Digital twin energy efficiency optimization method for regional power distribution system

Abstract

The invention belongs to the technical field of power systems and automation thereof, and discloses a digital twin energy efficiency optimization method for a regional power distribution system, which comprises the following steps of: acquiring real-time operation data of core equipment in the power distribution system, quantifying data quality through a multi-dimensional evaluation mechanism and adding a corresponding label; adjusting data acquisition priority and communication resource allocation in combination with the dynamic scene to generate a trusted data set; dividing cells, electing a temporary coordinator, constructing a communication topology between a local digital twin and the cells, and generating a dynamic tough cell architecture diagram; an optimization strategy set is generated through matching, and an autonomous optimization instruction set is generated through local digital twin forward simulation; in combination with a preset degradation plan, when resources in cells are insufficient, a cross-cell game cooperation mechanism is triggered, and a cross-cell cooperation instruction set and a security check report are generated; performance analysis is carried out through a multi-dimensional evaluation system, a problem root is traced and positioned, and a whole process link is optimized and iterated.

Description

Technical Field

[0001] This invention relates to the field of power systems and their automation technology, and more specifically, to a digital twin energy efficiency optimization method for regional power distribution systems. Background Technology

[0002] With the advancement of the "dual carbon" target, regional power distribution systems are exhibiting a complex pattern of high penetration of new energy sources and multi-faceted interaction between power sources, grids, loads, and storage. Digital twin technology has become a core support for energy efficiency optimization. However, it still faces many challenges in actual operation, such as insufficient reliability of multi-source data, rigid acquisition and resource allocation mechanisms in dynamic scenarios, and a lack of scientific organization for cross-regional collaboration. These issues lead to low accuracy in energy efficiency optimization, delayed response, and prominent safety risks.

[0003] To address these energy efficiency optimization issues, existing solutions still have some shortcomings: data processing lacks dynamic adaptation and multi-dimensional quality assessment, resulting in weak ability to generate reliable datasets; cell partitioning is mostly based on fixed substation boundaries, lacking resilience, while digital twins are mostly modeled in full detail, with low lightweighting and poor adaptability to edge deployment; the policy library is limited, the matching mechanism is imperfect, and there is a lack of digital twin simulation verification, making it difficult to guarantee policy feasibility; cross-cell collaboration relies on simple degradation methods, with insufficient security verification dimensions, resulting in low security and efficiency of collaboration. Summary of the Invention

[0004] To overcome the aforementioned deficiencies of the prior art and to achieve the above objectives, the present invention provides the following technical solution: a digital twin energy efficiency optimization method for a regional power distribution system, comprising: S1: Acquire real-time operating data of core equipment in the power distribution system, quantify data quality through a multi-dimensional evaluation mechanism and attach corresponding labels, and adjust data acquisition priority and communication resource allocation in combination with dynamic scenarios to generate a reliable dataset; S2: Based on a trusted dataset, cells are partitioned using real-time electrical coupling, a temporary coordinator is elected, and a lightweight local digital twin and cell-to-cell communication topology are constructed to generate a dynamic resilient cell architecture diagram and a coordinator configuration table. S3: Based on the dynamic resilience cell architecture diagram and coordinator configuration table, each cell coordinator generates an optimization strategy set through a dual-strategy library matching mechanism according to the dual-factor matching mechanism, and generates an autonomous optimization instruction set through local digital twin forward simulation; S4: Based on the autonomous optimization instruction set and combined with the pre-set degradation plan, when the resources within the cell are insufficient, the cross-cell game collaboration mechanism is triggered. Through multi-layer security boundary verification, mutual assistance transactions are achieved, and a cross-cell collaboration instruction set and security verification report are generated. S5: Based on the execution effect of cross-cell collaborative instruction set and the resource status of each cell, performance analysis is conducted through a multi-dimensional evaluation system to trace and locate the root cause of the problem and optimize and iterate the entire process.

[0005] Furthermore, the method of quantifying data quality and attaching corresponding labels through a multi-dimensional evaluation mechanism includes: Collect real-time operating data of four types of core equipment in the power distribution system: source-side equipment, grid-side equipment, load-side equipment, and storage-side equipment; Data quality is quantified through a five-dimensional evaluation mechanism that considers equipment accuracy, transmission latency, physical verification, historical stability, and multi-source consistency, thereby obtaining the reliability of each data point. Then, based on the quality and reliability, a dynamic reliability label is attached to each piece of data and associated with the corresponding device identifier, and the data is integrated into a multi-source dataset.

[0006] Furthermore, the method for generating the trusted dataset includes... Based on a multi-source dataset with dynamic credibility labels, the current running scenario is identified by a preset dynamic scenario library and a multi-source triggering mechanism, and the scenario priority is determined by combining the quality and credibility of the data. Based on scenario priority, the data collection priority of each device is adjusted; and the running scenarios in the dynamic scenario library are mapped to the preset communication resource pool, and communication resources for each device's data are allocated according to the mapping relationship. Simultaneously, valid data and abnormal data are divided according to quality and reliability. All valid data are integrated to generate a reliable dataset with scene association. Abnormal data is isolated and the reasons are labeled to generate an abnormal isolation list.

[0007] Furthermore, the method for electing a temporary coordinator includes: Based on the trusted dataset and the anomaly isolation list, devices associated with abnormal data are removed, and the remaining devices are marked as reliable devices; Based on the step-by-step verification rules of electrical coupling → communication quality → load balancing, the reliable equipment in adjacent transformer areas is divided into dynamically resilient cells, forming a cell cluster. Candidate devices with data processing and command issuance capabilities are selected from each cell. An independent election is conducted within each cell according to an election scoring mechanism to obtain the optimal candidate device as the temporary coordinator for the corresponding cell.

[0008] Furthermore, the method for constructing a lightweight local digital twin and inter-cell communication topology, and generating a dynamic resilient cell architecture diagram and coordinator configuration table includes: Based on the cell cluster and coordinator election results, all devices in the cell cluster are divided into high-confidence devices and low-confidence devices according to quality reliability. Each cell coordinator extracts highly reliable device parameters and builds a lightweight local digital twin that retains core functions; Simultaneously, the inter-cell communication topology is constructed according to the structure of the core layer and the edge layer, and the communication links are optimized to generate a dynamic resilient cell architecture diagram that marks cell boundaries, devices, and communication relationships; the performance and communication parameters of each coordinator are recorded in a structured manner in sync to generate a coordinator configuration table.

[0009] Furthermore, the method for generating the optimization strategy set includes: Based on the dynamic resilience cell architecture diagram and coordinator configuration table, the pre-built dual strategy library for normal operation and stress testing is invoked. The coordinator extracts the fault type features, impact range features, and resource constraint features of the current scenario from the multi-source dataset, quantifies them, calculates the similarity with the scenarios in the dual-strategy library, and matches them to select scenarios that meet the similarity criteria. Simultaneously, the coordinator retrieves the historical decision records of the models constructed by various algorithms within the local digital twin to obtain the accuracy assessment of the model's credibility. For strategies targeting scenarios that meet similarity criteria, scenario similarity and model credibility are used as two factors. A two-factor matching mechanism is used to select suitable strategy combinations to form an optimized strategy set.

