Power simulation method and device for micro-grid system

Abstract

The invention discloses a power simulation method and device for a micro-grid system, and relates to the technical field of micro-grid systems, and the method comprises the steps: determining an energy storage-consumption coefficient of the micro-grid system based on a plurality of dynamic energy storage models, the current power consumption data of an island and corresponding power consumption equipment; the method comprises the following steps: determining a plurality of key power-off factors based on a previous power-off event of an island, and determining a power simulation event of the micro-grid system according to the plurality of key power-off factors, an energy storage-consumption coefficient of the micro-grid system and a plurality of dynamic energy storage models. An island power management and control event is determined according to a plurality of island power management and control measures, a micro-grid system and a plurality of energy storage paths, and a plurality of sub power management and control projects are determined based on identification of the island power management and control event. The autonomous power management and control system is determined based on the cooperative relationship among the plurality of sub-power management and control projects, the abnormal power data of the micro-grid system and the plurality of dynamic energy storage models, and the accuracy of the autonomous power management and control system is improved.

Description

Technical Field

[0001] This invention relates to the technical field of microgrid systems, and more particularly to a power simulation method and apparatus for microgrid systems. Background Technology

[0002] Microgrid systems are widely used in people's lives. In this context, isolated islands with corresponding facilities are gradually being developed. Microgrid systems can be applied to these islands. In existing technologies, current power consumption data of the islands is collected, and the corresponding power consumption level is determined based on the identification of the current power consumption data. Based on the power consumption level and the island, corresponding power management projects are determined. However, the impact of abnormal power data of the microgrid system and multiple dynamic energy storage models is ignored, resulting in low accuracy of the existing power management system. Summary of the Invention

[0003] The purpose of this invention is to overcome the shortcomings of the prior art. This invention provides a power simulation method and apparatus for microgrid systems.

[0004] This invention provides a power simulation method for a microgrid system, comprising: When the microgrid system is located on an island, the current power grid system of the microgrid system is determined based on multiple working data of the microgrid system, and multiple energy storage paths are determined based on the current power grid system, the construction engineering map of the island, and the corresponding multiple energy storage nodes. Based on the island's weather data, wind data, and multiple energy storage paths, several dynamic energy storage models with different dimensions are determined. Based on these dynamic energy storage models, the island's current power consumption data, and the corresponding power consumption equipment, the energy storage-consumption coefficient of the microgrid system is determined. Based on past power outage events on the island, several key power outage factors were identified. Based on these key power outage factors, the energy storage-consumption coefficient of the microgrid system, and several dynamic energy storage models, power simulation events for the microgrid system were determined. Based on the identification of the power simulation event, multiple sub-abnormal power scenarios are determined, and corresponding island power control measures are determined based on each sub-abnormal power scenario, multiple dynamic energy storage models, and the current construction progress of the island. Based on multiple island power management measures, microgrid systems, and multiple energy storage pathways, island power management events are identified. Based on the identification of island power management events, multiple sub-power management projects are identified. Based on the collaborative relationships of multiple sub-power management projects, abnormal power data of microgrid systems, and multiple dynamic energy storage models, an autonomous power management system for microgrid systems located on islands is determined.

[0005] This invention provides a power simulation device for a microgrid system, which is applied to the aforementioned power simulation method for microgrid systems. The power simulation device for the microgrid system includes: The energy storage path module is used to determine the current grid system of the microgrid system based on multiple working data of the microgrid system when the microgrid system is located on an island, and to determine multiple energy storage paths based on the current grid system, the construction engineering map of the island, and the corresponding multiple energy storage nodes. The energy storage-consumption coefficient module is used to determine multiple dynamic energy storage models of different dimensions based on the island's weather data, wind data, and multiple energy storage paths. Based on multiple dynamic energy storage models, the island's current power consumption data, and corresponding power consumption equipment, the energy storage-consumption coefficient of the microgrid system is determined. The power simulation event module is used to identify multiple key power outage factors based on past power outage events on the island, and to determine the power simulation events of the microgrid system based on multiple key power outage factors, the energy storage-consumption coefficient of the microgrid system, and multiple dynamic energy storage models. The island power management and control measures module is used to identify multiple sub-abnormal power scenarios based on the identification of the power simulation event, and to determine the corresponding island power management and control measures based on each sub-abnormal power scenario, multiple dynamic energy storage models and the current construction progress of the island. The autonomous power management system module is used to determine island power management events based on multiple island power management measures, microgrid systems, and multiple energy storage paths. Based on the identification of island power management events, multiple sub-power management projects are determined. Based on the collaborative relationship between multiple sub-power management projects, abnormal power data of microgrid systems, and multiple dynamic energy storage models, the autonomous power management system of microgrid systems located on islands is determined.

[0006] Compared with the prior art, the beneficial effects of the present invention are: In this embodiment of the invention, when the microgrid system is located on an island, the current power grid structure of the microgrid system is determined based on multiple operating data of the microgrid system. Multiple energy storage paths are determined based on the current power grid structure, the island's construction engineering diagram, and the corresponding multiple energy storage nodes. Multiple dynamic energy storage models of different dimensions are determined based on the island's weather data, wind data, and multiple energy storage paths. The energy storage-consumption coefficient of the microgrid system is determined based on the multiple dynamic energy storage models, the island's current power consumption data, and the corresponding power-consuming equipment. Multiple key power outage factors are determined based on past power outage events on the island. Power simulation events of the microgrid system are determined based on these key power outage factors, the energy storage-consumption coefficient of the microgrid system, and the multiple dynamic energy storage models. The introduction of the energy storage-consumption coefficient of the microgrid system allows for multi-dimensional control of energy storage and consumption, incorporating a holistic consideration of multiple key power outage factors, the energy storage-consumption coefficient of the microgrid system, and multiple dynamic energy storage models, thereby improving the accuracy of power simulation events for the microgrid system.

[0007] Therefore, based on the identification of the power simulation event, multiple sub-abnormal power scenarios are determined. Corresponding island power management measures are then determined based on each sub-abnormal power scenario, multiple dynamic energy storage models, and the current construction progress of the island. Island power management events are determined based on these multiple island power management measures, the microgrid system, and multiple energy storage paths. Multiple sub-power management projects are then identified based on the identification of these island power management events. Finally, an autonomous power management system for the microgrid system located on the island is determined based on the collaborative relationships among these multiple sub-power management projects, the abnormal power data of the microgrid system, and multiple dynamic energy storage models. The introduction of island power management events further manages these multiple sub-power management projects, achieving a holistic consideration of the collaborative relationships among these projects, the abnormal power data of the microgrid system, and multiple dynamic energy storage models, thus improving the accuracy of the autonomous power management system. Attached Figure Description

[0008] Figure 1 This is a flowchart illustrating the power simulation method for a microgrid system in an embodiment of the present invention. Figure 2 This is a flowchart illustrating step S11 in the power simulation method for a microgrid system according to an embodiment of the present invention. Figure 3 This is a flowchart illustrating step S12 in the power simulation method for a microgrid system according to an embodiment of the present invention. Figure 4 This is a flowchart illustrating step S13 in the power simulation method for a microgrid system according to an embodiment of the present invention. Figure 5 This is a flowchart illustrating step S14 in the power simulation method for a microgrid system according to an embodiment of the present invention. Figure 6This is a flowchart illustrating step S15 in the power simulation method for a microgrid system according to an embodiment of the present invention. Figure 7 This is a schematic diagram of the structural composition of the power simulation device for a microgrid system in an embodiment of the present invention. Detailed Implementation

[0009] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0010] Please see Figures 1 to 7 A power simulation method for microgrid systems, applied to microgrid scenarios; the power simulation method for microgrid systems includes: Step S11: When the microgrid system is located on an island, determine the current power grid system of the microgrid system based on multiple working data of the microgrid system, and determine multiple energy storage paths according to the current power grid system, the construction engineering map of the island, and the corresponding multiple energy storage nodes; Step S12: Determine multiple dynamic energy storage models of different dimensions based on the island's weather data, wind data, and multiple energy storage paths. Determine the energy storage-consumption coefficient of the microgrid system based on the multiple dynamic energy storage models, the island's current power consumption data, and the corresponding power consumption equipment. Step S13: Based on past power outage events on the island, identify multiple key power outage factors, and determine the power simulation events of the microgrid system based on multiple key power outage factors, the energy storage-consumption coefficient of the microgrid system, and multiple dynamic energy storage models; Step S14: Based on the identification of the power simulation event, determine multiple sub-abnormal power scenarios, and determine the corresponding island power control measures based on each sub-abnormal power scenario, multiple dynamic energy storage models, and the current construction progress of the island; Step S15: Determine island power management events based on multiple island power management measures, microgrid systems, and multiple energy storage paths; determine multiple sub-power management projects based on the identification of island power management events; and determine the autonomous power management system of the microgrid system located on the island based on the collaborative relationship between multiple sub-power management projects, abnormal power data of the microgrid system, and multiple dynamic energy storage models.

