terminals and multiple independent business systems, with different data ownership and acquisition logic. Heterogeneous refers to data with significant differences in sampling frequency, time base, spatial coordinate system, and storage format. It includes both structured time-series numerical data and semi-structured communication messages and unstructured equipment ledger documents, which cannot be directly input into the digital twin model for calculation and simulation.

[0039] Following this, spatiotemporal alignment and format unification were performed. Spatiotemporal alignment consisted of two parallel processes: time dimension alignment and spatial dimension alignment. Time dimension alignment addressed the issues of inconsistent sampling periods and timestamp references among different devices. Using a unified standard time axis as a reference, data with different sampling frequencies (seconds, minutes, hours, etc.) were unified in granularity through linear interpolation, resampling, and other methods. This ensured that electrical operation data and environmental data at the same moment had completely consistent time labels, eliminating data time misalignment. Spatial dimension alignment involved aligning the latitude and longitude spatial coordinates in the geographic information system and the logical node numbers in the power grid topology. The asset numbers in the equipment ledger are mapped and bound one-to-one, giving each physical device a unified spatial and identity identifier. This ensures that the electrical topology, equipment location, and attribute data correspond accurately in the spatial dimension. Format unification is achieved through operations such as communication protocol parsing, field standardization mapping, and unstructured data structure conversion. This transforms the original data, which contains different formats such as power communication protocol binary messages, CSV tables, JSON interface messages, and unstructured ledger text, into a standardized structured dataset that conforms to the input specifications of the digital twin model. This breaks down data format barriers between different systems and achieves data form unification.

[0040] Based on this, further bad data identification and removal operations are performed. For invalid and abnormal data such as numerical jumps, missing data, and topological contradictions caused by factors such as electromagnetic interference, communication link interruption, and acquisition terminal failure in the distribution network, a multi-dimensional joint verification mechanism is established to carry out accurate identification. First, the collected values ​​are compared with the rated parameters of the equipment and historical operating extreme values ​​through threshold verification to initially screen out abnormal data that jumps out of the limit. Then, the consistency between the switch status and the power flow distribution is verified through topological logic verification to identify contradictory data that does not conform to the topological logic of the power grid.

[0041] Next, transient abrupt changes that do not conform to the load fluctuation pattern are screened out through temporal continuity verification. Finally, the final abnormal data confirmation is completed by combining state estimation residual verification. Identified invalid and bad data are directly removed. At the same time, missing data of a small number of key nodes are reasonably supplemented by interpolation of nearby measurement points and fitting of historical data from the same period to avoid the impact of missing key data on subsequent calculations. After the above full process, the fusion preprocessing is completed, which is to transform the originally scattered, isolated, inconsistent, and error-prone multi-source heterogeneous data into a unified dataset that is consistent in time and space, has a standard format, and is of reliable quality. It can directly provide qualified input data for the subsequent construction of digital twin models.

[0042] Furthermore, in the method provided in this application embodiment, constructing a digital twin model of the distribution network based on the fused preprocessed data includes: constructing an initial digital twin model of the distribution network based on the fused preprocessed data; and binding and mapping the physical distribution network measurement data with the initial digital twin model through a communication protocol to obtain the digital twin model of the distribution network.

[0043] Specifically, firstly, relying on the pre-processed equipment ledger parameters, network topology relationships, geospatial coordinates, historical operating sequence data, and equipment factory characteristic parameters, a layered virtual mirror framework of all elements of the distribution network is constructed. First, a topology layer is built, strictly replicating the spatial location and electrical connection logic of all physical entities such as distribution lines, distribution transformers, sectionalizing switches, distributed power generation grid connection points, and user load nodes in a 1:1 ratio according to the actual wiring relationships and spatial layout of the physical distribution network. Then, an equipment parameter layer is built, matching corresponding rated electrical parameters and operating characteristics for each virtual entity, covering line impedance and ground capacitance parameters, transformer turns ratio and no-load load loss characteristics, switch opening and closing action delay, distributed power generation output response curve, and load time-varying fluctuation law, etc. Subsequently, a simulation calculation layer with built-in power flow calculation and state solution algorithms is embedded, and a behavioral rule layer such as equipment action rules and safety boundary constraints is superimposed to form a complete model infrastructure.