[0010] Furthermore, the method for generating the autonomous optimization instruction set includes: The optimization strategy set is broken down into device-level operations and mapped to a local digital twin; The coordinator calls the local digital twin to perform forward simulation at a preset time step, dynamically monitoring the constraint satisfaction during the simulation process; For a strategy that is fully simulated, the feasibility of the strategy is evaluated according to the multi-dimensional credibility evaluation rules. Strategies that meet the criteria are judged as feasible strategies, and strategies that do not meet the criteria are revised and re-evaluated through a second simulation. The final feasible strategy is converted into device-executable instructions and encapsulated into a standardized, categorized, and ordered autonomous optimization instruction set.

[0011] Furthermore, the methods for triggering the cross-cell game cooperation mechanism when intracellular resources are insufficient include: By comparing the resource requirements of the autonomous optimization instruction set with the real-time available resources within the cell, the resource gap is calculated to determine whether the resources within the cell are sufficient: If sufficient, maintain the original autonomous optimization instruction set; if insufficient, execute the pre-set multi-level degradation plan and recalculate the resource gap; if still insufficient, trigger the cross-cell game collaboration mechanism, and select high-trust neighboring cells as collaborative cells based on the inter-cell communication topology. Resource-deficient cells and collaborative cells interact regarding resource needs and costs. Based on game theory, a collaborative solution is found that allows both parties to accept a certain amount of resources and costs, thus forming a cross-cell collaborative solution.

[0012] Furthermore, the generation methods for the cross-cell collaborative instruction set and security verification report include: Based on the cross-cell collaboration scheme, the security of the collaboration scheme is verified layer by layer by calling the local digital twin through a three-layer security boundary of electrical security, communication security and transaction compliance. Once all verifications are passed, the mutual assistance transaction is confirmed and the transaction parameters are fixed. A cross-cell collaborative instruction set with a collaborative identifier is generated, and a security verification report containing verification results and risk warnings is compiled simultaneously.

[0013] Furthermore, the method of conducting performance analysis through a multi-dimensional evaluation system to trace and locate the root cause of problems and optimize and iterate throughout the entire process includes: Based on the execution effect of cross-cell collaborative instructions and the resource status of each cell, an evaluation system is constructed through four dimensions of indicators: collaborative effect, system performance, economy and robustness. The system analyzes the overall performance and identifies the root causes of low-rated indicators. A full-process optimization report is generated for the entire process of root cause optimization iteration data collection rules, cell generation parameters, dual-strategy library and game parameters, forming a closed loop.

[0014] The technical effects and advantages of the digital twin energy efficiency optimization method for regional power distribution systems proposed in this invention are as follows: This invention focuses on the optimization needs of the entire regional power distribution chain. First, by constructing a five-dimensional quality assessment and dynamic scenario adaptation mechanism, a reliable dataset with scenario association is generated, solving the problems of data reliability and collection efficiency, and providing high-quality data support for subsequent optimization. Second, dynamic cells are divided according to layer-by-layer verification rules to build a lightweight twin and an efficient communication topology, improving cell response speed and twin deployment efficiency, and adapting to the real-time needs of the edge side. Then, relying on a dual-strategy library and a dual-factor matching mechanism, an autonomous instruction set is generated through twin simulation and multi-dimensional verification to ensure strategy adaptability and feasibility and avoid ineffective energy loss. Next, by combining a three-level degradation contingency plan and a cross-cell game mechanism, collaboration is achieved through three-layer security verification, balancing collaboration efficiency and security, and solving the problem of resource-scarce scenarios. Finally, a four-dimensional evaluation system and a periodic iteration mechanism are constructed to form a continuous optimization closed loop, improving system energy efficiency and robustness, and adapting to complex power distribution scenarios.

[0015] This invention significantly improves the energy efficiency optimization accuracy, dynamic response speed, and operational stability of regional power distribution systems through end-to-end design, making it suitable for diverse power distribution scenarios. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of a digital twin energy efficiency optimization method for a regional power distribution system according to the present invention; Figure 2This is a schematic diagram of the two-factor matching strategy screening process in the digital twin energy efficiency optimization method for a regional power distribution system according to the present invention. Figure 3 This is a schematic diagram of the digital twin energy efficiency optimization system for a regional power distribution system according to the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Example 1 Please see Figure 1 and Figure 2 As shown in this embodiment, a digital twin energy efficiency optimization method for a regional power distribution system includes: S1: Acquire real-time operating data of core equipment in the power distribution system, quantify data quality through a multi-dimensional evaluation mechanism and attach corresponding labels, and adjust data acquisition priority and communication resource allocation in combination with dynamic scenarios to generate a reliable dataset.

[0019] It should be noted that this technology addresses the issues of insufficient credibility of multi-source data and rigid data acquisition and communication resource allocation mechanisms in dynamic scenarios. Through multi-dimensional quality assessment and dynamic scenario adaptation, it generates a reliable dataset, providing data support for subsequent optimization.

[0020] S2: Based on a trusted dataset, cells are partitioned using real-time electrical coupling, a temporary coordinator is elected, and a lightweight local digital twin and cell-to-cell communication topology are constructed to generate a dynamic resilient cell architecture diagram and a coordinator configuration table.

[0021] It should be noted that this method addresses the lack of a scientific and resilient cell organization method for cross-regional equipment collaboration and the poor adaptability of digital twin technology. It constructs dynamic resilient cells and lightweight twins to lay the foundation for efficient collaboration and real-time simulation.

[0022] S3: Based on the dynamic resilience cell architecture diagram and coordinator configuration table, each cell coordinator generates an optimization strategy set through a dual-strategy library matching mechanism using a two-factor matching mechanism, and generates an autonomous optimization instruction set through forward simulation of the local digital twin.

[0023] It should be noted that, in order to address the problems of optimization strategies being detached from actual operating conditions, having low accuracy, and exhibiting delayed response, a feasible autonomous optimization instruction set is generated through two-factor matching and twin simulation verification.

[0024] S4: Based on the autonomous optimization instruction set and combined with the pre-set degradation plan, when the resources within the cell are insufficient, the cross-cell game collaboration mechanism is triggered. Through multi-layer security boundary verification, mutual assistance transactions are achieved, and a cross-cell collaboration instruction set and security verification report are generated.

[0025] It should be noted that this technology is designed to address the lack of a trustworthy game theory mechanism and the significant security risks associated with cross-regional collaboration. By employing game theory collaboration and multi-layered security verification, it enables secure and efficient cross-cell mutual assistance in resource-constrained scenarios.

[0026] S5: Based on the execution effect of cross-cell collaborative instruction set and the resource status of each cell, performance analysis is conducted through a multi-dimensional evaluation system to trace and locate the root cause of the problem and optimize and iterate the entire process.

[0027] It should be noted that, in order to address the problem of energy efficiency optimization failing to improve continuously and the system having poor adaptability, closed-loop evaluation and full-process iteration are used to continuously improve the system's optimization capabilities and robustness.

