Intelligent control method and system of district energy storage system for microgrid black start

Abstract

This application relates to the field of industrial big data and discloses an intelligent control method and system for a distribution network energy storage system for black start in a microgrid. The method includes: acquiring the initial state records of each energy storage unit in the microgrid; performing distributed calculations on location and capacity differences to generate compensation coefficient data; determining the initial power allocation ratio and load difference value to adjust the initial power allocation ratio; obtaining the adjusted power allocation ratio; determining the balanced power output configuration; acquiring the updated operating status of each energy storage unit; and generating a coordinated control sequence to control the power output of each energy storage unit by adjusting the information interaction frequency. This application generates compensation coefficients and determines the initial power allocation ratio through distributed calculation, achieving differentiated power configuration for multiple energy storage units and avoiding distribution imbalances caused by location and capacity differences; it adjusts the power allocation ratio based on load difference values ​​and adjusts the information interaction frequency according to operating status deviations, thereby improving black start stability.

Description

Technical Field

[0001] This invention belongs to the field of industrial big data technology, specifically a method and system for intelligent control of distribution area energy storage systems for black start of microgrids. Background Technology

[0002] In modern power systems, microgrids serve as a crucial means of ensuring energy security and improving power supply reliability, and their black-start capability directly impacts the rapid recovery of the power system. Black start refers to the process by which a power system, after a large-scale power outage caused by a fault, gradually restores power supply without relying on the external power grid, solely through its internal self-starting power sources. With the widespread application of distributed energy resources, the need for multiple energy storage units to collaboratively participate in black start-up is becoming increasingly urgent. Existing control methods can already achieve basic single-unit start-up or simple parallel operation.

[0003] However, existing methods lack overall control capabilities in complex scenarios involving the coordination of multiple energy storage units. Poor information exchange between units and imbalanced power distribution lead to system fluctuations and local instability, making it difficult to guarantee the success rate of black starts and recovery efficiency for users. Summary of the Invention

[0004] To address the above problems, this invention provides an intelligent control method for a distribution area energy storage system for microgrid black start, which solves the technical problems of poor information exchange between units and power distribution imbalance leading to system fluctuations and local instability.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows: Acquire the initial state records of each energy storage unit in the microgrid, the initial state records including location information, capacity status and load data; Based on the initial state record, distributed calculations are performed on the location and capacity differences between each energy storage unit to generate compensation coefficient data. The compensation coefficient data characterizes the relative deviation of each energy storage unit in the energy storage system. The compensation coefficient data is the proportion of the comprehensive deviation value of each energy storage unit to the sum of the comprehensive deviation values ​​of all energy storage units. The comprehensive deviation value is obtained by weighting the location and capacity differences of each energy storage unit. The initial power allocation ratio of each energy storage unit is determined based on the inverse relationship between the compensation coefficient data and the power allocation share, and the load difference value of each energy storage unit is determined based on the initial power allocation ratio and the load data. The initial power allocation ratio is adjusted based on the load difference value to obtain the adjusted power allocation ratio, and the balanced power output configuration is determined based on the adjusted power allocation ratio. The updated operating status of each energy storage unit is obtained according to the power output configuration, and the updated operating status includes the voltage response value and frequency response value of each energy storage unit. A time-series operating state diagram is constructed by recording the voltage response value and the frequency response value in chronological order. Based on the operating state diagram, a monitoring loop is executed to determine the deviation between the voltage response value and the corresponding voltage response value in the initial state record, and between the frequency response value and the corresponding frequency response value in the initial state record. The information interaction frequency between each energy storage unit is adjusted according to the deviation amount, and an adjustment record is generated. The adjustment record includes: the timestamp of the adjustment, the voltage deviation and frequency deviation of each energy storage unit that triggered the adjustment, the information interaction frequency value of each energy storage unit before the adjustment, and the information interaction frequency value of each energy storage unit after the adjustment. The adjustment record is saved to the distributed storage node. Based on the adjustment records, a coordinated control sequence is generated through rule extraction or sequence pattern mining methods. The coordinated control sequence is used to control the power output of each energy storage unit to achieve stable power distribution and suppression of system fluctuations during the black start process of the microgrid.