[0044] Based on this, offline calibration training is carried out using historical operating data. Multiple sets of typical historical operating sections are input into the model framework for simulation calculation. The output power flow distribution, node voltage, and other results are compared with the actual historical operating data of the corresponding time period. Through parameter iterative correction algorithms, the equivalent parameters of the equipment and the load model coefficients are gradually adjusted to continuously reduce the deviation between the simulation results and the real operating data until the offline simulation accuracy of the model reaches the preset standard. This forms the initial digital twin model. The initial digital twin model is a virtual mirror base of the distribution network built only based on static parameters and historical data. It has not yet established a dynamic linkage link with the real-time operating status of the physical distribution network. It only has offline structural representation and basic simulation calculation capabilities and cannot update the status synchronously with the operation changes of the physical power grid in real time.

[0045] After the initial model is built, a virtual-physical two-way data interaction channel is established based on a standardized communication protocol adapted to the power industry scenario. The communication protocol is a unified standard for realizing two-way data transmission between the physical distribution network field acquisition terminal, measurement and control device and twin model platform. The distribution network scenario usually uses the IEC61850 power communication protocol to realize data interaction between equipment in the station. Combined with the MQTT Internet of Things transmission protocol to adapt to the widely distributed distribution network terminal equipment, it can ensure low latency and high reliability real-time transmission of multi-source measurement data, while supporting the downlink transmission of subsequent dispatch instructions.

[0046] Subsequently, relying on a stable communication link, binding and mapping operations are carried out. First, a unique identifier is established for each physical device in the physical distribution network and its corresponding virtual entity in the initial model. By binding the device asset number, terminal communication address and virtual object ID one by one, the precise targeting of data transmission is ensured, avoiding data mismatch. After the identifier binding is completed, all the measurement data such as node voltage, branch active and reactive power, switch on / off status and equipment load rate transmitted back in real time from the physical distribution network are synchronously assigned to the running attribute port of the corresponding entity in the virtual model according to a unified timestamp. This drives the running status of the virtual entity to be updated synchronously with the actual operation of the physical device. At the same time, a synchronization verification mechanism is established to ensure that the time axis of the measurement data is precisely aligned with the model simulation step size, avoiding state deviation caused by timing misalignment. After continuous data synchronization and state linkage calibration, the initial model is upgraded into a distribution network digital twin model that can reproduce the operating status of the physical power grid in real time and supports online simulation and deduction.

[0047] Furthermore, in the method provided in this application embodiment, the abnormal diagnosis of the distribution network based on the distribution network digital twin model includes: performing state estimation in the distribution network digital twin model based on real-time measurement data to obtain the operating state; monitoring the operating safety boundary indicators based on the operating state, and identifying the distribution network abnormality in the distribution network digital twin model.

[0048] Specifically, the first step is to synchronously inject the multi-node measurement data transmitted in real time from the physical distribution network into the digital twin model of the distribution network. State estimation calculations are then performed within the model's built-in electrical calculation engine. State estimation is a core calculation method that optimally approximates the actual operating state of the power grid based on the distribution network topology constraints and redundant measurement data. Because the distribution network's on-site measurement points are unevenly distributed, and the original collected data generally contains random noise and minor errors, it is impossible to directly stitch them together to form a complete and reliable global operating profile. Therefore, the calculation process uses the precise network topology structure built into the digital twin model. Using the rated electrical parameters as a fixed calculation benchmark, a weighted least squares statistical algorithm is employed to correct and complete the input real-time measurement data. This not only filters out random interference errors remaining in the acquisition link but also reasonably extrapolates and completes the parameters of terminal nodes and branches with insufficient measurement coverage. Ultimately, a complete and high-precision operating profile covering all nodes and branches of the entire network is obtained, which includes the voltage amplitude and phase angle of each node, the active and reactive power of each branch, the real-time load rate of lines and transformers, and the topological connectivity status of the entire network switches, thus fully restoring the current real operating conditions of the distribution network.