[0011] refer to Figure 2 In step S11, the specific steps are as follows: S111: Configure a corresponding microgrid system on the island, and feed back the electrical energy output by the microgrid system to the construction project on the island. Monitor the microgrid system in real time and collect multiple working data of the microgrid system, which cover the scene dimensions of ocean, sun and wind fields. S112: Determine the current power grid system of the microgrid system based on multiple working data of the microgrid system, the corresponding scenario dimensions, and the current working mode of the microgrid system; collect the construction engineering map of the island, determine the corresponding power consumption path based on the construction engineering map and the cable distribution map of the island, and determine multiple energy storage paths based on the power consumption path, the current power grid system, and the corresponding multiple energy storage nodes.

[0012] In the embodiments of this application, a corresponding microgrid system is configured on the island, and the electrical energy output by the microgrid system is fed back to the island's construction project. The microgrid system is monitored in real time, and multiple working data of the microgrid system are collected. These multiple working data cover the scene dimensions of ocean, sun, and wind field.

[0013] At this point, a cyber-physical coupling (CPS) relationship is established between the microgrid physical system and the island load entities. The power output of the microgrid is no longer a unidirectional energy flow, but a two-way information interaction is formed with the power consumption behavior of the construction projects on the island through the Advanced Measurement System (AMI) and Power Monitoring Unit (PMU). This requires the deployment of high-precision monitoring equipment at the microgrid's grid connection point (PCC) and key load feeders to capture power quality parameters (such as voltage harmonics, frequency deviation, and three-phase imbalance) and power flow data in real time.

[0014] The collected data exhibits significant heterogeneity, including: Electrical quantities: from SCADA systems, PMUs, and smart meters, such as voltage, current, active / reactive power, frequency, and phase angle. This type of data is characterized by high sampling rates and accurate timestamps; Status quantities: from the device's own controllers (such as wind turbine converters, photovoltaic inverters, and BMS), such as switch status (on / off), device operating mode (charging / discharging / standby), alarm codes, and temperature; Environmental quantities: from professional meteorological stations and marine monitoring buoys, such as Global Horizontal Irradiance (GHI), Direct Normal Irradiance (DNI), wind speed, wind direction, air density, seawater temperature, and salinity.

[0015] To integrate heterogeneous data, time alignment and spatial correlation are introduced. Time synchronization: All acquisition devices, especially PMU and IED, must ensure that all data streams have a unified high-precision time base (error in the microsecond range), which is a prerequisite for subsequent causal analysis and dynamic simulation. Data preprocessing: This is performed at the edge or in the cloud and mainly includes: outlier removal (such as using the 3σ rule or isolated forest algorithm to identify and remove sensor fault data), data interpolation (performing linear or spline interpolation on data lost due to short-term communication interruptions), and data smoothing / filtering (such as using a Kalman filter to reduce noise in wind speed data and extract the true trend).

[0016] By introducing scene-dimensional labeling, the system uses a rule engine to transform massive amounts of raw data into structured information with clear context. When each dataset is accompanied by scene labels such as "strong wind weather + high solar irradiance + equipment operation", the system has context awareness capabilities. This not only greatly enhances the value of the data, but also facilitates subsequent intelligent decision-making, pattern recognition, and predictive analysis.

[0017] Furthermore, based on multiple working data of the microgrid system, corresponding scenario dimensions, and the current working mode of the microgrid system, the current power grid system is determined; construction engineering drawings of the island are collected, and the corresponding power consumption paths are determined based on the construction engineering drawings and cable distribution maps of the island. Multiple energy storage paths are determined based on the power consumption paths, the current power grid system, and the corresponding multiple energy storage nodes, which takes into account the overall consideration of power consumption paths, the current power grid system, and the corresponding multiple energy storage nodes, ensuring the accuracy of multiple energy storage paths.

[0018] At this point, a topology diagram reflecting the real-time operating status of the microgrid is constructed; it is not a static wiring diagram, but a weighted directed graph, where the nodes are buses, generators, and loads, and the edges are switches, lines, and transformers; its core is the combination of topology analysis and state estimation.

[0019] Based on the collected switch status quantities (opening and closing positions of circuit breakers and disconnectors), the system uses depth-first search (DFS) or breadth-first search (BFS) algorithms to determine network connectivity and divide the network into different electrical islands. For example, it determines whether the system operates on a single island or is split into multiple isolated islands. After determining the topology, the system uses collected voltage, current, power, and other measurement data to calculate the voltage amplitude and phase angle of all buses in the entire network using state estimation algorithms such as weighted least squares (WLS). This not only filters out measurement noise but also fills in measurement blind spots, obtaining a consistent and most reliable operating state for the entire network. Combined with scene dimension tags, the system can identify the current operating mode, such as "diesel generator standby mode," "wind and solar maximum output mode," and "energy storage frequency regulation mode." This mode will be used as a global attribute of the topology map.

[0020] The system extracts electrical connections from geographic information to construct a physical, static power supply routing network. This primarily relies on Geographic Information Systems (GIS) and graph database technologies. Inputs include construction drawings of the island (CAD / DWG format) and cable distribution maps (GIS data containing cable type, length, and laying path). The system parses the drawings using CADAPI or GIS tools to identify key entities: substations, distribution rooms, cable trenches, cable wells, and distribution boxes in various buildings. These entities are used as nodes, and cables as edges, and imported into graph databases such as Neo4j. Each edge (cable) has the following attributes: type (determining impedance), length, rated current carrying capacity, and laying method (determining heat dissipation). In this way, a complete static power supply routing network with fully defined physical attributes is constructed.

[0021] By overlaying the dynamic power grid system with static consumption paths and introducing energy storage nodes, a path search problem with practical scheduling value is planned. This is a multi-constraint, multi-objective problem. The system merges and aligns the generated dynamic topology map with the generated static physical path map. Simultaneously, the planning of energy storage paths must meet a series of real-time constraints: electrical constraints: the current carrying capacity of the lines cannot exceed limits, and the node voltage cannot exceed boundaries; equipment constraints: the charging and discharging power of the energy storage cannot exceed the rated value of its PCS (Power Conversion System), and the SOC must be within the safe range; topological constraints: the path must be electrically connected. An improved Dijkstra algorithm is used for path search. Here, the "path cost" is no longer a simple distance but a comprehensive cost function, which can include: line loss cost, energy storage scheduling cost (such as penalties for SOC deviation from ideal values), and path reliability (such as whether the area traversed by the path is susceptible to typhoons).

[0022] refer to Figure 3 In step S12, the specific steps are as follows: S121: Real-time monitoring of islands, and collection of island weather and wind data based on multiple island detections; determination of corresponding island data combinations based on island weather data, wind data and detection time, and determination of multiple dynamic energy storage models of different dimensions based on each island data combination and multiple energy storage paths; S122: Collect current power consumption data of the island, identify multiple sub-power consumption events based on the identification of the current power consumption data of the island, and determine the combination of consumption content according to the multiple sub-power consumption events and the corresponding power consuming equipment; S123: Determine the energy storage content combination based on multiple dynamic energy storage models and the current power consumption data of the island, and determine the energy storage-consumption coefficient of the microgrid system based on the consumption content combination, the energy storage content combination, and the energy storage effect coefficient of the microgrid system.

[0023] In the embodiments of this application, islands are monitored in real time, and weather and wind data of islands are collected based on multiple detections of islands; corresponding island data combinations are determined based on island weather data, wind data and detection time, and multiple dynamic energy storage models of different dimensions are determined based on each island data combination and multiple energy storage paths, which is compatible with the overall consideration of each island data combination and multiple energy storage paths, and ensures the accuracy of multiple dynamic energy storage models of different dimensions.

[0024] At this point, the goal of establishing a comprehensive and precise sensing system is to capture every key environmental variable affecting energy production on the island. This involves more than just deploying a few sensors; it means building a three-dimensional monitoring network covering the sky, land, and sea. On the ground, integrated automatic weather stations collect basic meteorological parameters such as temperature, humidity, and air pressure. Through high-precision total radiation meters and direct radiation meters, they record the core data that determines photovoltaic output—global horizontal irradiance (GHI) and direct normal irradiance (DNI)—at a frequency of 1Hz. In the high altitude, in addition to traditional wind measuring equipment, the most crucial element is the deployment of phased array lidar (LiDAR) at the height of the wind turbine hub. In the island's practical application, this radar can scan every 30 seconds, generating a wind speed profile within a 3-kilometer radius ahead, enabling advanced wind resource prediction—something traditional methods cannot achieve. All this data from different devices and at different frequencies is collected in real time through an encrypted 5G private network to the microgrid energy management system (MEMS) platform, providing a highly reliable raw data stream for subsequent processing.