[0006] In addition, to achieve the above objectives, this application also proposes an intelligent control system for a distribution area energy storage system for microgrid black start, the system comprising: an initial state recording acquisition module, a compensation coefficient generation module, an initial power allocation determination module, a power allocation ratio adjustment module, an updated state acquisition module, a state diagram construction module, a deviation determination module, an interaction frequency adjustment module, and a coordinated control generation module. The initial state record acquisition module is used to acquire the initial state records of each energy storage unit in the microgrid. The initial state records include location information, capacity status, and load data. The compensation coefficient generation module is used to perform distributed calculations on the location and capacity differences between each energy storage unit based on the initial state record, and generate compensation coefficient data. The compensation coefficient data characterizes the relative deviation of the energy storage unit in the energy storage system. The compensation coefficient data is the proportion of the comprehensive deviation value of each energy storage unit to the sum of the comprehensive deviation values ​​of all energy storage units. The comprehensive deviation value is obtained by weighting the location and capacity differences of each energy storage unit. The initial power allocation determination module is used to determine the initial power allocation ratio of each energy storage unit based on the inverse relationship between the compensation coefficient data and the power allocation share, and to determine the load difference value of each energy storage unit based on the initial power allocation ratio and the load data. A power allocation ratio adjustment module is used to adjust the initial power allocation ratio based on the load difference value, obtain the adjusted power allocation ratio, and determine the balanced power output configuration based on the adjusted power allocation ratio. The update status acquisition module is used to acquire the updated operating status of each energy storage unit according to the power output configuration. The updated operating status includes the voltage response value and frequency response value of each energy storage unit. The state diagram construction module is used to construct a time-seriesd operating state diagram by recording the voltage response value and the frequency response value in chronological order. The deviation determination module is used to execute a monitoring loop based on the running state diagram to determine the deviation between the voltage response value and the corresponding voltage response value in the initial state record, and between the frequency response value and the corresponding frequency response value in the initial state record; The interaction frequency adjustment module is used to adjust the information interaction frequency between each energy storage unit according to the deviation amount and generate an adjustment record. The adjustment record includes: the timestamp of the adjustment, the voltage deviation and frequency deviation of each energy storage unit that triggered the adjustment, the information interaction frequency value of each energy storage unit before the adjustment, and the information interaction frequency value of each energy storage unit after the adjustment. The adjustment record is saved to the distributed storage node. The coordination control generation module is used to generate a coordination control sequence based on the adjustment record through rule extraction or sequence pattern mining methods. The coordination control sequence is used to control the power output of each energy storage unit to achieve stable power distribution and suppression of system fluctuations during the black start process of the microgrid.

[0007] The technical solution proposed in this application involves acquiring the location information, capacity status, and load data of each energy storage unit. Distributed calculations are then performed to generate compensation coefficients based on the location and capacity differences between units, achieving adaptive correction of power allocation. Based on the initial power allocation ratio and load difference value determined by the compensation coefficients, the output configuration is further adjusted to obtain balanced power output, effectively avoiding local overload or fluctuations caused by uneven load. Furthermore, the information exchange frequency is dynamically adjusted and a coordinated control sequence is generated according to the deviation between the updated operating status and the initial status record, improving the collaborative response capability and system stability of multiple energy storage units during black start-up, thereby enhancing the recovery efficiency and operational safety of the microgrid during black start-up. Attached Figure Description

[0008] Figure 1 This is a flowchart illustrating an embodiment of the intelligent control method for a transformer energy storage system for black start microgrids in this application. Figure 2 This is a flowchart illustrating Embodiment 2 of the intelligent control method for a transformer energy storage system for black start of a microgrid, as provided in this application. Figure 3 This is a flowchart illustrating Embodiment 3 of the intelligent control method for a distribution area energy storage system for black start microgrids provided in this application. Detailed Implementation

[0009] To enable those skilled in the art to better understand the technical solution, the present invention will be described in detail below with reference to embodiments. The description in this part is only exemplary and explanatory, and should not be used to limit the scope of protection of the present invention in any way. In existing technologies, the intelligent control methods for distribution area energy storage systems for microgrid black start have the following drawbacks: They lack the ability to collaboratively utilize the location differences, capacity status, and load data among multiple energy storage units, making it impossible to scientifically quantify the deviations between units during the startup phase, resulting in an unreasonable initial power allocation ratio; path simulation and information interaction do not fully consider the location differences between units and real-time load fluctuations, leading to inaccurate load difference assessments and difficulty in effectively adjusting power output configurations; and the deviation between the updated operating status and the initial status record is not used in a timely manner to dynamically adjust the information interaction frequency, causing the coordinated control sequence to be unable to adapt to system fluctuations, ultimately affecting the recovery efficiency and operational stability of black start.