[0049] Based on the global operational status, the system continuously monitors and compares operational safety boundary indicators in real time. These indicators are a set of critical thresholds pre-set according to the rated parameters of distribution network equipment, national power supply quality standards, and power safety operation regulations. They cover multiple aspects, including the upper and lower limits of allowable deviations of node voltage, the maximum load rate limits of lines and distribution transformers, the allowable threshold of three-phase imbalance, and power quality indicators of distributed power grid connection points. They serve as a rigid benchmark for quantitatively determining whether the distribution network is in a safe and controllable range. During the monitoring process, the various operational parameters output by the state estimation are matched and verified one by one with the corresponding safety boundary indicators. At the same time, the system conducts power flow transmission analysis by combining the topological correlation of the digital twin model. This not only directly identifies explicit distribution network anomalies where operational parameters exceed safety thresholds, such as branch overload, node voltage exceeding limits, and excessive three-phase imbalance, but also identifies latent anomalies that are continuously approaching the safety boundary through parameter change trends, as well as topological anomalies where switch status and power flow distribution logic contradict each other. Furthermore, relying on the topological positioning capability of the twin model, the system accurately locates the specific location, scope of impact, and severity level of anomalies, thus completing the accurate identification of all types of distribution network anomalies.

[0050] Furthermore, in the method provided in this application embodiment, based on the distribution network anomaly, situation prediction is performed in the distribution network digital twin model to obtain the weak nodes in resilience, including: based on the distribution network anomaly, the distribution network power flow distribution in the distribution network digital twin model for future periods is deduced; based on the distribution network power flow distribution, a preset fault scenario is simulated in the distribution network digital twin model to evaluate the weak nodes in resilience of the distribution network.

[0051] Specifically, firstly, by combining the load time-series fluctuation patterns, distributed power output change trends derived from meteorological forecasts, and user electricity consumption behavior characteristics contained in the fusion preprocessed data, and overlaying the location, fault type, and current severity of currently identified anomalies, the distribution network operation status for multiple consecutive time periods is simulated forward along the time axis in the electrical simulation engine built into the distribution network digital twin model. The power flow distribution of the distribution network at each time step is solved. The power flow distribution is a core set of electrical characteristics characterizing the operating conditions of all nodes and branches of the distribution network, encompassing the voltage amplitude of all nodes in the entire network. Complete parameters such as phase angle, active and reactive power of each distribution line and transformer branch, real-time load rate of equipment, and power flow direction are the core basis for judging the operating status and risk level of the distribution network. The cumulative evolution effect of the current anomaly will be continuously incorporated during the simulation process. For example, the heavy load anomaly of a certain branch will gradually increase the load rate as the evening peak load rises, and will be transmitted to adjacent branches through the network topology, simultaneously increasing the power supply pressure of surrounding nodes. The simulation fully reproduces the transmission path, diffusion range and aggravation degree of the abnormal state over time, and finally outputs the continuous power flow distribution simulation results for the future all time period.

[0052] Subsequently, pre-set fault scenarios are injected into the digital twin model of the distribution network section by section to conduct impact simulation. The pre-set fault scenarios are typical fault sets pre-constructed by combining the historical fault statistics of the distribution network, equipment health status rating, and external environmental risk level. They include single-point fault types such as single line disconnection, single distribution transformer tripping, and distributed power grid connection point disconnection, as well as compound fault types such as multi-equipment cascading outages caused by extreme weather and protection maloperation after a fault. During the simulation, various faults are applied to the initial sections of the power flow distribution at different time periods one by one. The dynamic response process of the distribution network after the fault is triggered is simulated according to the actual relay protection action logic and switch action sequence. The full-dimensional results such as voltage drop range, power loss load scale, power flow transfer path, and equipment overload cascading risk after the fault occur are fully calculated. The actual impact of different faults under different operating conditions can be reproduced without any intervention in the physical distribution network.

[0053] After completing the full-scenario fault simulation, the quantitative assessment and screening of weak nodes in the distribution network are carried out based on the simulation results of multiple scenarios and time periods. Weak nodes refer to key nodes or branches that have the lowest operational safety margin, the weakest fault tolerance, and the greatest impact on the reliability of the entire power supply after a fault occurs under the dual effects of abnormal evolution and fault impact. The assessment process constructs a quantitative indicator system from three core dimensions: fault tolerance, fault impact range, and power supply recovery difficulty. Specifically, it includes sub-indicators such as the magnitude of node voltage exceeding the limit, the duration of equipment overload, the proportion of power loss load affected by the fault, and the time for power transfer and recovery after the fault. All nodes and branches in the entire network are ranked by weighted scoring of multiple indicators. At the same time, the location with the most significant risk superposition effect is further identified by combining the transmission path of abnormal evolution, and finally the weak nodes in the distribution network are accurately located.