[0028] Methods for quantifying data quality and attaching corresponding labels through multi-dimensional evaluation mechanisms include: Collect real-time operating data of four types of core equipment in the power distribution system: source-side equipment, grid-side equipment, load-side equipment, and storage-side equipment; Source-side equipment: If it is photovoltaic power generation, the equipment is a photovoltaic inverter (data includes the generator set's real-time output, active power, reactive power, maximum tracking voltage range, etc.). If it is wind power generation, the equipment is a wind power converter (data includes real-time output, power regulation range, wind speed, etc.). Grid-side equipment: Lines (data includes three-phase voltage, three-phase current, resistance, reactance, etc.), distribution transformer related data (including load rate, top oil temperature, etc.), switchgear related data (including opening and closing status, protection action signals, etc.). Load-side equipment: data related to smart meters for residential users (including real-time power consumption, interruptible load capacity, etc.), data related to industrial users (including real-time load of production workshops, inverter operating status, etc.), and data related to data centers (including IT load power, UPS output status, etc.). Energy storage-side equipment: Battery energy storage related data (including battery SOC (remaining capacity), charging and discharging current, conversion efficiency, etc.), flywheel energy storage related data (including speed, output power, etc.). It should be noted that the specific equipment types of the four core equipment categories of source, grid, load and storage will vary depending on the business scenario (such as a single thermal power generation scenario or a mixed scenario with thermal power, photovoltaic and hydropower generation). Here, we take photovoltaic power generation and wind power generation as examples and only list some equipment types of the core equipment on each side. Data quality is quantified through a five-dimensional evaluation mechanism that considers device accuracy, transmission latency, physical verification, historical stability, and multi-source consistency. Specifically: The accuracy dimension of the equipment is assigned a value based on the accuracy level of the data acquisition terminal (according to the standard GB776-76 "General Technical Conditions for Measurement and Indicating Instruments"); for example: a 0.2-level terminal (such as a high-precision PMU) gets 100 points, a 0.5-level terminal gets 80 points, a 1.0-level terminal gets 60 points, and a 2.0-level terminal (such as an old sensor) gets 40 points. The transmission latency dimension is assigned a value based on the transmission latency of data from the acquisition terminal to the edge gateway; Example: ≤50ms (meets real-time control requirements) gets 100 points, 50-100ms gets 80 points, 100-200ms gets 50 points, and >200ms (latency exceeds the standard) gets 30 points. The physical verification dimension verifies the rationality of the data based on the physical laws of the power grid; for example: 100 points for complete compliance (e.g., total power of the transformer area = sum of power of each branch meter ±3%), 70 points for slight deviation (±3%-5%), 40 points for significant deviation (±5%-10%), and 10 points for serious deviation (>±10%). The historical stability dimension is assigned a value based on the number of anomalies in the data within a recent period (e.g., 30 days); Example: ≤2 anomalies = 100 points, 3-5 anomalies = 70 points, 6-10 anomalies = 40 points, >10 anomalies = 20 points; Multi-source consistency dimension compares multi-source collected data of the same monitoring object to identify collection deviations; Example: deviation ≤2% gets 100 points, 2%-5% gets 80 points, 5%-10% gets 50 points, >10% gets 30 points; The five-dimensional scores are weighted and fused to obtain the quality and credibility score of each data point (out of 100). Furthermore, a dynamic credibility tag (including quality credibility score and scores for each dimension) is attached to each piece of data and associated with a device identifier to achieve quantifiable and traceable data quality.

[0029] Methods for generating trustworthy datasets include Based on the operating characteristics of the regional power distribution system, a two-level dynamic scenario library of basic scenarios and emergency scenarios is preset, and the judgment rules, core data types and scenario priority levels of each scenario are clearly defined. Exemplary core content of the scenario library: [Scenario Type: Extreme Weather Warning; Triggering Method: External Signal Triggering; Judgment Rules: Receive typhoon / cold wave warning (orange or above) from meteorological department, and the icing monitoring value of power distribution lines in the area is ≥5mm; Core Data Types: Line icing thickness, wind speed, equipment temperature; Scenario Priority: P0 (highest)] [Scenario Type: Normal Operation; Triggering Method: Periodic Triggering; Judgment Rule: No sudden scenario triggering, load fluctuation ≤10% during the flat period from 9:00 to 17:00 throughout the day; Core Data Type: New Energy Forecast Value (referring to the forecast value of power generation of new energy power plants), Energy Storage Charging and Discharging Plan; Scenario Priority: P1] [Scenario type: Equipment maintenance; Triggering method: Manual command triggering; Judgment rule: Receive maintenance work order (such as transformer maintenance), the equipment involved is in the "under maintenance" status; Core data types: current of the equipment under maintenance, load of adjacent equipment; Scenario priority: P2] Based on multi-source datasets, the current running scenario is identified through a dynamic scenario library and a multi-source triggering mechanism; The multi-source triggering mechanism combines real-time data, external signals, and manual commands to ensure comprehensive scene recognition and avoids misjudgments through cross-validation. Specifically: Real-time data triggering: The edge gateway scans the multi-source dataset at a fixed frequency, extracts the core data marked by the dynamic scene library, compares it with the judgment rules of the dynamic scene library, and triggers scene recognition if the rules are met. External signal triggering: Receive early warning signals from meteorological departments and higher-level dispatch centers via API interface to directly trigger corresponding emergency scenarios; Manual command triggering: Maintenance personnel issue maintenance commands through the maintenance platform, match the "equipment maintenance" scenario, and associate the real-time data verification status of the equipment under maintenance; When multiple scenarios are triggered simultaneously, a score is assigned to the scenario priority based on the quality and reliability of the data (out of 100, e.g., P0=100, P1=80, P2=60). The scenario priority scores are then weighted and combined with the quality and reliability scores to obtain the final scenario priority. Based on scenario priority, the data collection priority of each device is adjusted, specifically by adjusting the data collection frequency and upload order to ensure that core data is collected first. Specifically, in the P0 scenario, the frequency of core data acquisition is increased (such as the SOC of an energy storage system); in the P1 scenario, data is acquired at the initial frequency, and the core data is uploaded in advance; in the P2 scenario, non-core data is delayed until the off-peak hours (such as 2-4 am) for batch uploading to reduce communication resource consumption. Map the running scenarios in the dynamic scenario library to the pre-set communication resource pool, which is a pre-planned communication resource pool of types such as 5G_uRLLC slice, 5G_eMBB slice, LoRa local network; Example mapping methods: P0 level scenarios are allocated 5G_uRLLC slices (covering the entire area); P1 level scenarios are allocated 5G_eMBB slices (covering the core area); P2 level scenarios are allocated LoRa local network or off-peak fiber optic (covering the designated maintenance area). Allocate communication resources for each device's data according to the mapping relationship; Simultaneously, valid data and abnormal data are divided into two categories based on quality reliability using a threshold segmentation method. For example, if the quality reliability threshold is set to 60, all data with a score <60 are classified as abnormal data, and all data with a score ≥60 are classified as normal data. Integrate all valid data and organize it in three dimensions: device type, scene type, and timestamp to generate a trusted dataset with scene association, including: device ID, collection time, data value (e.g., 500kW), dynamic trust label, and scene. Isolate and summarize abnormal data to generate a structured list of abnormal data isolation, including but not limited to: abnormal data ID, device ID, and dynamic trustworthiness tag; By generating multi-source datasets with scene associations and summarizing abnormal data to form an isolation list, reliable and scene-adaptive data support is provided for subsequent cell construction.