[0010] Therefore, this application provides a solution: by acquiring the location information, capacity status, and load data of each energy storage unit, distributed calculations are performed on the location and capacity differences between units to generate compensation coefficients, thus solving the problem of lack of basis for initial power allocation; the initial power allocation ratio is determined based on the compensation coefficients, and the output configuration is dynamically adjusted in combination with the load difference value to obtain balanced power output, effectively avoiding local overload and system fluctuations; the updated operating status is obtained according to the power output configuration, and the information interaction frequency is dynamically adjusted based on the deviation from the initial state record to generate a coordinated control sequence, thereby realizing full-link closed-loop control from status acquisition and deviation quantification to power allocation and interaction optimization in the process of multi-energy storage unit collaboration, significantly improving the recovery efficiency and operational safety of microgrid black start.

[0011] It should be noted that the executing entity of each embodiment of this application can be a computing service system with data processing, network communication, and program execution functions, such as an electronic system capable of realizing the above functions, or an intelligent control system for a distribution area energy storage system for microgrid black start. The following description uses an intelligent control system for a distribution area energy storage system for microgrid black start (hereinafter referred to as "the system") as an example to illustrate the following embodiments.

[0012] Based on this, this application provides an intelligent control method for a distribution area energy storage system for microgrid black start, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the intelligent control method for a distribution area energy storage system for black start microgrids according to this application.

[0013] In this embodiment, the intelligent control method for the distribution area energy storage system for microgrid black start includes steps S10~S90: Step S10: Obtain the initial state record of each energy storage unit in the microgrid. The initial state record includes location information, capacity status, and load data.

[0014] It should be understood that before the black start process begins, it is necessary to acquire the initial state records of all energy storage units participating in the startup of the microgrid. The initial state records include data in at least three dimensions: location information, capacity status, and load data. Location information refers to the geographical coordinates or electrical node location of each energy storage unit in the microgrid, expressed in latitude and longitude or feeder number, used to quantify the spatial distance between units; capacity status refers to the percentage of the energy storage unit's current available capacity relative to its rated capacity, expressed as a percentage, reflecting the unit's remaining energy level; load data refers to the current power output value undertaken by the energy storage unit, expressed in kilowatts, representing the unit's real-time output in the system. This data is acquired through a distributed sensor network deployed on each energy storage unit. The sensors read data from the internal battery management system according to a preset sampling period (e.g., once per second) and transmit the data to the central coordination node via low-power wide-area network technologies (such as LoRa) or industrial Ethernet.

[0015] To ensure data comparability across different units, a unified node communication protocol is used to format the data. For example, a custom protocol based on JSON defines fields including unit identifier, latitude and longitude coordinates, capacity percentage, and load power. A cyclic redundancy check (CRC) algorithm is used to verify the integrity of each data packet, and packets failing the check are automatically retransmitted. After receiving data from each unit, the central coordination node decrypts and parses it to obtain the initial state record of each energy storage unit. If the capacity status of a unit is significantly lower than the average (e.g., more than 10% lower than the average capacity of all units), the system triggers a load balancing prediction, prioritizing adjustments to the power allocation for that unit in subsequent steps. After the above collection, formatting, transmission, and aggregation processing, the initial state records of all energy storage units are completely stored in the central database, serving as the basis for subsequent difference calculations and power allocation.

[0016] Step S20: Based on the initial state record, perform distributed calculations on the location differences and capacity differences between each energy storage unit to generate compensation coefficient data. The compensation coefficient data characterizes the relative deviation of the energy storage unit in the energy storage system. The compensation coefficient data is the proportion of the comprehensive deviation value of each energy storage unit to the sum of the comprehensive deviation values ​​of all energy storage units. The comprehensive deviation value is obtained by weighting the location differences and capacity differences of each energy storage unit.

[0017] It should be noted that location difference refers to the spatial distance between two energy storage units, which can be calculated using the Euclidean distance formula, that is, the straight-line distance is calculated based on the latitude and longitude coordinates of the two units. The greater the distance, the greater the line loss and response delay in the power transmission process between the units. Capacity difference refers to the difference in the current available capacity of the two units, expressed in kilowatt-hours or as a percentage. This difference reflects the degree of imbalance in energy reserves between the units.