[0025] All devices are synchronized with GPS at the nanosecond level to ensure precise alignment of data on the timeline. The system performs deep cleaning and calibration of the data, such as using a Kalman filter to smooth the wind speed data from the lidar, removing high-frequency noise, and correcting for solar occlusion in the radiation data. After this series of processing steps, the system packages all environmental elements into a high-dimensional feature vector at each unified time reference (e.g., every 5 seconds). This high-dimensional feature vector is what you call the "island data combination." This structured data object not only includes the current GHI, wind speed, and temperature, but also incorporates ultra-short-term forecast information from LiDAR, providing rich and accurate input for subsequent dynamic modeling.

[0026] Based on the combination of data from various islands and multiple energy storage paths, multiple dynamic energy storage models with different dimensions are determined to form a flexible model cluster, rather than a single universal model. This process unfolds from three dimensions: time, scenario, and path. In terms of time, the system will activate the corresponding model according to different prediction cycles. The ultra-short-term model (second to minute level) uses ARMA or Kalman filtering to guide energy storage to participate in primary frequency regulation. The short-term model (minute to hour level) combines numerical weather prediction (NWP) and machine learning algorithms for peak shaving and valley filling and economic dispatch.

[0027] In terms of scenarios, the system uses clustering algorithms to identify typical weather patterns from massive historical "island data combinations," such as "sunny and windless summer" or "typhoon passing through." When a new data combination arrives, the system immediately matches the corresponding scenario and activates the scenario's dedicated, offline-optimized energy storage operation strategy library. In terms of paths, the model is further refined to customize control strategies for energy storage paths connecting different power sources and loads. For example, the energy storage model connecting to the southern photovoltaic power station will use PV control mode to stabilize the grid connection voltage based on minor changes in GHI.

[0028] Furthermore, current power consumption data of the island is collected, and multiple sub-power consumption events are identified based on the identification of the current power consumption data of the island. The combination of consumption content is determined according to the multiple sub-power consumption events and the corresponding power consuming equipment. This takes into account the overall consideration of multiple sub-power consumption events and the corresponding power consuming equipment, and ensures the accuracy of the combination of consumption content.

[0029] At this point, the data source mainly relies on smart meters (AMI) and power quality monitoring devices (PQMF). Unlike traditional meters used for billing, meters used for event monitoring need to support higher data sampling rates. In order to capture the rapid dynamic process of events such as motor starting and elevator lifting, the sampling rate needs to reach at least 4 sampling points per second (4Hz) or higher. For scenarios that need to analyze power quality issues such as harmonics, even kHz-level sampling is required.

[0030] Monitoring points should be deployed in a hierarchical manner; a master meter should be installed at the main incoming line of the island to capture macroscopic changes in the total load of the entire island; regional meters should be installed on the power distribution trunk lines in major load areas (such as residential areas, docks, and hotels); and dedicated monitoring points should be installed on the feeders of key high-power equipment (such as seawater desalination systems and quay cranes). This hierarchical deployment provides a basis for subsequent incident tracing.

[0031] Based on the identification of current power consumption data of the island, multiple sub-power consumption events are determined. A power consumption event typically has three basic characteristics: a start point, an end point, and morphological characteristics during the event (such as step, ramp, pulse, periodic oscillation). At this time, the first derivative (gradient) of the power time series is calculated. When the absolute value of the gradient exceeds a preset threshold, it is considered to be the start or end point of an event. This method is simple and efficient and suitable for detecting step events. A fixed-size window is used to slide on the time series, and the statistical characteristics of the data within the window (such as mean and variance) are calculated. When the feature value changes significantly, event detection is triggered. For more complex events, Hidden Markov Models (HMM) or Long Short-Term Memory Networks (LSTM) can be used. By training on historical event data, the model can learn to recognize complex patterns of different types of events, even if these patterns are masked by noise.

[0032] The system compares the detected event characteristics (such as location, power change, and duration) with information in the equipment archive. The equipment archive stores information such as the rated power, typical operating cycle, and geographical location of all electrical equipment on the island. This is an advanced technology in the absence of a dedicated equipment table. It infers the start-up and shutdown status of specific internal equipment by analyzing steady-state and transient characteristics such as current harmonics and voltage sags in the master table or regional table. Once the equipment corresponding to the event is determined, the system generates a structured data object, namely the "consumption content combination," which describes in detail the electrical characteristics and physical meaning of the event.

[0033] Therefore, the combination of energy storage contents is determined based on multiple dynamic energy storage models and the current power consumption data of the island. The energy storage-consumption coefficient of the microgrid system is determined based on the combination of consumption contents, the combination of energy storage contents, and the energy storage effect coefficient of the microgrid system. This approach takes into account the overall consideration of the combination of consumption contents, the combination of energy storage contents, and the energy storage effect coefficient of the microgrid system, thus ensuring the accuracy of the energy storage-consumption coefficient of the microgrid system.

[0034] At this point, the system will transform the abstract model instructions output by S121 into a specific and measurable "energy storage content combination". It will deconstruct the real-time operating capability and potential of the energy storage system at the current moment into four core dimensions: power capability (how much power can it charge and discharge), energy capability (how long can it last), response capability (how fast can it react), and quality capability (how well can it regulate power quality). For example, at 14:30:00 on the island, the system will generate a capability vector containing specific values ​​such as maximum discharge power of 900kW, sustainability for 2 hours, and response time of 80 milliseconds, based on the SOC, SOH, and PCS status reported by the BMS.

[0035] The system introduces a dynamic "energy storage performance coefficient" to revise this capacity list because the energy storage system is not always in optimal condition. It comprehensively evaluates whether the SOC deviates from the ideal range, the degree of battery aging (SOH), the current thermal state, and recent operating history. In the case of the island, since the SOC, SOH, and temperature are all in excellent condition, the system calculates a performance coefficient as high as 0.995.

[0036] The supply-side "capacity list" and the demand-side "consumption combination" are now facing off. The system will perform fine matching from multiple dimensions such as power, energy, and time. It will compare the maximum discharge power of the energy storage with the sum of the peak power of all load events to assess whether the energy is sufficient to cover the total demand, and even analyze whether the available duration of the energy storage can cover the duration of critical events. In the case of the island, the system found that the 900kW power and 1800kWh energy of the energy storage are sufficient to meet the total demand of the dock crane (280kW) and the upcoming evening peak of residents (400kW), with margins in both power and energy. However, the 2-hour battery life of the energy storage cannot independently cover the evening peak of up to 3 hours, exposing the risk in the time dimension.

[0037] The system uses a weighted algorithm to merge the matching results of these dimensions (including negative time margin) into an original comprehensive scalar value, and then corrects it with the previously calculated "energy storage effect coefficient"; the final value of 0.068 is sent to S13, which precisely tells the subsequent risk assessment module that the current energy storage system of the island microgrid, although it still has a power and energy surplus when dealing with the upcoming load peak, cannot independently support the entire period, and there is a clear time gap risk. It is necessary to immediately start the backup power supply or implement demand-side management.

[0038] Specifically, for the island, the energy storage configuration is: P_discharge_max (maximum discharge power) = 900kW, E_duration (continuous discharge duration) = 2h; the consumption configuration is: E_Dock_1430 event from S122, steady_state_power (steady-state power) = 280kW; at the same time, the system identifies that the evening peak in the residential area is about to begin, and the predicted load is 400kW; the total peak power demand is 280 + 400 = 680kW; the total energy demand (1 hour of dock operations + 3 hours of evening peak in residential areas) is approximately (280*1) + (400*3) = 1480kWh.

[0039] Dimensional matching: Power matching: 900kW > 680kW, power is sufficient, power margin is (900-680) / 900=24.4%; Energy matching: the available energy of energy storage is 900kW×2h=1800kWh; 1800kWh > 1480kWh, energy is also sufficient, energy margin is (1800-1480) / 1800=17.8%; Time matching: the 2-hour battery life of energy storage cannot fully cover the 3-hour demand of residents during the evening peak, there is a 1-hour energy gap, and the time margin is negative.