[0054] Furthermore, in the method provided in this application embodiment, the executable strategy is confirmed by performing a sand table simulation based on the weakly resilient node, including: generating a set of candidate scheduling strategies based on the weakly resilient node with the objectives of operational safety, operational economy, and renewable energy consumption; inputting the set of candidate scheduling strategies into the digital twin model of the distribution network for sand table simulation verification; if the virtual simulation result does not meet the preset safety constraints, the candidate strategy is eliminated and re-verified; if the virtual simulation result meets the preset safety constraints, it is confirmed as the executable strategy.

[0055] Specifically, the optimization objectives are multi-dimensional, including operational safety, operational economy, and renewable energy consumption. Based on the risk causes, impact scope, and current operational status of vulnerable nodes, a candidate scheduling strategy set is generated by coordinating all adjustable resources within the distribution network. This set consists of multiple differentiated scheduling adjustment schemes designed specifically for the risks of vulnerable nodes. The control measures cover multiple dimensions, including distribution network reconfiguration and load transfer path adjustment, active and reactive power output optimization of distributed generation, reactive power compensation device switching and capacity regulation, and precise load control in demand-side response. Each strategy corresponds to different resource combinations and parameter adjustment levels, covering various risk response paths from mild intervention to deep adjustment. Simultaneously, each strategy is initially matched with corresponding economic calculations and absorption capacity estimates, providing a multi-dimensional alternative basis for subsequent simulation and verification.

[0056] Subsequently, each scheme in the candidate scheduling strategy set is input into the distribution network digital twin model one by one for sand table simulation verification. Sand table simulation verification is a verification method that fully reproduces the entire process of strategy execution in a purely virtual simulation environment, based on the topology, electrical parameters and dynamic response characteristics of the digital twin model and the physical power grid. The entire process does not require trial operation in the real physical power grid, and can verify the actual effect of the strategy and identify potential secondary risks without loss. During the simulation, the current real-time operating section and the situation prediction data of the corresponding time period are loaded as the initial operating conditions to accurately restore the real operating environment of the weak nodes.

[0057] Subsequently, according to the action sequence set by the strategy, various control operations such as switch opening and closing, output adjustment, and load adjustment are simulated. Simultaneously, the power flow redistribution, node voltage changes, equipment load rate fluctuations, renewable energy consumption ratio, and network loss level of the distribution network after the strategy is executed are solved step by step over time. The entire dynamic process of the strategy from startup to steady state is fully reproduced, and complete virtual simulation results are output.

[0058] After the simulation is completed, the virtual simulation results of each strategy are rigidly verified against the preset safety constraints. The preset safety constraints are inviolable hard thresholds set in advance based on the distribution network safety operation regulations, equipment rated parameters, and national power supply quality standards. They cover multiple core indicators such as the allowable deviation range of node voltage, the maximum load rate limit of lines and distribution transformers, the allowable value of three-phase imbalance, the coordination conditions of relay protection actions, and the power supply reliability requirements. If any indicator in the virtual simulation result of a candidate strategy exceeds the preset safety constraints, such as causing secondary operational risks such as overload of adjacent branches, voltage exceeding the limit of terminal nodes, or timing conflicts of protection coordination after the strategy is executed, the unqualified candidate strategy will be directly eliminated.

[0059] Simultaneously, based on the risk points exposed in this simulation, targeted strategy optimizations are carried out. For example, the power transfer path of network reconstruction is adjusted to distribute load pressure, the output adjustment range of distributed power sources is corrected to match voltage tolerance, and the matching switching action of reactive power compensation devices is supplemented to improve voltage quality. New candidate strategies after adjustment are generated and re-substituted into the digital twin model for simulation verification, forming an iterative process of generation-simulation-verification-optimization-re-simulation until the adjusted strategy meets all safety constraints. If the virtual simulation results of the candidate strategies all meet the preset safety constraints and achieve the expected standards in terms of operational economy, renewable energy consumption level, and other optimization objectives, they are included in the qualified strategy pool. If multiple strategies meet the safety constraints, a multi-objective weighted scoring algorithm is used to comprehensively rank indicators such as safety margin, network loss level, renewable energy consumption rate, and control cost, and the scheme with the best comprehensive benefits is selected as the final executable strategy.

[0060] Furthermore, in the method provided in this application embodiment, the executable strategy is sent to the distribution network for execution, and the distribution network digital twin model is evolved and corrected based on the actual operating state. This includes: collecting the actual operating state of the distribution network after executing the executable strategy, and comparing the actual operating state with the virtual simulation result in the distribution network digital twin model; if the difference between the actual operating state and the virtual simulation result exceeds a preset threshold, an emergency correction mechanism is triggered, and a correction strategy is generated in the distribution network digital twin model and sent for execution.