[0018] Since the number of energy storage units may be large, this application adopts a distributed computing framework to process the paired differences between units in parallel to improve computational efficiency. A synchronization update mechanism is introduced during the calculation process to ensure that data from all units participate in the calculation within the same time window, avoiding additional deviations caused by inconsistent data acquisition times. Specifically, with a preset time period (e.g., 1 second) as the synchronization rhythm, each unit broadcasts its own location coordinates and capacity percentage to adjacent nodes, and each node independently calculates its own location and capacity differences with other nodes.

[0019] Weights are assigned to location differences and capacity differences, namely location weights and capacity weights. These weights can be set according to the system design objectives, for example, a location weight of 0.4 and a capacity weight of 0.6, to highlight the decisive role of capacity status in the black start process. The comprehensive deviation value between each pair of units is obtained according to the weighted summation formula. After summing all pairwise deviation values, the proportion of each unit's comprehensive deviation value to the sum of all unit comprehensive deviation values ​​is defined as the compensation coefficient data for that unit.

[0020] The compensation coefficient reflects the relative deviation of the unit in the entire energy storage system. If the compensation coefficient of a unit exceeds the preset threshold range (e.g., the normal range is 0.9 to 1.1), it indicates that the state of the unit deviates significantly from the average level of the group, and an adjustment command needs to be sent in subsequent steps for correction.

[0021] Step S30: Determine the initial power allocation ratio of each energy storage unit based on the inverse relationship between the compensation coefficient data and the power allocation share, and determine the load difference value of each energy storage unit based on the initial power allocation ratio and the load data.

[0022] It should be noted that the initial power allocation ratio refers to the power output share preset for each energy storage unit according to the inverse relationship of the compensation coefficient, without considering real-time load fluctuations.

[0023] It should be understood that a higher compensation coefficient indicates a greater deviation in that unit, and its output share should be appropriately reduced during power allocation to avoid exacerbating system imbalance. Conversely, units with lower compensation coefficients can handle more power output. For example, by taking the reciprocal of the compensation coefficients of all units and then normalizing them, the initial power allocation ratio of each unit can be obtained. The initial power allocation ratio determines the percentage of total output power that each unit should output under ideal conditions.

[0024] However, during black start, the actual load data of each unit changes in real time, reflecting the actual power output (kilowatts) currently undertaken by each unit. Multiplying the initial power allocation ratio by the total system power demand yields the theoretical allocated power value for each unit. This theoretical allocated power value is then compared with the actual load data of that unit; the difference between the two is the load difference value. The load difference value can be an absolute value (kilowatts) or a relative value (percentage), and its physical meaning is the degree of deviation between the current actual load and the ideal allocation. If the actual load of a unit is much higher than its theoretical allocation value, the load difference value for that unit is positive and large, indicating that the unit may be in an overload state; conversely, it is negative, indicating that the unit is in a light load state.

[0025] This embodiment uses the load difference value as a quantitative indicator to determine whether the system has unbalanced fluctuations. It should be noted that the compensation coefficient data and the load difference value describe the unbalanced state between units from two dimensions: static difference and dynamic load, respectively. The two complement each other and together constitute the basis for power adjustment.

[0026] Step S40: Adjust the initial power allocation ratio based on the load difference value to obtain the adjusted power allocation ratio, and determine the balanced power output configuration based on the adjusted power allocation ratio.

[0027] After calculating the load difference value of each energy storage unit, the initial power allocation ratio is adjusted based on this value to obtain a balanced power output configuration. Specifically, if the load difference value of a certain unit is too large (for example, a large positive value indicates that the unit is overloaded), its power output share needs to be reduced, and the reduced share is redistributed to units with negative load difference values ​​(light load) so that the load of all units tends to be balanced.

[0028] In practice, the system determines whether the absolute value of the load difference for each unit exceeds a preset load difference threshold. This threshold can be set based on system capacity and power regulation accuracy, for example, as 10% of the rated power of a single unit or an absolute value such as 5 kilowatts. When the load difference for a unit exceeds this threshold, the system identifies a risk of local instability and initiates a power share redistribution process. This redistribution employs a group iterative logic, such as a genetic algorithm or a multi-objective optimization algorithm, with the goal of minimizing overall system fluctuations, iteratively adjusting the power allocation ratio for each unit. In each iteration, the standard deviation of the overall system load distribution is recalculated based on the current power configuration; if the standard deviation decreases, the adjustment is retained.