[0040] Final coefficient calculation: The system uses a weighted average method to calculate a comprehensive scalar value (assuming a power weight of 0.4, an energy weight of 0.4, and a time weight of 0.2): Original coefficient = (0.244 × 0.4) + (0.178 × 0.4) + (-0.5 × 0.2) = 0.0976 + 0.0712 - 0.1 = 0.0688; This positive but low value indicates that the system is basically balanced, but there are hidden dangers; it is corrected using the energy storage effect coefficient: Final energy storage-consumption coefficient = 0.0688 × 0.995 ≈ 0.068; This final coefficient of 0.068 is sent to S13; it accurately tells the subsequent risk assessment module that the current energy storage system of the island microgrid has a power and energy margin when dealing with the upcoming load peak, but it cannot independently support the entire evening peak, and there is a risk in the time dimension. It is necessary to start the backup power supply or implement demand-side management.

[0041] refer to Figure 4 In step S13, the specific steps are as follows: S131: Based on the microgrid system and island database, identify past power outage events on the islands, identify multiple power outage scenarios based on the identification of past power outage events on the islands, and identify corresponding key power outage factors based on the detection of each power outage scenario, so as to collect multiple key power outage factors. S132: Collect multiple dynamic energy storage models, determine the first level of power simulation content based on the multiple dynamic energy storage models, the corresponding key power outage factors, and the overall shape of the island; determine the second level of power simulation content based on the multiple dynamic energy storage models, the energy storage-consumption coefficient of the microgrid system, and the overall shape of the island. S133: Collect the scenario influence coefficients preset for the island, determine the power simulation events of the microgrid system based on the first level of power simulation content, the second level of power simulation content and the scenario influence coefficients preset for the island. The power simulation events present the virtual power data of the island under the control of the microgrid system, and optimize the power decision content of the microgrid system in different scenarios based on the virtual power data.

[0042] In the embodiments of this application, past power outage events of the islands are determined based on the microgrid system and the island database. Multiple power outage scenarios are determined based on the identification of past power outage events of the islands. Corresponding key power outage factors are determined based on the detection of each power outage scenario. This process collects multiple key power outage factors, which takes into account the overall consideration of the detection of each power outage scenario and ensures the accuracy of the corresponding key power outage factors.

[0043] At this point, the system uses an ETL (Extract-Transform-Load) tool to align and merge data from the SCADA system, protection information substation, equipment management system (EMS), and work order system. The key to alignment is a unified timestamp. The system uses a Named Entity Recognition (NER) model to extract key entities from the text descriptions of the operation and maintenance logs. The merged time-series data and the structured information extracted by NLP are combined to form a power outage event object.

[0044] Specifically, the island database contains a power outage record from 2024, with the text description: "At 20:30 on August 15, Typhoon Haiyan passed through. The 10kV harbor line swayed due to excessive wind speed, causing a phase-to-phase short circuit. The overcurrent protection stage I tripped, resulting in a power outage at the dock and the western residential area. Power was successfully restored at 21:45." The system reconstructed this into the following digital event object through NLP and SCADA data fusion.

[0045] The system extracts a set of feature vectors from each power outage event object to describe its essence, including numerical features (such as duration and maximum wind speed) and categorical features (such as cause type). The latter is converted into numerical values ​​through one-hot encoding. Unsupervised learning algorithms such as K-Means and DBSCAN are used to cluster the feature vectors. The algorithm automatically groups events with similar features into one category, and each category represents a typical power outage scenario. Each clustering result is analyzed, and its typical features are described in concise language.

[0046] Specifically, the system clustered 50 power outage events on the island over the past 5 years, resulting in 4 clusters and generating the following scenario profiles: Cluster 1 ("Typhoon weather external damage"): Contains 12 events; Profile: {Average duration: 120 minutes, Average maximum wind speed: >30m / s, Cause: Line fault / tower tilt, Percentage: 24%}; Cluster 2 ("Evening peak energy storage depletion"): Contains 8 events; Profile: {Average duration: 45 minutes, Average wind speed: <5m / s, Average SOC: <20%, Cause: Low voltage / low frequency load shedding, Percentage: 16%}; Cluster 3 ("Equipment protection malfunction / failure"): Contains 6 events; Profile: {Average duration: 25 minutes, Main cause: Protection device malfunction / transformer failure, Percentage: 12%}; Cluster 4 ("Planned maintenance"): Contains 24 events; Profile: {Average duration: 180 minutes, Cause: Planned power outage, Percentage: 48%}.

[0047] For each scenario, the system backtracks all high-frequency data from a period prior to its occurrence (e.g., the previous 2 hours); uses association rule mining methods such as the Apriori algorithm to search for frequently occurring itemsets in the time series data; to eliminate spurious correlations, the system also employs more advanced causal inference methods such as the Granger Causality Test to verify whether a genuine causal chain exists between these conditions; the combination of preconditions with high confidence verified through causal inference is identified as the "key power outage factor" for that scenario, which is a structured rule set that can be monitored in real time.

[0048] Specifically, the system focuses on the scenario of "evening peak energy storage depletion" and retrieves all data from the three hours prior to the occurrence of eight events in this scenario. Through Granger causality testing, it is found that "GHI < 50W / m²" (i.e., sunset) and "residential load growth rate > 10kW / min" are significant Granger causes of "SOC decline rate > 5% / min". The system identifies a high-confidence key power outage factor combination. When the island's real-time operating data meets this combination of conditions, the system knows that one of the most dangerous scenarios in history is about to repeat itself, thus enabling early warning and intervention.

[0049] Furthermore, multiple dynamic energy storage models are collected, and the first level of power simulation content is determined based on these models, the corresponding key power outage factors, and the overall shape of the island. The second level of power simulation content is determined based on these models, the energy storage-consumption coefficient of the microgrid system, and the overall shape of the island. This comprehensive approach, which considers multiple dynamic energy storage models, the energy storage-consumption coefficient of the microgrid system, and the overall shape of the island, ensures the accuracy of the second level of power simulation content.

[0050] At this point, the triggering condition for the history-driven fault simulation comes directly from the "key power outage factors" extracted by S131. Once the real-time monitoring data meets or is about to meet this combination of factors, the simulation is activated. The system will call the dynamic energy storage model generated by S121, but here it is used to simulate the dynamic response under extreme external conditions. For example, when the "sudden drop in wind speed" factor is injected, the wind turbine power prediction model will output a power curve that drops sharply. The geographical and electrical topology of the island (output of S112) is used as the physical boundary condition for the simulation. The simulation engine must accurately model cable parameters, transformer impedance, etc., to calculate the physical processes such as power flow, voltage, and frequency of the entire network after the fault injection.

[0051] Specifically, the trigger is as follows: The current time on the island is 18:10, sunset has occurred, the system monitors a GHI of 20W / m², the residential load is increasing at a rate of 12kW / min, and the energy storage SOC is 32%, which is about to trigger the key power outage factor in the "evening peak energy storage depletion" scenario in S131; Fault injection: The system starts the first simulation and injects the following events into the digital twin model: Meteorological event: Force the wind turbine input speed to be linearly reduced from the current 6m / s to 3m / s within 5 minutes; Equipment status event: Set the initial SOC value of the energy storage model to 32% and apply a "low SOC" label, limiting its maximum discharge power to 60% of the rated value.

[0052] Simulation Execution: Based on an accurate topology model of the island containing long-distance cables, the simulation begins. Results show that the wind turbine output drops from 300kW to 50kW within 5 minutes; the energy storage system discharges at its maximum power of 540kW (900kW×60%), but cannot make up for the power deficit; the voltage at the end of the long cable connecting to the dock drops to 0.92pu within 8 minutes, triggering the low-voltage load shedding device; the system frequency drops to 49.2Hz within 10 minutes, entering an emergency state; Output: The first stage of power simulation content is generated, which is a set of catastrophic time-series data that accurately depicts the collapse process of the island microgrid when the historical tragedy is repeated.

[0053] The core input of the state-driven evolution simulation is the "energy storage-consumption coefficient" calculated by S123. This coefficient itself contains information about the vulnerability of the current supply and demand balance. The system also calls the dynamic energy storage model of S121, but here it uses their predictive function. The simulation engine performs iterative calculations within a time window (such as the next 90 minutes): at each step, it calculates the power difference based on the predicted generation and load, dynamically adjusts the charging and discharging power of the energy storage and updates the SOC based on the balance capacity revealed by the energy storage-consumption coefficient. The overall shape of the island serves as a physical constraint to ensure that every power flow conforms to the physical laws of the power grid.