[0061] Specifically, the executable strategy, which has undergone full-dimensional security verification, is first decomposed according to the instruction specifications of the on-site monitoring and control equipment. The macro-scheduling requirements in the strategy, such as network reconstruction, output adjustment, and load control, are decomposed into specific control instructions that can be executed by individual devices, such as switch opening and closing, output parameter adjustment, and compensation device switching. Then, through the previously established virtual and physical two-way communication protocol link, the control instructions are accurately sent to various controllable execution devices in the distribution network, such as feeder automation terminals, distributed power control systems, reactive power compensation devices, and demand-side response terminals. The devices are driven to complete the corresponding operations in sequence according to the preset action sequence, so as to realize the smooth implementation of the scheduling strategy in the physical distribution network.

[0062] Subsequently, throughout the entire process of strategy execution and the steady-state operation phase, the measurement terminals deployed on-site will continuously collect the actual operating status of the distribution network after the execution of the executable strategy. This includes full-volume operating data such as voltage amplitude of each node in the entire network, active and reactive power of each branch, real-time load rate of equipment, switch topology location, and actual output of distributed power sources. The data will be synchronously transmitted back to the digital twin model of the distribution network with a unified timestamp, providing real data support for virtual-real comparison. The transmitted actual operating data will be compared and verified point by point with the virtual simulation results of the corresponding time period in the twin model. The virtual simulation results refer to the set of expected operating indicators simulated and output in the digital twin model during the sand table simulation verification phase of the executable strategy. These indicators include full-dimensional expected data such as power flow distribution, voltage level, load status, and renewable energy absorption rate after strategy execution. They are the core benchmark for measuring the effectiveness of strategy execution and the accuracy of model simulation.

[0063] Subsequently, the comparison process uses a pre-set threshold as a quantitative benchmark for deviation judgment. The pre-set threshold is a deviation tolerance range defined by combining the safe operation procedures of the distribution network, the safety margin of the equipment and the simulation accuracy requirements of the twin model. It includes several sub-indicators such as node voltage deviation rate, branch power calculation error, equipment load rate deviation amplitude, and new energy output deviation value, which are used to distinguish between simulation errors within the normal range and significant deviations that require intervention.

[0064] If the difference between the actual operating state and the virtual simulation result is within the preset threshold range, it indicates that the strategy execution effect is consistent with the simulation expectation and the power grid operation state is safe and controllable. At this time, the system will include the actual operating data of the entire process of this strategy execution into the model's iterative training dataset, and fine-tune and optimize the simulation parameters such as the equivalent electrical parameters of the equipment, the time-varying characteristics of the load, and the response delay of the equipment action in the digital twin model, so as to reduce the inherent deviation between the model simulation output and the real operating state, and complete the normalized evolution correction of the model. Evolution correction refers to the digital twin model relying on the continuously accessed real operating feedback data to dynamically iterate its own simulation calculation logic and entity attribute parameters, gradually eliminate the difference between the static modeling parameters and the real operating characteristics of the physical power grid, realize the dynamic process of continuous improvement of model accuracy with the accumulation of operating data and synchronous evolution with the physical power grid state, and ensure that the model can always accurately map the real operating law of the physical power grid.

[0065] If the difference between the actual operating state and the virtual simulation results exceeds a preset threshold, it is usually caused by uncertainties not fully covered in the modeling, such as sudden fluctuations in on-site load, deviations in equipment response characteristics, and sudden changes in the external environment. For example, the actual load growth rate may be higher than the predicted value, causing the branch load rate to exceed the simulation expectation, or the actual output of distributed power sources may be lower than the simulation value due to cloud cover, causing voltage deviations. In this case, the system will immediately trigger an emergency correction mechanism. The emergency correction mechanism is a rapid response logic for exceeding the standard operating deviation. It does not need to wait for regular scheduling iterations. It directly uses the current real operating section as the initial condition and quickly conducts secondary simulation and safety constraint verification in the digital twin model. It generates targeted compensatory correction strategies, such as adding reactive power compensation devices to correct voltage deviations, fine-tuning the output of distributed power sources to match load changes, and optimizing load transfer paths to disperse overload pressure. After rapid safety verification, the correction strategies are immediately sent to the physical distribution network for execution through the communication link, quickly pulling the operating state back to a safe and controllable range and avoiding the continuous expansion of deviations that could lead to operational risks.