[0030] Methods for electing a temporary coordinator include: Based on the trusted dataset and the anomaly isolation list, devices associated with abnormal data are removed, and the remaining devices are marked as reliable devices; For reliable devices, read the distribution network topology information and construct a device association matrix consisting of device ID, adjacent devices, impedance value, and signal strength. Pre-calibrate the layer-by-layer verification rules and thresholds for electrical coupling → communication quality → load balancing; it should be noted that → is the sequence symbol. Among them, electrical coupling parameters: to ensure that the power transmission loss within the cell meets the expected (e.g., ≤5%), an electrical coupling threshold (e.g., impedance ≤0.5Ω) is set. Communication quality parameters: To meet the needs of intracellular data interaction (e.g., packet loss rate ≤ 1%), set communication quality thresholds (e.g., signal strength ≥ -80dBm). Load balancing parameters: To reserve a certain amount of computing power (e.g., 40% computing power) to cope with sudden optimization needs, set a load balancing threshold (e.g., CPU utilization < 60%). According to the preset layer-by-layer verification rules, the reliable equipment in adjacent transformer areas is divided into dynamically resilient cells, forming a cell cluster. Specifically, the method for dividing the layer-by-layer verification rules is as follows: The first step is to verify the electrical coupling rules: using the substation as the anchor point, extract the equipment clusters that meet the line impedance standards to form preliminary electrical units; if a device is associated with two clusters at the same time, it is assigned to the cluster with the larger total number of associated devices; if a cluster covers more than three substations, it is split according to the minimum impedance. The second step is communication quality rule filtering: For the initial electrical unit, devices with substandard signal strength are removed; if there are fewer than 5 devices remaining in the unit, it is merged with the adjacent electrical unit (after merging, the electrical coupling rule verification needs to be performed again). The third step is to optimize the load balancing rules: collect the CPU utilization rate of edge devices (such as distribution area gateways and energy storage controllers) within the initial electrical unit. If the CPU utilization rate exceeds the standard, split the devices according to their types to ensure that the CPU utilization rate meets the load balancing threshold after splitting. Ultimately, a dynamic resilient cell cluster is formed, covering 1-3 substations with tight electrical coupling, stable communication, and balanced computing power; Candidate devices with data processing and command issuance capabilities are selected from each cell (prioritizing energy storage controllers (with built-in BMS computing power) and distribution area edge gateways (with integrated edge computing modules), while excluding devices with high CPU utilization or low SOC to ensure the reliability of candidate device performance). The election scoring mechanism is defined as follows: Based on a trusted dataset, election scoring indicators for candidate devices are extracted, including computing power indicators, energy storage capacity indicators, and quality reliability indicators; these indicators are then quantified into scores using expert experience and weighted and fused to obtain the final comprehensive election score. Independent elections are conducted within each cell according to the election scoring mechanism. The candidate device with the highest comprehensive election score is selected as the optimal candidate device, and the optimal candidate device is selected as the temporary coordinator for the corresponding cell. Real-time monitoring of cell size (1-3 zones are normal) and coordinator computing power redundancy (such as CPU utilization <50% is normal). An alarm will be triggered if either indicator is abnormal. When an alarm is triggered, if the cell size exceeds 3 zones, the boundary device is split according to the farthest electrical distance; if the coordinator's CPU utilization suddenly increases, the backup candidate device (the second highest scorer) is activated to take over, and the original coordinator is demoted to a normal node.

[0031] Methods for constructing lightweight local digital twins and inter-cell communication topologies, and generating dynamic, resilient cell architecture diagrams and coordinator configuration tables include: Based on the cell cluster and coordinator election results, high-reliability devices (≥70 points) are selected through a preset quality reliability threshold (e.g., 70 points). Each cell coordinator extracts parameters of high-confidence devices from multi-source datasets and categorizes them by device type; Based on high-reliability equipment parameters, and following the principle of retaining core functions and simplifying unnecessary details, a lightweight local digital twin is built to retain core functions: the simplification direction is to ignore the internal micro-mechanisms of the equipment, retain macro-characteristics (such as the SOC-charge and discharge power relationship), simplify electromagnetic transient processes, and focus on steady-state power flow calculation; the architecture is based on the power flow calculation model and includes a device layer (mapping the physical devices and parameters within the cell), a topology layer (reproducing the electrical connection relationship between devices), and a simulation layer (integrating core general algorithm models for predicting line loss, renewable energy consumption, voltage, etc. (such as power flow calculation, power balance verification, etc.)), built on the OpenDSS open source library; The local digital twin initializes and inputs device parameters, ensuring consistency between the digital twin and the initial state of the physical system based on real-time data; the accuracy is verified by comparing physical measurements with simulation values ​​(e.g., deviation ≤ 5%); if the deviation exceeds the standard, the source of the parameters is traced back to extract higher quality and more reliable parameters to correct the local digital twin. Synchronization is based on the spatial distribution and electrical association of cells, and a hybrid topology of core layer and edge layer is used to construct the inter-cell communication topology: each coordinator in the core layer is directly connected according to the principle of shortest electrical distance, and non-coordinator devices in the edge layer only communicate with the coordinator of their own cell (avoiding redundant connections across cells); after the coordinator summarizes the data of its own cell, it pushes it to the associated cells according to preset rules (such as only pushing "the status of resources that need mutual assistance"). The method for testing and optimizing communication links is as follows: coordinators send test messages (including timestamps) to each other, and count the transmission latency (≤100ms is acceptable) and packet loss rate (≤1% is acceptable). If both meet the predetermined standards, the communication link is considered to be acceptable. For communication links with excessive latency, communication repeaters (such as LoRa gateways) are deployed to enhance the signal. For links with excessive packet loss rate, retransmission is performed. Generate a dynamic resilience cell architecture diagram, including: cell boundaries, coordinator location, and electrical and communication connections between cells; Synchronously record the performance and communication parameters of each coordinator in a structured manner, and generate a coordinator configuration table, including basic information (cell ID to which the coordinator belongs, device ID, etc.), performance parameters (quantified scores of election scoring indicators), and communication parameters (IP address, list of associated cell coordinators (directly connected nodes), etc.).