[0054] Specifically, the input is: the energy storage-consumption coefficient calculated by S123 is 0.068, indicating that there is an energy gap in the system over time; the simulation execution is as follows: the system starts the second simulation, with the time window set to the next 90 minutes; the load forecast is as follows: the load forecast model shows that the residential load will reach a peak of 450kW in 30 minutes; the power generation forecast is as follows: the wind and solar forecast model shows that in the next hour, the photovoltaic output will drop to zero, and the wind power will remain at around 300kW; the iterative simulation is as follows: T+30min: the evening peak arrives, the total load reaches 730kW, and the total wind and solar output is 300kW. The energy storage system discharges at a power of 430kW; at T+75min: the SOC of the energy storage system drops to 25% after continuous discharge, reaching the safety threshold, and the system automatically stops discharging; at this time, the power deficit is 730kW-300kW=430kW; at T+76min: since the power deficit cannot be filled, the system frequency begins to drop rapidly, and the voltage becomes unstable; output: the second level of power simulation content is generated, which is a set of predictive time series data of the system heading towards instability; it is not as drastic as the first level of simulation, but clearly indicates that the system will enter an emergency state due to energy depletion in about 75 minutes.

[0055] Therefore, by collecting the scenario impact coefficients preset for the island, and determining the power simulation events of the microgrid system based on the first-level power simulation content, the second-level power simulation content, and the scenario impact coefficients preset for the island, the power simulation events present virtual power data of the island under the control of the microgrid system. Based on this virtual power data, the power decision content of the microgrid system in different scenarios is optimized. This approach is compatible with the overall consideration of the first-level power simulation content, the second-level power simulation content, and the scenario impact coefficients preset for the island, ensuring the accuracy of the power simulation events of the microgrid system. At the same time, the energy storage-consumption coefficient of the microgrid system is introduced to manage energy storage and consumption in multiple dimensions. This approach is compatible with the overall consideration of multiple key power outage factors, the energy storage-consumption coefficient of the microgrid system, and multiple dynamic energy storage models, thereby improving the accuracy of the power simulation events of the microgrid system.

[0056] At this point, the scenario impact coefficient preset for the island is collected. The scenario impact coefficient α (within the range of [0,1]) can be designed as a multivariable function. The closer α is to 1, the more attention is paid to historical risks; the closer it is to 0, the more attention is paid to the current state. Its dynamic adjustment logic can be based on a variety of triggering conditions: for example, when an extreme weather forecast is received, the system can automatically increase α because the probability of extreme physical failures is much higher than normal operation. Conversely, when the "energy storage-consumption coefficient" calculated by S123 is very low, it indicates that the system itself is in a vulnerable state. At this time, more attention should be paid to its own evolution trend, and α can be decreased. In addition, the system can learn and adjust the setting strategy of α based on the historical decision-making effect through an expert rule base or a more advanced reinforcement learning model.

[0057] Specifically, the current time on the island is 18:10, the weather is clear, and there is no extreme weather warning. However, the energy storage-consumption coefficient calculated by S123 is 0.068, which is a low value, indicating that the system itself has a significant risk of energy shortage. According to the internal state triggering rules, the system sets the scenario influence coefficient α to 0.25. This means that in the subsequent fusion calculation, the system will give the second simulation (state-driven) 75% weight, while the first simulation (history-driven) only accounts for 25% weight. The decision focuses more on solving the energy crisis that is about to occur.

[0058] The system must ensure that the time series data output from the first and second simulations are perfectly aligned on the time axis. For each time point t on the time axis, the system extracts the corresponding variable values ​​(such as bus voltage, system frequency, energy storage SOC, etc.) from the two simulation contents and performs a weighted calculation according to the formula: fused value (t) = first simulation value (t) × α + second simulation value (t) × (1-α). By combining the fused values ​​of all variables at all time points, a complete new time series dataset is formed, namely the "power simulation event". It contains both the shadow of the worst historical situation and reflects the current development trajectory, and is a future prediction after "risk weighting".

[0059] Specifically, both simulations generated data from 18:10 to 19:40, with a time step of 1 second. Taking the key variable "central energy storage SOC" as an example, at the time point T+75min: the first simulation (history-driven) predicted SOC of 5% (already collapsed); the second simulation (state-driven) predicted SOC of 25% (reaching the safety threshold). Using α=0.25 for fusion: fused SOC(T+75)=5%×0.25+25%×0.75=20%; the system performed the same weighted fusion on all key variables, ultimately generating a completely new power simulation event. This event predicts that the system will see the energy storage SOC drop to 20% in about 75 minutes. Although it will not collapse immediately like in the historical tragedy, it has entered an extremely dangerous emergency state and must be intervened immediately.

[0060] The system has a built-in power decision strategy library containing a variety of predefined control schemes, such as activating standby diesel generators, implementing demand-side response (DR), adjusting energy storage control strategies, or performing network reconfiguration. The system applies each strategy individually or in combination to the virtual data of "power simulation events" and reruns the simulation. The effectiveness of each strategy is evaluated through a multi-objective evaluation function that includes technical safety, economy, and equipment health. The system finds the optimal strategy under the multi-objective evaluation through genetic algorithms, particle swarm optimization, or simple traversal comparison. This optimal strategy and its expected effect constitute the final output power decision content.

[0061] Specifically, the system tested three strategies on the merged power simulation events: Strategy A (Diesel Generator Only): A 500kW diesel generator was started at T+60min; simulation results showed that the system was stable, but fuel costs were high; Strategy B (DR Only): The DR agreement with the hotel was initiated at T+45min, reducing the load by 150kW; simulation results showed that the SOC eventually dropped to 22%, the system barely maintained its position, but the risk remained high; Strategy C (DR + Energy Storage Optimization): DR was initiated at T+45min to reduce the load by 150kW, while the energy storage control strategy was adjusted; simulation results showed that the SOC eventually remained at 28%, the system voltage and frequency were very stable, and the cost was moderate; Optimization Output: The multi-objective evaluation function determined that Strategy C was the optimal solution; Final Decision Package: The final output of the "Power Simulation Event" from S133 not only included the merged prediction data but also the corresponding power decision content.

[0062] refer to Figure 5 In step S14, the specific steps are as follows: S141: Collect the power simulation event, determine multiple scene anomaly markers based on the identification of the power simulation event, determine the corresponding sub-abnormal power scene based on the tracing of each scene anomaly marker, and collect multiple sub-abnormal power scenes; S142: Collect island database, determine the construction information space within the island based on the identification of the island database, determine the current construction progress of the island based on the detection of the construction information space within the island, and determine the island power control area based on each sub-abnormal power scenario and the current construction progress of the island. S143: Determine the island power control content based on each sub-abnormal power scenario and multiple dynamic energy storage models. Determine the corresponding island power control measures based on the mapping relationship between the island power control content, island power control area and power control measures. At this time, determine the island power control coefficient based on the identification of the island power control measures. The island power control coefficient changes dynamically with the changes in the island's internal construction.

[0063] In the embodiments of this application, the power simulation event is collected, multiple scene anomaly markers are determined based on the identification of the power simulation event, and corresponding sub-abnormal power scenarios are determined by tracing the anomaly markers of each scene, so as to collect multiple sub-abnormal power scenarios, which is compatible with the overall consideration of tracing the anomaly markers of each scene and ensures the accuracy of the corresponding sub-abnormal power scenarios.

[0064] At this point, the system takes the power simulation event output by S133 (a multi-dimensional time series data containing future predictions) as input; the rule base has a large number of "IF-THEN" rules based on power system operation criteria, covering multiple aspects such as voltage quality, frequency stability, line safety, equipment status and power balance; the rule engine traverses the entire simulation event time series in a sliding window manner, and once a rule is triggered, it immediately generates a structured "scene anomaly marker" containing the time, location, physical quantity, current value and severity level of the anomaly.

[0065] The system uses the power grid topology map generated in S112 to perform graph traversal starting from the location of the anomaly marker to trace the topology source. At the same time, it combines the dynamic energy storage model in S121 and the load model in S122 to perform reverse power flow calculation to quantify the power deficit. Through analysis, the system can construct a clear anomaly chain, such as: low energy storage capacity (root cause) > insufficient energy storage power output > system power deficit > voltage / frequency collapse (result).

[0066] The system matches the analyzed "anomaly chain pattern" with a predefined "anomaly scenario pattern library". This pattern library stores causal chain templates for various typical faults, such as "islanding instability caused by main grid loss" and "voltage dip caused by heavy load input". Once a match is successful, the system defines a "sub-anomaly power scenario". This scenario is a highly condensed information body that contains core information such as the scenario type, root cause, scope of impact, and key evolution stages.

[0067] Specifically, the power simulation event prediction output by S133 shows the following system status at T+75min: the bus voltage in the western residential area is 0.92pu, the system frequency is 49.1Hz, and the central energy storage SOC is 20%. The rule engine scans this time point, triggers multiple rules, and generates multiple anomaly markers: MARK_1, MARK_2, and MARK_3.