[0066] At the same time, the system will conduct in-depth analysis of the causes of the deviation based on the deviation data, locate the parameter links in the model with insufficient simulation accuracy, and make targeted corrections to the corresponding equipment response parameters, environmental impact weights, load prediction coefficients, etc. in the twin model. The deviation scenario will be included in the model's typical scenario library to complete the accurate iterative optimization of the model and avoid the recurrence of similar deviations.

[0067] In summary, the intelligent dispatching method for distribution networks based on digital twins provided in this application has the following technical effects: By fusing and preprocessing multi-source heterogeneous data from the distribution network to build a standardized data foundation, a dynamic digital twin model that is virtually and physically bound to the physical power grid is constructed. Based on the twin model, full-condition anomaly diagnosis, situation prediction and location of weak nodes are carried out sequentially, and sand table simulation is used to verify and screen executable scheduling strategies. In addition, the twin model is modified and corrected in a closed loop based on actual operating data. This effectively solves the above-mentioned technical problems and achieves the technical effect of improving the efficiency of scheduling decisions and the economic efficiency of power grid operation while ensuring the safe operation of the distribution network.

[0068] Example 2, based on the same inventive concept as the digital twin-based intelligent dispatching method for distribution networks in the preceding examples, such as... Figure 2 As shown in the embodiment of this application, a distribution network intelligent dispatching system based on digital twins is provided. The system includes: The fusion module 11 is used to acquire multi-source heterogeneous data of the distribution network and perform fusion preprocessing on the multi-source heterogeneous data; the construction module 12 is used to construct a digital twin model of the distribution network based on the fusion preprocessed data; the diagnosis module 13 is used to perform anomaly diagnosis on the distribution network based on the digital twin model of the distribution network; the prediction module 14 is used to perform situation prediction in the digital twin model of the distribution network based on distribution network anomalies to obtain weak nodes; the simulation module 15 is used to perform sand table simulation verification based on the weak nodes of resilience to confirm executable strategies; and the execution module 16 is used to send the executable strategies to the distribution network for execution and to evolve and correct the digital twin model of the distribution network based on the actual operating status.

[0069] Furthermore, the fusion module 11 is also used to perform the following steps: collect the operation measurement data records of the distribution network and the external environment data records, perform spatiotemporal alignment and format unification, and identify and remove bad data to complete the fusion preprocessing.

[0070] Furthermore, the construction module 12 is also used to perform the following steps: based on the fusion preprocessed data, construct an initial digital twin model of the distribution network; bind and map the physical distribution network measurement data with the initial digital twin model through a communication protocol to obtain the distribution network digital twin model.

[0071] Furthermore, the diagnostic module 13 is also used to perform the following steps: perform state estimation in the digital twin model of the distribution network based on real-time measurement data to obtain the operating status; monitor the operating safety boundary indicators based on the operating status, and identify the distribution network anomalies in the digital twin model of the distribution network.

[0072] Furthermore, the prediction module 14 is also used to perform the following steps: based on the distribution network anomaly, to extrapolate the distribution network power flow distribution for future periods in the distribution network digital twin model; and to simulate preset fault scenarios in the distribution network digital twin model according to the distribution network power flow distribution, and to evaluate the weak nodes in the distribution network resilience.

[0073] Furthermore, the simulation module 15 is also used to perform the following steps: with the goals of operational safety, operational economy and new energy consumption, generate a set of candidate scheduling strategies based on the weakly resilient nodes; input the set of candidate scheduling strategies into the digital twin model of the distribution network for sand table simulation verification; if the virtual simulation result does not meet the preset safety constraints, the candidate strategy is eliminated and re-verified; if the virtual simulation result meets the preset safety constraints, it is confirmed as the executable strategy.

[0074] Furthermore, the execution module 16 is also used to perform the following steps: collecting the actual operating status of the distribution network after executing the executable strategy, and comparing the actual operating status with the virtual simulation result in the distribution network digital twin model; if the difference between the actual operating status and the virtual simulation result exceeds a preset threshold, an emergency correction mechanism is triggered, a correction strategy is generated in the distribution network digital twin model and issued for execution.