[0032] The methods for generating optimization strategy sets include: Based on the dynamic resilience cell architecture and coordinator configuration table, the pre-built dual strategy library for normal operation and stress testing is invoked. Among them, the dual strategy library for normal operation and stress testing is based on cell-level operating scenarios. The pre-built normal operation strategy library (adapted to stable scenarios) and stress testing strategy library (adapted to sudden scenarios) are stored in the format of scenario tags, strategy content and constraints. The constraints are strictly related to the device parameters within the cell (such as SOC upper and lower limits, line load rate, etc., taken from high-confidence device parameters). Example of a standard strategy library: Scenario tag: Midday PV power generation; Strategy content: Energy storage charging + inverter reactive power compensation + adjustable load shifting; Constraints: SOC≤90%, line load rate≤75%; Example of a stress test strategy library: Scenario tag: Photovoltaic sudden drop; Strategy content: Emergency discharge of energy storage + closing of the interconnection switch + disconnection of tertiary loads; Constraints: Discharge power ≤ rated power of energy storage, voltage deviation ≤ ±7%; The coordinator extracts and quantifies the scene features of the current scenario from the multi-source dataset, including fault type features (determined based on the device status data within the cell (e.g., no fault, photovoltaic inverter disconnected from the grid), such as 100 points for no fault and 80 points for a single device fault), impact range features (quantified based on the cell boundary of the dynamic resilience cell architecture diagram, according to the number of affected transformer areas within the cell (e.g., 100 points for 1 transformer area, 80 points for 2-3 transformer areas)), and resource constraint features (assessing the available resources within the cell (e.g., SOC, adjustable load capacity, etc.), such as 100 points for sufficient resources and 70 points for moderately insufficient resources). The similarity (value 0-1) between the quantified scene features and each scene in the dual-strategy library is calculated. The similarity is calculated as follows: the product of the current scene feature score and the scene feature score in the dual-strategy library is weighted and fused with the corresponding scene feature weight. Then, the weighted fusion result is divided by (current scene total score × dual-strategy library scene total score). The weights of each scene feature can be preset based on expert experience. Match the current scenario with each scenario in the dual-strategy library based on their similarity. Select scenarios in the dual-strategy library with a similarity greater than or equal to a preset similarity threshold (e.g., 75%) as similarity-compliant scenarios to ensure strategy adaptability. At the same time, the coordinator retrieves the historical decision records of the models built by each algorithm in the local digital twin, calculates the accuracy, and evaluates the model's credibility through the accuracy. Specifically, a tolerance threshold for the deviation between the predicted and actual values ​​is set for each model (for example, taking the line loss model as an example, if the tolerance threshold is set to 5%, then the deviation between the predicted and actual line loss values ​​is ≤5%). Accuracy = (Number of times the deviation between model prediction and actual value is ≤ tolerable threshold) ÷ Total number of times × 100%; The weights of each model are assigned according to their accuracy (the sum of the weights is 1), and the formula is: Weight of a model = Accuracy of that model ÷ Sum of the accuracies of all models; For strategies targeting scenarios that meet similarity criteria, scenario similarity and model credibility are used as two factors. A two-factor matching mechanism is used to select suitable strategy combinations to form an initial set of optimized strategies. The two-factor matching mechanism is as follows: the scene similarity and the model average credibility (model average credibility = the sum of the accuracy of each model × the corresponding weight) are weighted and fused to obtain a comprehensive score. The comprehensive scores are sorted in descending order, and the strategies corresponding to the top 1-3 dual-strategy library scenes (the number of which needs to be equal to the number of intracellular platform areas) are selected and combined to form an optimized strategy set.

[0033] The methods for generating autonomous optimization instruction sets include: Based on the characteristics of cell operation and the timeliness requirements of strategy, the core simulation parameters are preset, including the time step (dynamically adjusted according to the scenario (usually 5 minutes / step for normal scenarios and 1 minute / step for emergency scenarios), covering the entire execution cycle of the strategy) and input data (including the current equipment-related status in the trusted dataset (such as photovoltaic output, load power, SOC, etc.) and the current real-time environmental parameters (light intensity, temperature, used to correct photovoltaic prediction deviations)). Core constraints are extracted from the parameters of highly reliable equipment and combined with corresponding constraint thresholds preset in advance by experts to serve as red line standards for simulation verification. These include safety constraints (line load rate, transformer top oil temperature, energy storage charging and discharging power, etc.) and economic constraints (line loss rate (threshold can be set based on the historical best value of the cell), energy storage charging and discharging efficiency, etc.). The optimization strategy set is broken down into device-level executable operation instruction prototypes to clarify the operation object, action type, and parameter range; for example, on the source side: photovoltaic inverter "active power limit (0-rated value)" and "reactive power compensation (-0.4~0.4Mvar)"; Mapping the disassembled operation command parameters to the device and simulation layers of the local digital twin requires ensuring that the analog inputs match the characteristics of the physical devices. For example, “the charging power of the energy storage system is 500kW” is mapped to the “charging and discharging power interface” of the local digital twin energy storage device and associated with its SOC-power curve (e.g., when SOC≥90%, the charging power is automatically limited to 200kW). The coordinator calls the local digital twin to perform forward simulation at preset time steps. At each time step, the current operation parameters are input (e.g., step 1: energy storage charging 300kW, photovoltaic output limited to 1000kW). The simulation layer outputs the node voltage, line power, equipment load rate, etc. at that moment based on the power flow calculation model. It automatically iterates to the next time step, using the simulation result of the previous step as the initial state (e.g., energy storage SOC in step 2 = SOC in step 1 + charging amount × efficiency), until the entire strategy cycle is covered. The system monitors the constraint satisfaction status during the simulation process in real time and sets up a three-level early warning mechanism: all indicators are within the constraint threshold and are considered normal; indicators equal to the constraint threshold are judged to require an early warning, the cause of the deviation is recorded (such as the load forecast being too high) and an early warning is issued; indicators exceeding the constraint threshold are judged to be out of bounds, the simulation is immediately paused, and the corresponding operation parameter is marked as a "risk point". For a fully simulated strategy, the scores of the three dimensions of security compliance, economy and stability are evaluated according to the multi-dimensional feasibility assessment rules. Then, the scores are weighted and fused to obtain a comprehensive feasibility score, which is used as the strategy feasibility. Among them, safety compliance: 100 points for no over-limit violations, 10 points deducted for each warning, and 30 points deducted for each over-limit violation; economy: 2 points are added for every 0.1% decrease in line loss rate based on the benchmark value; stability: 100 points for voltage deviation standard deviation ≤1%, and 10 points are deducted for every 0.5% increase. Set a policy credibility threshold range. If the policy feasibility is greater than or equal to the maximum value of the policy credibility threshold range, it is judged as a qualified policy. If the policy feasibility is less than the maximum value of the policy credibility threshold range but greater than or equal to the minimum value of the policy credibility threshold range, it is judged as a substandard policy. If the policy feasibility is less than the minimum value of the policy credibility threshold range, the policy is removed and marked as infeasible. Alternative policies are added from the suboptimal policies, and the simulation process is repeated. The strategies that meet the standards are deemed feasible, while those that do not meet the standards are corrected (by fine-tuning parameters for warning items, such as reducing the energy storage charging power from 300kW to 250kW when the line load rate is warned) and then re-simulated and re-evaluated. The final feasible strategy is converted into device-executable instructions and encapsulated in a standardized format of device ID + timestamp + operation type + parameter value + checksum (to ensure instruction integrity): Then, the standardized instructions are categorized by device type and execution urgency, and encapsulated into a standardized, categorized, and sorted autonomous optimized instruction set; Among them, the equipment types are classified as: source-side instruction set, network-side instruction set, load-side instruction set, and storage-side instruction set (for easy batch distribution). Execution urgency priority: Emergency scenario instructions (such as emergency energy storage discharge) are marked P0 (execute immediately), and normal scenario instructions (such as load shifting) are marked P1 (execute by timestamp).