[0068] The system analyzed the three markers mentioned above; it identified that MARK_1 and MARK_2 occurred at the same time and were highly correlated; through power flow calculation, the system found a huge power deficit of 400kW in the western bus; the system checked the energy storage model and found that because the SOC had dropped to 20%, its BMS triggered low SOC protection, limiting the maximum discharge power to 200kW, which could not fill the deficit; the anomaly chain constructed by the system clearly pointed to the failure of the energy storage system due to energy depletion, which was the root cause of the collapse of the entire western power grid.

[0069] The system matches the anomaly chain of "low energy storage capacity > power deficit > voltage / frequency collapse" with the "energy storage depletion type collapse" template in the pattern library, achieving a matching accuracy of up to 95%. The system defines a sub-anomaly power scenario. This well-defined sub-anomaly scenario is much clearer than the initial three isolated markers. It accurately describes the nature, process, and impact of the problem, providing perfect input for the subsequent development of precise control measures.

[0070] Furthermore, an island database is collected, and the construction information space within the island is determined based on the identification of the island database. The current construction progress of the island is determined based on the detection of the construction information space within the island. The power control area of ​​the island is determined based on each sub-abnormal power scenario and the current construction progress of the island. This overall consideration of each sub-abnormal power scenario and the current construction progress of the island ensures the accuracy of the power control area of ​​the island.

[0071] At this point, the system constructs this information space by integrating four key data sources: Geographic Information System (GIS), Building Information Modeling (BIM), Electrical Topology Model (from S112), and Asset Management System (EAM). Through a spatial data engine and a unified ID system, these four types of data are aligned and associated. For example, the "hotel power distribution room" in the BIM model is bound to the geographic coordinates in the GIS, the transformer asset ID in the EAM, and the node ID in the electrical topology, forming a fused information space of "one object, multiple attributes".

[0072] The system defines a "status label" for each building or area, such as: [Planning, Foundation Construction, Interior Decoration, Operational, Maintenance]. It uses various technologies for real-time monitoring: automatically updating the status using IoT sensors deployed in the construction area; synchronizing with the Project Management System (PMS) via API to obtain the latest project milestones; inferring whether a building is operational by analyzing smart meter data; and even using computer vision technology to analyze images taken by drones or cameras to automatically identify construction features such as tower cranes and scaffolding.

[0073] The system acquires the "sub-abnormal power scenario" of S141 and extracts its impact range (e.g., "western island"). In the information space, the polygon of this impact range is overlaid with the geographic layers of all buildings and areas. All entities overlaid within the impact range are logically filtered according to their "status labels," with the core principle being to select only those entities that can effectively respond to the power system. For example, only entities with the status "in operation" are selected, excluding areas under "foundation construction" or "not in operation." The system generates one or more discontinuous, functional "power control areas," which are collections of multiple independently controllable "power nodes."

[0074] Specifically, the "Island Construction Information Space" is an integrated digital platform. On this platform, clicking the "Hotel" icon on the GIS map allows users to: view its BIM model and see through it to examine the 3D structure of its basement power distribution room; view its EAM information, such as "dry-type transformer, capacity 1250kVA, commissioning date 2023-05-20"; and see on the electrical topology diagram that it is connected to the "WH-FDR-03" feeder of the "Western Busbar". This information space connects isolated "data silos" into an organic whole.

[0075] The system monitors the construction progress of the western part of the island in multiple ways: Hotel: Its smart meter data shows a normal daily load curve, and the status label is "in operation"; Commercial Street: The system reads its PMS status as "interior decoration" through API, and the IoT sensor at the entrance of its construction area shows activity, confirming the status label as "interior decoration"; at the same time, its electricity application process has not been completed, and it is in an "unconnected" electrical state; First Residential Area: The meter data for Phase I is normal, and the status is "in operation"; Drone images of Phase II plots show that pile drivers are working, and the status label is "foundation construction".

[0076] Input: The sub-anomaly scenario determined by S141, whose impact range is "Western Island"; Overlay and Filtering: The system overlays the "Western Island" range with the information space to obtain the hotel, commercial street, Phase I of the First Residential Area, and Phase II of the First Residential Area; Filtering Starts: Hotel: Status "In Operation", selected; it has the electrical and control infrastructure to participate in Demand Side Response (DR); Commercial Street: Status "Under Renovation" and "Not Connected" Electrically, excluded; Issuing DR commands to it is meaningless; Phase I of the First Residential Area: Status "In Operation", selected; its smart meters support load control; Phase II of the First Residential Area: Status "Foundation Construction", excluded; Final Area Division: The island power control area ZONE_A determined by the system is not the entire "Western Island", but a precise, discrete set of entities: {"Hotel", "Phase I of the First Residential Area"}. This area definition is passed to S143 to ensure that subsequent control measures can accurately and effectively act on the correct targets.

[0077] Therefore, the island power management content is determined based on each sub-abnormal power scenario and multiple dynamic energy storage models. The corresponding island power management measures are determined based on the mapping relationship between the island power management content, the island power management area, and the power management measures. At this time, the island power management coefficient is determined based on the identification of the island power management measures. This island power management coefficient changes dynamically with the changes in the island's internal construction, which is compatible with the overall consideration of the identification of island power management measures and ensures the accuracy of the island power management coefficient.

[0078] At this point, the system has a built-in "scenario-control content" knowledge base, which defines the macro strategies to be adopted under different abnormal scenarios based on power system theory and expert experience. For example, for the "energy storage depletion-type cascading failure" scenario, the knowledge base will map control content such as "load reduction" and "increasing backup power supply". At the same time, the dynamic energy storage model of S121 provides key feasibility constraints here. If the energy storage itself is already in a deep discharge state, then the priority of control content that depends on energy storage (such as "increasing reactive power support") will be reduced or marked as "infeasible".

[0079] The system maintains a "content-measure" mapping knowledge base, which defines the specific technical means to implement each type of control content; for example, "load reduction" can be mapped to "implementing demand-side response (DR)" or "implementing tiered load reduction"; the system combines the "island power control area" determined by S142 to perform adaptability screening of measures; it queries the asset database to confirm which equipment in the area has the ability to participate in the corresponding measures (such as whether a DR agreement has been signed, whether remote control is supported), thereby determining the list of executable measures.

[0080] The system defines an "island power control coefficient," which is a performance indicator that quantifies the deviation between the actual and predicted effects of control measures. After the measures are implemented, the system continuously collects actual power grid operation data and compares it with the expected "post-control" curve. By calculating the control coefficient as the ratio of actual effect value to expected effect value, the system can assess the accuracy of each action. This coefficient is persistently stored and correlated with changes in island construction. For example, the system records that "when the second phase of the first residential area is put into operation, the load characteristics of its first phase change, resulting in a 10% decrease in the effectiveness of air conditioning group control." When similar scenarios are encountered in the future, the system will automatically use this learned knowledge to make more accurate predictions for decision-making, realizing an intelligent closed loop where control strategies are dynamically optimized according to changes in the island's operational status and environment.

[0081] Specifically, the input scenario is the "energy storage depletion-type cascading failure" determined by S141; the system queries the knowledge base to determine the corresponding island power control content for this scenario as: load reduction (the goal is to reduce at least 400kW before T+75min) and activation of backup power; at the same time, the system queries the energy storage model of S121 and finds that its current SOC is only 20%, which can no longer provide additional support, so the focus of control is completely shifted to the load side and backup power.

[0082] The input is the control content {"Reducing load (400kW)", "Starting backup power"} and the control area ZONE_A={"Hotel", "Phase 1 of Residential Area"}; Load reduction: The system decides to adopt the "Execute DR" measure; The query finds that the hotel can reduce 150kW, and the Phase 1 Residential Area can reduce 100kW through intelligent air conditioning control, for a total of 250kW, which is still insufficient; The system continues to search and finds that the quay crane in the dock area has a 15-minute idle period after T+70min, which can temporarily interrupt the power supply and reduce 250kW; Backup power start: The system decides to adopt the "Remote command start" measure, with the target being "Standard backup diesel generator No. 1 (500kW)"; The island power control measures determined by the system are a list containing specific targets, objects and parameters, which accurately guides subsequent actions.

[0083] After the above measures were implemented as planned, the system detected a deviation between the actual effect and the expectation. The hotel's actual DR was reduced by 155kW, with a control coefficient of 155 / 150≈1.03. The air conditioning group control in the first residential area was reduced by 85kW, with a control coefficient of 85 / 100=0.85. The system analysis found that the temperature was low at the time, and the air conditioning compressor itself was operating at a lower power than expected. The system associated and stored the coefficient 0.85 with tags such as "Phase I of the First Residential Area", "Air Conditioning Group Control", and "Temperature <25°C". In the future, when formulating similar strategies, the system will consider meteorological factors and make more accurate predictions of DR effects, thereby achieving self-evolution and continuous optimization.