[0075] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A distribution network intelligent dispatching method based on digital twins, characterized in that, include: Acquire multi-source heterogeneous data from the power distribution network and perform fusion preprocessing on the multi-source heterogeneous data; A digital twin model of the distribution network is constructed based on the preprocessed data. Anomaly diagnosis of the distribution network is performed based on the aforementioned digital twin model of the distribution network; Based on distribution network anomalies, situation prediction is performed in the digital twin model of the distribution network to identify nodes with weak resilience. Based on the aforementioned weak resilience nodes, a sand table simulation was conducted to verify and confirm the executable strategy. The executable strategy is deployed to the distribution network for execution, and the digital twin model of the distribution network is evolved and corrected based on the actual operating status.

2. The intelligent dispatching method for distribution networks based on digital twins as described in claim 1, characterized in that, Acquire multi-source heterogeneous data from the power distribution network and perform fusion preprocessing on the multi-source heterogeneous data, including: The system collects operational measurement data records of the power distribution network and external environmental data records, performs spatiotemporal alignment and format unification, and identifies and removes bad data to complete the fusion preprocessing.

3. The intelligent dispatching method for distribution networks based on digital twins as described in claim 1, characterized in that, A digital twin model of the distribution network is constructed based on the preprocessed data, including: Based on the preprocessed data, an initial digital twin model of the distribution network is constructed. The physical distribution network measurement data is bound and mapped to the initial digital twin model through a communication protocol to obtain the distribution network digital twin model.

4. The intelligent dispatching method for distribution networks based on digital twins as described in claim 1, characterized in that, Based on the aforementioned digital twin model of the distribution network, anomaly diagnosis of the distribution network is performed, including: Based on real-time measurement data, state estimation is performed in the digital twin model of the power distribution network to obtain the operating status; Based on the operating status, monitor the operating safety boundary indicators and identify the distribution network anomalies in the distribution network digital twin model.

5. The intelligent dispatching method for distribution networks based on digital twins as described in claim 1, characterized in that, Based on distribution network anomalies, situation prediction is performed in the digital twin model of the distribution network to identify nodes with weak resilience, including: Based on the aforementioned distribution network anomalies, the distribution network power flow distribution for future periods is inferred in the digital twin model of the distribution network; Based on the power flow distribution of the distribution network, a preset fault scenario is simulated in the digital twin model of the distribution network to assess the weak nodes in the distribution network's resilience.

6. The intelligent dispatching method for distribution networks based on digital twins as described in claim 1, characterized in that, Based on the aforementioned weak nodes, a sandbox simulation was conducted to verify and confirm executable strategies, including: With the goals of operational safety, operational economy, and new energy consumption, a set of candidate scheduling strategies based on the aforementioned weakly resilient nodes is generated; The candidate scheduling strategy set is input into the digital twin model of the distribution network for sand table simulation verification. If the virtual simulation result does not meet the preset security constraints, the candidate strategy is removed and re-verified. If the virtual simulation result satisfies the preset security constraints, it is confirmed as the executable strategy.

7. The intelligent dispatching method for distribution networks based on digital twins as described in claim 6, characterized in that, The executable strategy is distributed to the distribution network for execution, and the digital twin model of the distribution network is evolved and modified based on the actual operating status, including: The actual operating status of the distribution network after executing the executable strategy is collected, and the actual operating status is compared with the virtual simulation results in the digital twin model of the distribution network; If the difference between the actual operating state and the virtual simulation result exceeds a preset threshold, an emergency correction mechanism is triggered, and a correction strategy is generated in the digital twin model of the power distribution network and executed.

8. A distribution network intelligent dispatching system based on digital twins, characterized in that: The system is used to implement the digital twin-based intelligent dispatching method for distribution networks according to any one of claims 1 to 7, the system comprising: The fusion module is used to acquire multi-source heterogeneous data from the power distribution network and perform fusion preprocessing on the multi-source heterogeneous data; The building module is used to construct a digital twin model of the distribution network based on the fused preprocessed data; The diagnostic module is used to perform anomaly diagnosis on the distribution network based on the digital twin model of the distribution network; The prediction module is used to perform situation prediction in the digital twin model of the distribution network based on distribution network anomalies, and to identify nodes with weak resilience. The simulation module is used to perform sand table simulations and verifications based on the weak resilience nodes to confirm executable strategies; The execution module is used to distribute the executable strategy to the distribution network for execution and to evolve and correct the digital twin model of the distribution network based on the actual operating status.

Images