[0034] The ways in which cross-cell game cooperation mechanisms are triggered when resources within a cell are insufficient include: The coordinator compares the resource requirements of the autonomous optimization instruction set (e.g., an instruction requires 500kW of energy storage and charging) with the real-time available resources within the cell, calculates the resource gap (resource gap = (instruction resource requirement - available resources) ÷ instruction resource requirement × 100%), and determines whether the resources within the cell are sufficient. A preset resource sufficiency threshold (e.g., 10%) is set. If the resource gap is ≤10%, it indicates that intracellular autonomy is feasible and is judged as having sufficient resources. If the resource gap is >10%, it is judged as having a slight gap. If resources are sufficient, maintain the original autonomous optimization instruction set; if resources are insufficient, execute the pre-set multi-level degradation contingency plan. The multi-level degradation contingency plan includes three preset levels: mild, moderate, and severe. Examples of triggering conditions and implementation measures for each level are as follows: Mild deficit (resource deficit ≤ 20%): Prioritize reducing non-essential flexible loads (such as residents being able to interrupt air conditioning) and lowering photovoltaic power restrictions (increasing local output); Moderate shortage (20% < resource shortage ≤ 50%): Activate the cell's backup power supply (such as emergency discharge of energy storage to the lower limit of SOC), adjust the transformer taps (optimize voltage support); Severe shortage (resource shortage > 50%): Cut off tertiary loads (such as public lighting) and limit industrial load power (according to contract priority); After the coordinator executes the corresponding downgrade plan according to the resource gap level, it recalculates the resource gap and then re-determines the resource situation. If the resource shortage is still determined, the cross-cell game cooperation mechanism is triggered: The first step is to select high-trust neighboring cells with redundant real-time resource status and compliant communication links as cooperative cells based on the inter-cell communication topology. The criteria for determining high-trust neighboring cells are as follows: extract the historical collaboration data of each neighboring cell, and for each cell, take the ratio of the number of successful collaborations to the total number of collaborations as the trust level of the cell; neighboring cells with a trust level ≥ 0.8 are considered high-trust neighboring cells. The second step involves resource-deficient cells acting as demanders and collaborative cells acting as suppliers, with both parties interacting on resource needs and costs. Specifically, the collaborative cells selected from the demand direction send resource demand messages, including the gap type, demand duration, and acceptable cost limit. Then, after receiving the resource demand message, the supplier provides feedback on the amount of resources available, unit cost, and response preparation time based on its own resource redundancy. Both parties calculate their own costs and benefits: The demand side calculates the net benefit of collaboration = the reduction in losses after filling the resource gap (such as the benefit of reduced line loss) - the total cost of resource allocation; the supply side calculates the net benefit of collaboration = the benefit that can be obtained from resource allocation - the increase in its own operating costs (such as the cycle loss of energy storage discharge). The third step is to solve the collaborative scheme of acceptable resource quantity and cost for both parties based on the principles of game theory (i.e., game algorithm). The collaborative scheme acceptable to both parties is taken as the cross-cell collaborative scheme. It needs to satisfy the following conditions at the same time: the net collaborative benefit of the demand side is ≥0 (collaboration is better than no collaboration) and the net collaborative benefit of the supply side is ≥0 (providing resources is profitable). It should be noted that the cross-cell game cooperation mechanism achieves a Nash equilibrium state by means of the Nash equilibrium theory in game theory, in which neither party can increase its payoff by unilaterally adjusting its plan. The core condition for solving Nash equilibrium (optimal strategy for both parties) is that, with the other party's strategy unchanged, neither party can increase its own revenue by unilaterally adjusting its strategy (cooperative power P or unit power price c). The core calculation logic is as follows: The optimal strategy for the demand side (given P, find the optimal c): Net collaborative benefits on the demand side It is negatively correlated with c (the lower the c, the better). The higher the value, the more likely it is to be (c), but c must meet the conditions for supplier participation ( (≥0); Therefore, the optimal pricing strategy for the demand side is "not exceeding its own revenue coefficient". (Refers to the revenue coefficient per unit power, determined by the load importance of cell A according to a linear function) (otherwise) ≤0, no synergistic force): c≤ ; The supplier's optimal strategy (given c, find the optimal P): Net benefits of collaboration with the supplier Taking the derivative with respect to P, let the derivative be 0 (maximization condition): Solving for the optimal power of the supplier. (must meet) ≥0, that is, c≥ Otherwise, suppliers have no incentive to provide resources; in, This is the cost increment factor. The base cost coefficient is determined by the equipment characteristics of B, that is, , The cost of providing P power to the cooperating cell B (i.e., the increased cost of its own operation, such as the cycle loss of energy storage discharge and the line transmission loss, which is positively correlated with P). Combining the optimal strategies of both parties, the equilibrium condition is: price c satisfies ≤c≤ (Ensure that both parties benefit); (Optimal output from the supplier); the final equilibrium point must satisfy the condition that "the net revenue of both parties is not negative": =( -c)× ≥0, ; It should be noted that in practical applications, to reduce computational complexity, c is usually fixed as an intermediate value negotiated by both parties (e.g., ...). (Take the average of cost and benefit), and then substitute it into... The power is calculated using the formula.

[0035] The methods for generating cross-cell collaborative instruction sets and security verification reports include: Based on the cross-cell collaboration scheme, the security of the collaboration scheme is verified layer by layer by calling the local digital twin through a three-layer security boundary of electrical security, communication security and transaction compliance. Specifically, electrical safety verification is the core layer verification. The coordinator of the resource-insufficient cell and the cooperating cell synchronously calls the local digital twin to simulate the full-cycle electrical state of the cross-cell cooperative scheme: input cooperative power (e.g., cell B delivers 200kW to cell A) and duration, and simulate and calculate the tie line current, the voltage of the two nodes, and the load rate of related equipment; if all indicators are within the preset electrical safety boundaries (e.g., load rate 70%≤75%, voltage deviation 3%≤5%), the electrical safety is marked as passed; if there are over-limits (e.g., load rate 80%), the over-limit equipment and deviation value are output (e.g., the load rate of equipment Line-003 exceeds 5%), and fed back to the cooperative scheme formation stage to adjust the cooperative power; Communication security verification is a link-layer verification that verifies the security and stability of cross-cell command transmission: the two parties exchange command samples to be executed, verify the sender's identity through a digital certificate (to prevent forgery), and verify the command checksum after decryption (to ensure it has not been tampered with); test commands are sent multiple times in succession, and the average latency (≤50ms), encryption and decryption time (≤10ms), and packet loss rate (≤0.5%) are statistically analyzed. If all criteria are met, the communication is secure; otherwise, link optimization is triggered (such as switching to a backup encrypted channel). Transaction compliance verification is a protocol-level verification that compares the collaborative scheme with the transaction compliance boundaries: confirming that the transaction duration is within the maximum acceptable transaction time; verifying whether the profit distribution ratio is consistent with the game outcome; if all are met, the transaction is compliant and passes; otherwise, it is returned for adjustment (such as correcting the price to the compliant range). Summarize the multi-layered verification results and generate a security verification summary table. Record the three-layer verification results. Only when all verifications pass can the transaction confirmation process begin. Otherwise, repeat the verification (up to 3 times). If it still fails, the collaboration will be terminated. Both coordinators exchange transaction confirmation messages through an encrypted channel, solidifying core transaction parameters, including resource quantity (e.g., "cell B transmits 150kW of active power to cell A for 2 hours"), division of responsibilities (e.g., "overload of the tie line during transmission shall be borne by both parties according to their respective power ratios"), and emergency clauses (e.g., "in the event of a sudden failure, cell A has the right to disconnect the tie switch first"). The messages are archived after being signed by both coordinators (associated with batch IDs for easy traceability). Based on fixed transaction parameters, a cross-cell collaborative instruction set with collaborative identifiers is generated; Simultaneously compile a security verification report that includes verification results, risk warnings, and instruction associations (correspondence between collaborative instruction sets and verification results) that includes three layers of security boundaries.