[0084] refer to Figure 6 In step S15, the specific steps are as follows: S151: Collect multiple island power control measures, determine the first control coefficient based on the multiple island power control measures and the microgrid system, determine the second control coefficient based on the multiple island power control measures and the corresponding energy storage path, and determine the island power control event based on the mapping relationship between the first control coefficient, the second control coefficient and the island power control event; S152: Based on the island power control events and the corresponding power control areas, determine multiple sub-power control items, and based on the multiple sub-power control items and the abnormal power data of the microgrid system, determine the abnormal control content; S153: Determine the energy storage management content based on the energy storage status of multiple sub-power management projects, multiple dynamic energy storage models, and the collaborative relationship of multiple dynamic energy storage models; determine the autonomous power management system of the microgrid system located on the island based on the abnormal management content, the energy storage management content, and the current power consumption data of the island.

[0085] In the embodiments of this application, multiple island power management measures are collected, a first management coefficient is determined based on the multiple island power management measures and the microgrid system, a second management coefficient is determined based on the multiple island power management measures and the corresponding energy storage path, and island power management events are determined based on the mapping relationship between the first management coefficient, the second management coefficient and island power management events. This approach takes into account the overall consideration of the mapping relationship between the first management coefficient, the second management coefficient and island power management events, ensuring the accuracy of island power management events.

[0086] At this point, the first control coefficient is a comprehensive normalized score (usually between 0 and 1), calculated based on a weighted evaluation of multiple key performance indicators, including voltage stability, frequency stability, and power balance. The system employs rapid digital simulation, applying the control measures of S143 as input to the power grid model of S112 to quickly calculate the future dynamic response of the system. A multi-objective decision function (such as weighted summation) combines the performance of the above dimensions into a first control coefficient C1 = w1 × Score_voltage + w2 × Score_frequency + w3 × Score_balance. Here, Score_voltage is the voltage stability score; Score_frequency is the frequency stability score; and Score_balance is the power balance score. These three coefficients (w1, w2, w3) represent the "importance" of voltage, frequency, and power balance in the overall score. Their values ​​are set by the system designer or operation and maintenance expert based on the operating characteristics of the power grid and the current focus.

[0087] The second control coefficient is also a normalized score, which is calculated based on key indicators of energy storage operation, including SOC trajectory health, cycle life loss and economy; the system calls the dynamic energy storage model of S121, especially the battery aging model and cost model contained therein; by comparing the energy storage operation trajectory under the two scenarios of "implementing measures" and "not implementing measures", the degree of improvement of the above dimensions is quantified and synthesized into the second control coefficient.

[0088] The system maintains an internal "coefficient-event level" mapping library. For example, C1>0.9 and C2>0.9 can be mapped to "Level 1 optimization event". The system creates an event object and encapsulates all key information such as event ID and level, list of control measures, evaluation coefficient, expected target, timestamp and lifecycle in it.

[0089] Furthermore, multiple sub-power control projects are identified based on the island's power control events and the corresponding power control areas. Abnormal control content is determined based on the abnormal power data of multiple sub-power control projects and the microgrid system. This approach takes into account the overall consideration of abnormal power data from multiple sub-power control projects and the microgrid system, ensuring the accuracy of abnormal control content.

[0090] At this point, the system parses the list of control_measures in the control event and instantiates each measure (such as DR, start diesel generator) into an independent sub-control project object; it assigns a set of standardized attributes to each sub-project object to form a structured task card, including project ID, type, execution target (i.e., the specific entity in the "power control area" determined by S142), execution parameters, status, and the intelligent agent responsible for execution.

[0091] The system retrieves scenario anomaly markers generated by S141 (such as low voltage and low frequency), which are "abnormal factors" that the system needs to address. The system uses grid topology and power flow knowledge to perform causal impact analysis on each sub-project to determine which abnormal indicators will be positively affected after its execution. For example, power generation projects (starting a diesel generator) will increase the system's power output, which is directly related to "frequency anomaly" and "voltage anomaly". Load projects (DR) will reduce power consumption, which is also related to these anomalies. Once the association is established, the system generates anomaly control content for the sub-project and quantifies its "control objectives".

[0092] Therefore, the energy storage management content is determined based on the energy storage status of multiple sub-power management projects, multiple dynamic energy storage models, and the collaborative relationships of multiple dynamic energy storage models. Based on the anomaly management content, energy storage management content, and the current power consumption data of the island, an autonomous power management system for the microgrid system located on the island is determined. This system is compatible with the overall consideration of anomaly management content, energy storage management content, and the current power consumption data of the island, ensuring the accuracy of the autonomous power management system for the microgrid system located on the island. Simultaneously, island power management events are introduced to further manage multiple sub-power management projects, achieving an overall consideration of the collaborative relationships of multiple sub-power management projects, abnormal power data of the microgrid system, and multiple dynamic energy storage models, thus improving the accuracy of the autonomous power management system.

[0093] At this point, the system generates strategies by integrating multi-source inputs; it acquires the sub-power management items of S152 to understand the action plans of all other entities, which provides "context" for energy storage coordination; it refers to the dynamic energy storage model of S121 to obtain the real-time "physical condition" of energy storage (such as SOC, SOH, and maximum charging and discharging power) as "physical constraints" for formulating strategies; based on the optimal coordination relationship between energy storage and other equipment (such as diesel generators) defined in the model, it generates phased management content for energy storage; for example, in the "transition period" before diesel generator startup, energy storage is set to V / f (voltage / frequency) control mode and operates as an islanded main power source; in the "coordination period" after diesel generator grid connection, energy storage switches to PQ (active / reactive) control mode and enables the Virtual Synchronous Generator (VSG) algorithm, playing the role of "auxiliary regulator" and "stabilizer".

[0094] The autonomous power management system for a microgrid located on an island is constructed in three layers: the decision-making layer takes the control events of S151 as the highest command; the coordination layer takes the anomaly content of S152 and the energy storage strategy of S153 as the core, clarifying the coordination logic; the execution layer consists of multiple sub-projects of S152, which are responsible for specific execution; the system's "sensory perception" is the real-time power consumption data of the island, which will be continuously monitored after execution and compared with the simulation prediction of S133; if the actual effect is found to deviate from the expectation, the system has the ability to adaptively adjust and can dynamically fine-tune the parameters of certain sub-projects, forming a complete closed loop of perception-decision-execution-feedback; the final output is not a static plan, but a dynamic and continuously operating example of an autonomous power management system.

[0095] Please see Figure 7 , Figure 7 This is a schematic diagram of the structural composition of the power simulation device for a microgrid system in an embodiment of the present invention; the power simulation device for the microgrid system includes: The energy storage path module 21 is used to determine the current grid system of the microgrid system based on multiple working data of the microgrid system when the microgrid system is located on an island, and to determine multiple energy storage paths according to the current grid system, the construction engineering map of the island and the corresponding multiple energy storage nodes. The energy storage-consumption coefficient module 22 is used to determine multiple dynamic energy storage models of different dimensions based on the island's weather data, wind data, and multiple energy storage paths. Based on the multiple dynamic energy storage models, the island's current power consumption data, and the corresponding power consumption equipment, the energy storage-consumption coefficient of the microgrid system is determined. The power simulation event module 23 is used to determine multiple key power outage factors based on past power outage events of the island, and to determine the power simulation events of the microgrid system based on multiple key power outage factors, the energy storage-consumption coefficient of the microgrid system and multiple dynamic energy storage models. The island power management and control measures module 24 is used to determine multiple sub-abnormal power scenarios based on the identification of the power simulation event, and to determine the corresponding island power management and control measures based on each sub-abnormal power scenario, multiple dynamic energy storage models and the current construction progress of the island. The autonomous power management system module 25 is used to determine island power management events based on multiple island power management measures, microgrid systems, and multiple energy storage paths; to determine multiple sub-power management projects based on the identification of island power management events; and to determine the autonomous power management system of the microgrid system located on the island based on the collaborative relationship between multiple sub-power management projects, abnormal power data of the microgrid system, and multiple dynamic energy storage models.