[0036] The methods for performance analysis using a multi-dimensional evaluation system to trace and locate the root causes of problems and optimize and iterate throughout the entire process include: Based on the execution effect of cross-cell collaborative instructions and the resource status of each cell, an evaluation system is constructed using multi-dimensional indicators, including collaborative effect (resource gap filling rate, the ratio of actual filled resources to demand gap, which measures the effectiveness of cross-cell collaboration), system performance (voltage qualification rate, the ratio of the duration of voltage within the qualification range to the total evaluation time, which reflects network-side stability), economy (collaboration cost-benefit ratio, the ratio of total collaboration benefit to total collaboration cost), and robustness (strategy iteration adaptation rate, the ratio of the number of adapted strategies in new scenarios to the total number of strategies, which reflects the flexibility of the strategy library). The values ​​of each dimension indicator are calculated according to the constructed evaluation system. The performance is judged by the preset rating standards to obtain a single-dimensional evaluation table and mark the low-rated indicators. Example: Resource gap filling rate ≥90% is excellent, 70%-89% is good, 50%-69% is medium, and <50% is poor. For low-rated indicators (medium or poor), the associated processes are located by mapping the indicators to specific processes. Example: If the resource gap filling rate is low, the collaborative mechanism of the related game (such as too high trust threshold leading to a small number of selectable collaborative cells) and the collaborative scheme generation stage (such as too strict security boundary leading to limited collaborative power) will be affected. If the accuracy of instruction execution is low, it may be related to the local digital twin simulation process (e.g., insufficient simulation accuracy leading to unreasonable instruction parameters) and the data quality and reliability calculation process (e.g., low quality and reliability leading to misjudgment of device status). Based on the root causes, the entire process is optimized and iterated. In particular, to address issues such as low data reliability and mismatched collection frequency, the data collection rules are optimized (increasing the data collection frequency of key equipment (such as interconnection switches and energy storage controllers) and dynamically adjusting the quality reliability threshold (such as reducing the threshold for photovoltaic output data from 70 points to 65 points during periods of stable sunlight to reduce the rejection of valid data)). Based on issues such as low intra-cell coupling and unbalanced load, cell generation parameters are optimized (e.g., tightening the electrical coupling threshold to enhance the electrical correlation of devices within the cell; lowering the load balancing threshold to reserve more computing power to meet collaborative needs). Simultaneously, the dual strategy library and game parameters are optimized. The dual strategy library is supplemented with new strategies for low-rated scenarios (such as the "cross-cell energy storage joint discharge" strategy for the "sudden drop in photovoltaic power + high load" scenario), and the similarity matching weight is adjusted. The game parameters are optimized to reduce the collaboration cost of high-trust cells, incentivize more mutual transactions, and optimize the trust index. Based on the optimization results, a full-process optimization report is generated, including: evaluation conclusions (original system performance shortcomings), root cause analysis (specific sources of key problems, optimization measures (adjustment parameters and expected effects of each stage). Establish a regular iteration mechanism (e.g., weekly evaluation + monthly iteration) to form a closed loop of evaluation-source tracing-optimization, and continuously improve the system's collaborative efficiency and stability.

[0037] Example 2 Please see Figure 3 As shown, for parts not described in detail in this embodiment, please refer to the description in Embodiment 1. A digital twin energy efficiency optimization system for a regional power distribution system is provided, comprising: Trusted processing module: acquires real-time operating data of core equipment in the power distribution system, quantifies data quality through a multi-dimensional evaluation mechanism and attaches corresponding labels, and adjusts data acquisition priority and communication resource allocation in combination with dynamic scenarios to generate a trusted dataset; Cell and Digital Twin Building Module: Based on a trusted dataset, cells are partitioned using real-time electrical coupling, a temporary coordinator is elected, and a lightweight local digital twin and cell communication topology is constructed to generate a dynamic and resilient cell architecture diagram and coordinator configuration table. Autonomous instruction generation module: Based on the dynamic resilience cell architecture diagram and coordinator configuration table, each cell coordinator generates an optimized policy set through a dual-policy library matching mechanism according to a two-factor matching mechanism, and generates an autonomous optimized instruction set through forward simulation of the local digital twin; Cross-cell game collaboration module: Based on the autonomous optimization instruction set and combined with the pre-set degradation plan, when the resources in the cell are insufficient, the cross-cell game collaboration mechanism is triggered. Through multi-layer security boundary verification, mutual assistance transactions are achieved, and a cross-cell collaboration instruction set and security verification report are generated. Full-process performance evaluation module: Based on the execution effect of cross-cell collaborative instruction set and the resource status of each cell, it conducts performance analysis through a multi-dimensional evaluation system, traces and locates the root cause of the problem, and optimizes and iterates the entire process.

[0038] Example 3 This embodiment discloses an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the operation mode of the digital twin energy efficiency optimization system for a regional power distribution system described above.

[0039] Since the electronic device described in this embodiment is the electronic device used to implement the digital twin energy efficiency optimization method for a regional power distribution system according to the embodiments of this application, those skilled in the art can understand the specific implementation method and various variations of the electronic device in this embodiment based on the digital twin energy efficiency optimization method for a regional power distribution system described in the embodiments of this application. Therefore, how the electronic device implements the method in the embodiments of this application will not be described in detail here. Any electronic device used by those skilled in the art to implement the digital twin energy efficiency optimization method for a regional power distribution system according to the embodiments of this application falls within the scope of protection of this application.

[0040] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters and thresholds in the formulas are set by those skilled in the art according to the actual situation.

[0041] The above description is merely a preferred embodiment of the present invention, and the scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for users of ordinary technical skills, any improvements and modifications made without departing from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A digital twin energy efficiency optimization method for a regional power distribution system, characterized in that, include: S1: Acquire real-time operating data of core equipment in the power distribution system, quantify data quality through a multi-dimensional evaluation mechanism and attach corresponding labels, and adjust data acquisition priority and communication resource allocation in combination with dynamic scenarios to generate a reliable dataset; S2: Based on a trusted dataset, cells are partitioned using real-time electrical coupling, a temporary coordinator is elected, and a lightweight local digital twin and cell-to-cell communication topology are constructed to generate a dynamic resilient cell architecture diagram and a coordinator configuration table. S3: Based on the dynamic resilience cell architecture diagram and coordinator configuration table, each cell coordinator generates an optimization strategy set through a dual-strategy library matching mechanism according to the dual-factor matching mechanism, and generates an autonomous optimization instruction set through local digital twin forward simulation; S4: Based on the autonomous optimization instruction set and combined with the pre-set degradation plan, when the resources within the cell are insufficient, the cross-cell game collaboration mechanism is triggered. Through multi-layer security boundary verification, mutual assistance transactions are achieved, and a cross-cell collaboration instruction set and security verification report are generated. S5: Based on the execution effect of cross-cell collaborative instruction set and the resource status of each cell, performance analysis is conducted through a multi-dimensional evaluation system to trace and locate the root cause of the problem and optimize and iterate the entire process.

2. The digital twin energy efficiency optimization method for a regional power distribution system according to claim 1, characterized in that, The method of quantifying data quality and attaching corresponding labels through a multi-dimensional evaluation mechanism includes: Collect real-time operating data of four types of core equipment in the power distribution system: source-side equipment, grid-side equipment, load-side equipment, and storage-side equipment; Data quality is quantified through a five-dimensional evaluation mechanism that considers equipment accuracy, transmission latency, physical verification, historical stability, and multi-source consistency, thereby obtaining the reliability of each data point. Then, based on the quality and reliability, a dynamic reliability label is attached to each piece of data and associated with the corresponding device identifier, and the data is integrated into a multi-source dataset.

3. The digital twin energy efficiency optimization method for a regional power distribution system according to claim 2, characterized in that, The methods for generating the trusted dataset include: Based on a multi-source dataset with dynamic credibility labels, the current running scenario is identified by a preset dynamic scenario library and a multi-source triggering mechanism, and the scenario priority is determined by combining the quality and credibility of the data. Adjust the data collection priority of each device based on scenario priority; The dynamic scene library is mapped to the pre-set communication resource pool, and communication resources for each device are allocated according to the mapping relationship. Simultaneously, valid data and abnormal data are divided according to quality and reliability. All valid data are integrated to generate a reliable dataset with scene association. Abnormal data is isolated and the reasons are labeled to generate an abnormal isolation list.

4. The digital twin energy efficiency optimization method for a regional power distribution system according to claim 3, characterized in that, The methods for electing a temporary coordinator include: Based on the trusted dataset and the anomaly isolation list, devices associated with abnormal data are removed, and the remaining devices are marked as reliable devices; Based on the step-by-step verification rules of electrical coupling → communication quality → load balancing, the reliable equipment in adjacent transformer areas is divided into dynamically resilient cells, forming a cell cluster. Candidate devices with data processing and command issuance capabilities are selected from each cell. An independent election is conducted within each cell according to an election scoring mechanism to obtain the optimal candidate device as the temporary coordinator for the corresponding cell.

5. The digital twin energy efficiency optimization method for a regional power distribution system according to claim 4, characterized in that, The methods for constructing a lightweight local digital twin and inter-cell communication topology, and generating a dynamic, resilient cell architecture diagram and coordinator configuration table include: Based on the cell cluster and coordinator election results, all devices in the cell cluster are divided into high-confidence devices and low-confidence devices according to quality reliability. Each cell coordinator extracts highly reliable device parameters and builds a lightweight local digital twin that retains core functions; Simultaneously, the inter-cell communication topology is constructed according to the structure of the core layer and the edge layer, and the communication links are optimized to generate a dynamic resilient cell architecture diagram that marks cell boundaries, devices, and communication relationships; the performance and communication parameters of each coordinator are recorded in a structured manner in sync to generate a coordinator configuration table.

6. The digital twin energy efficiency optimization method for a regional power distribution system according to claim 5, characterized in that, The optimization strategy set is generated in the following ways: Based on the dynamic resilience cell architecture diagram and coordinator configuration table, the pre-built dual strategy library for normal operation and stress testing is invoked. The coordinator extracts the fault type features, impact range features, and resource constraint features of the current scenario from the multi-source dataset, quantifies them, calculates the similarity with the scenarios in the dual-strategy library, and matches them to select scenarios that meet the similarity criteria. Simultaneously, the coordinator retrieves the historical decision records of the models constructed by various algorithms within the local digital twin to obtain the accuracy assessment of the model's credibility. For strategies targeting scenarios that meet similarity criteria, scenario similarity and model credibility are used as two factors. A two-factor matching mechanism is used to select suitable strategy combinations to form an optimized strategy set.

7. The digital twin energy efficiency optimization method for a regional power distribution system according to claim 6, characterized in that, The methods for generating the autonomous optimization instruction set include: The optimization strategy set is broken down into device-level operations and mapped to a local digital twin; The coordinator calls the local digital twin to perform forward simulation at a preset time step, dynamically monitoring the constraint satisfaction during the simulation process; For a strategy that is fully simulated, the feasibility of the strategy is evaluated according to the multi-dimensional credibility evaluation rules. Strategies that meet the criteria are judged as feasible strategies, and strategies that do not meet the criteria are revised and re-evaluated through a second simulation. The final feasible strategy is converted into device-executable instructions and encapsulated into a standardized, categorized, and ordered autonomous optimization instruction set.

8. The digital twin energy efficiency optimization method for a regional power distribution system according to claim 7, characterized in that, The methods for triggering the cross-cell game cooperation mechanism when resources within a cell are insufficient include: By comparing the resource requirements of the autonomous optimization instruction set with the real-time available resources within the cell, the resource gap is calculated to determine whether the resources within the cell are sufficient: If sufficient, maintain the original autonomous optimization instruction set; if insufficient, execute the pre-set multi-level degradation plan and recalculate the resource gap; if still insufficient, trigger the cross-cell game collaboration mechanism, and select high-trust neighboring cells as collaborative cells based on the inter-cell communication topology. Resource-deficient cells and collaborative cells interact regarding resource needs and costs. Based on game theory, a collaborative solution is found that allows both parties to accept a certain amount of resources and costs, thus forming a cross-cell collaborative solution.

9. The digital twin energy efficiency optimization method for a regional power distribution system according to claim 8, characterized in that, The generation methods for the cross-cell collaborative instruction set and security verification report include: Based on the cross-cell collaboration scheme, the security of the collaboration scheme is verified layer by layer by calling the local digital twin through a three-layer security boundary of electrical security, communication security and transaction compliance. Once all verifications are passed, the mutual assistance transaction is confirmed and the transaction parameters are fixed. A cross-cell collaborative instruction set with a collaborative identifier is generated, and a security verification report containing verification results and risk warnings is compiled simultaneously.

10. The digital twin energy efficiency optimization method for a regional power distribution system according to claim 9, characterized in that, The method of using a multi-dimensional evaluation system to conduct performance analysis, trace and locate the root causes of problems, and optimize and iterate throughout the entire process includes: Based on the execution effect of cross-cell collaborative instructions and the resource status of each cell, an evaluation system is constructed through four dimensions of indicators: collaborative effect, system performance, economy and robustness. The system analyzes the overall performance and identifies the root causes of low-rated indicators. A full-process optimization report is generated for the entire process of root cause optimization iteration data collection rules, cell generation parameters, dual-strategy library and game parameters, forming a closed loop.

Images