[0096] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

Claims

1. A power simulation method of a microgrid system, characterized by, The method comprises the following steps: When the micro-grid system is in an island, a current power grid system of the micro-grid system is determined based on multiple working data of the micro-grid system, multiple energy storage paths are determined according to the current power grid system, a construction engineering diagram of the island and corresponding multiple energy storage nodes; Multiple dynamic energy storage models of different dimensions are determined according to weather data, wind data and the multiple energy storage paths of the island, and an energy storage-consumption coefficient of the micro-grid system is determined based on the multiple dynamic energy storage models, current power consumption data of the island and corresponding power consumption equipment. The energy storage-consumption coefficient contains vulnerability information of current supply and demand balance; Multiple key power outage factors are determined based on previous power outage events of the island, and a power simulation event of the micro-grid system is determined according to the multiple key power outage factors, the energy storage-consumption coefficient of the micro-grid system and the multiple dynamic energy storage models; the power simulation event contains multiple-dimensional time series data predicted in the future; Multiple sub-abnormal power scenarios are determined according to the identification of the power simulation event, and corresponding island power management and control measures are determined according to each sub-abnormal power scenario, the multiple dynamic energy storage models and the current construction process of the island; the sub-abnormal power scenario contains the type, root cause, influence range and key evolution stage of the scenario; An island power management and control event is determined according to the multiple island power management and control measures, the micro-grid system and the multiple energy storage paths, and multiple sub-power management and control projects are determined based on the identification of the island power management and control event; a self-power management and control system of the micro-grid system in the island is determined based on the multiple sub-power management and control projects, abnormal power data of the micro-grid system and a cooperative relationship of the multiple dynamic energy storage models. 2.The power emulation method of the micro-grid system according to claim 1, wherein, The method comprises the following steps: The corresponding micro-grid system is configured on the island, and the power output by the micro-grid system is fed back to the island construction project, the micro-grid system is monitored in real time, and multiple working data of the micro-grid system are collected, which cover the scene dimensions of the ocean, the sun and the wind field; The current power grid system of the micro-grid system is determined based on the multiple working data of the micro-grid system, the corresponding scene dimensions and the current working mode of the micro-grid system; the construction engineering diagram of the island is collected, the corresponding power consumption path of the island is determined according to the construction engineering diagram of the island and the cable distribution diagram of the island, and the multiple energy storage paths are determined according to the power consumption path, the current power grid system and the corresponding multiple energy storage nodes. 3.The power emulation method of the microgrid system according to claim 1, wherein, The method comprises the following steps: The weather data and wind data of the island are collected according to the multiple detections of the island, and the corresponding island data combinations are determined based on the weather data, the wind data and the detection time of the island; the multiple dynamic energy storage models of different dimensions are determined according to each island data combination and the multiple energy storage paths. Collecting current power consumption data of the island, determining a plurality of sub-power consumption events based on the identification of the current power consumption data of the island, and determining a consumption combination according to the plurality of sub-power consumption events and the corresponding power consumption equipment; Determining a storage combination according to the plurality of dynamic storage models and the current power consumption data of the island, and determining a storage-consumption coefficient of the micro-grid system based on the consumption combination, the storage combination, and the storage effect coefficient of the micro-grid system. 4.The method of claim 1, wherein, The plurality of key outage factors are determined based on the past outage events of the island, and the power simulation event of the micro-grid system is determined according to the plurality of key outage factors, the storage-consumption coefficient of the micro-grid system, and the plurality of dynamic storage models, including: The past outage events of the island are determined based on the micro-grid system and the island database, the plurality of outage scenarios are determined according to the identification of the past outage events of the island, and the corresponding key outage factors are determined according to the detection of each outage scenario, so as to collect the plurality of key outage factors.

5. The power emulation method of the microgrid system according to claim 4, wherein, The plurality of key outage factors are determined based on the past outage events of the island, and the power simulation event of the micro-grid system is determined according to the plurality of key outage factors, the storage-consumption coefficient of the micro-grid system, and the plurality of dynamic storage models, further including: The plurality of dynamic storage models are collected, the first heavy power simulation content is determined according to the plurality of dynamic storage models, the corresponding key outage factors, and the overall shape of the island; and the second heavy power simulation content is determined according to the plurality of dynamic storage models, the storage-consumption coefficient of the micro-grid system, and the overall shape of the island; The scenario influence coefficient preset by the island is collected, the power simulation event of the micro-grid system is determined based on the first heavy power simulation content, the second heavy power simulation content, and the scenario influence coefficient preset by the island, the power simulation event presents the power virtual data of the island under the control of the micro-grid system, and the power decision content of the micro-grid system in different scenarios is optimized based on the power virtual data. 6.The power emulation method of the microgrid system of claim 1, wherein, The plurality of sub-abnormal power scenarios are determined according to the identification of the power simulation event, and the corresponding island power control measures are determined according to each sub-abnormal power scenario, the plurality of dynamic storage models, and the current construction process of the island, including: The power simulation event is collected, the plurality of scenario abnormal markers are determined based on the identification of the power simulation event, the corresponding sub-abnormal power scenario is determined according to the tracing of each scenario abnormal marker, and the plurality of sub-abnormal power scenarios are collected. 7.The power emulation method of the microgrid system according to claim 6, wherein, The plurality of sub-abnormal power scenarios are determined according to the identification of the power simulation event, and the corresponding island power control measures are determined according to each sub-abnormal power scenario, the plurality of dynamic storage models, and the current construction process of the island, further including: The island database is collected, the island construction information space is determined according to the identification of the island database, the current construction process of the island is determined based on the detection of the island construction information space, and the island power control area is determined according to each sub-abnormal power scenario and the current construction process of the island; According to the various sub-exceptional power scenarios and the plurality of dynamic energy storage models, the island power management content is determined, based on the mapping relationship of the island power management content, the island power management area and the power management measure, the corresponding island power management measure is determined, at this time, based on the identification of the island power management measure, the island power management coefficient is determined, and the island power management coefficient dynamically changes with the change of the island construction. 8.The method of Claim 1, wherein, According to the plurality of island power management measures, the micro-grid system and the plurality of energy storage paths, the island power management event is determined, based on the identification of the island power management event, the plurality of sub-power management projects is determined, based on the synergistic relationship of the plurality of sub-power management projects, the abnormal power data of the micro-grid system and the plurality of dynamic energy storage models, the autonomous power management system of the micro-grid system in the island is determined, including: A plurality of island power management measures are collected, based on the plurality of island power management measures and the micro-grid system, a first control coefficient is determined, based on the plurality of island power management measures and the corresponding energy storage path, a second control coefficient is determined, based on the mapping relationship of the first control coefficient, the second control coefficient and the island power management event, the island power management event is determined. 9.The power emulation method of the microgrid system according to claim 8, wherein, According to the plurality of island power management measures, the micro-grid system and the plurality of energy storage paths, the island power management event is determined, based on the identification of the island power management event, the plurality of sub-power management projects is determined, based on the synergistic relationship of the plurality of sub-power management projects, the abnormal power data of the micro-grid system and the plurality of dynamic energy storage models, the autonomous power management system of the micro-grid system in the island is determined, including: According to the island power management event and the corresponding power management area, a plurality of sub-power management projects are determined, based on the plurality of sub-power management projects and the abnormal power data of the micro-grid system, the abnormal management content is determined; Based on the plurality of sub-power management projects, the energy storage state of the plurality of dynamic energy storage models and the synergistic relationship of the plurality of dynamic energy storage models, the energy storage management content is determined; based on the abnormal management content, the energy storage management content and the current power consumption data of the island, the autonomous power management system of the micro-grid system in the island is determined.

10. A power simulation device of a microgrid system, characterized by, The power simulation device of the micro-grid system is applied to the power simulation method of the micro-grid system in any one of claims 1-9, and the power simulation device of the micro-grid system comprises: The energy storage path module is used for determining the current power grid system of the micro-grid system based on a plurality of working data of the micro-grid system when the micro-grid system is in the island, and determining a plurality of energy storage paths according to the current power grid system, the construction engineering drawing of the island and a plurality of corresponding energy storage nodes; The energy storage-consumption coefficient module is used for determining a plurality of dynamic energy storage models of different dimensions according to the weather data, the wind data of the island and the plurality of energy storage paths, and determining the energy storage-consumption coefficient of the micro-grid system based on the plurality of dynamic energy storage models, the current power consumption data of the island and the corresponding power consumption equipment; The power simulation event module is used for determining a plurality of key power failure factors based on the past power failure events of the island, and determining the power simulation event of the micro-grid system according to the plurality of key power failure factors, the energy storage-consumption coefficient of the micro-grid system and the plurality of dynamic energy storage models. The island power control measure module is configured to determine a plurality of sub-abnormal power scenarios according to the identification of the power simulation event, determine corresponding island power control measures according to each sub-abnormal power scenario, a plurality of dynamic energy storage models, and a current construction process of the island; The autonomous power control system module is configured to determine an island power control event according to a plurality of island power control measures, a micro-grid system, and a plurality of energy storage paths, determine a plurality of sub-power control projects based on the identification of the island power control event, and determine an autonomous power control system of the micro-grid system in the island based on the plurality of sub-power control projects, abnormal power data of the micro-grid system, and a cooperative relationship of the plurality of dynamic energy storage models.

Images