[0029] After multiple iterations (e.g., 100 iterations), the algorithm converges to a set of power allocation ratios that ensure the load difference values ​​of all units are less than a threshold; this is the adjusted power allocation ratio. Multiplying the adjusted ratio by the total system output power yields the power output value (kilowatts) for each unit. These output values ​​collectively constitute the balanced power output configuration. This configuration effectively avoids localized overloads or voltage / frequency fluctuations caused by uneven load distribution, providing stable power support for the black start process.

[0030] For example, the initial power allocation ratios are 20%, 23%, 18%, 28%, and 21%, and the corresponding load difference values ​​show that node 4 is overloaded. After iterative optimization by the genetic algorithm, the ratios are adjusted to 19%, 22%, 19%, 27%, and 23%, so that the load deviation of all nodes is reduced to within the threshold. The balanced output configurations are 190, 220, 190, 270, and 230 kilowatts, respectively.

[0031] Step S50: Obtain the updated operating status of each energy storage unit according to the power output configuration. The updated operating status includes the voltage response value and frequency response value of each energy storage unit.

[0032] Updating the operating status refers to the set of electrical parameters actually exhibited by each energy storage unit under a new power output configuration. These parameters include at least voltage response values ​​(volts) and frequency response values ​​(hertz), which are collected in real time by the unit's local voltage transformers and frequency measurement modules and reported to the central coordination node. Specifically, the balanced power output configuration is distributed to the actuators of each energy storage unit through the central coordination node. Upon receiving the instruction, each unit adjusts its power output, and simultaneously, the unit's local voltage transformers and frequency measurement modules initiate continuous acquisition mode, measuring the voltage response and frequency response values ​​at their output terminals in real time at a preset sampling period (e.g., 10 times per second). These real-time acquired electrical parameters constitute the updated operating status. Unlike the initial state record, the updated operating status reflects the dynamic response characteristics actually exhibited by each unit after power adjustment, rather than static configuration data. The acquired data is reported to the central coordination node through an encrypted communication channel (e.g., adaptive message transmission based on the MQTT (Message Queuing Telemetry Transport) protocol), forming a multi-dimensional state stream indexed by unit identifier and aligned to time.

[0033] Step S60: By recording the voltage response value and the frequency response value in chronological order, a time-seriesd operating state diagram is constructed.

[0034] It should be understood that after receiving voltage and frequency data from each unit, the central coordination node stores and integrates the data in chronological order. The system maintains a first-in, first-out (FIFO) circular queue for each energy storage unit. The queue length is determined by a preset backtracking window, for example, retaining data points collected every 0.1 seconds within the last 60 seconds, totaling 600 data points. Each storage unit contains three key pieces of information: a timestamp accurate to milliseconds, the measured voltage value, and the measured frequency value. All unit queues are aligned along the same time axis, and a multi-dimensional time-series operational status diagram is formed with time as the horizontal axis and voltage or frequency as the vertical axis. This operational status diagram can visually display the trajectory of each unit's electrical parameters over time as a curve, or it can be stored at the underlying level as a large-scale data matrix with time and unit identifier as a joint primary key. The core purpose of constructing this diagram is to provide a complete, reliable, and traceable historical data foundation for subsequent deviation calculations, trend analysis, and the prediction of fluctuation indicators.

[0035] Step S70: Execute a monitoring loop based on the operating state diagram to determine the deviation between the voltage response value and the corresponding voltage response value in the initial state record, and between the frequency response value and the corresponding frequency response value in the initial state record.

[0036] In practice, the system performs monitoring operations cyclically at a fixed monitoring cycle (e.g., once every 0.5 seconds). Within each monitoring cycle, the system extracts the latest data points from the operating status diagram. For each energy storage unit, its current voltage value is compared with the initial voltage value saved in the initial state record to calculate the voltage deviation; similarly, its current frequency value is compared with the initial frequency value to calculate the frequency deviation.

[0037] Voltage deviation and frequency deviation can be positive or negative, representing whether the current value is higher or lower than the initial reference value, respectively. The absolute value of the deviation directly reflects the degree to which the system deviates from the initial equilibrium point: the smaller the absolute value, the closer the system is to a stable state; the larger the absolute value or the more frequently the sign changes, the more unstable factors or fluctuations exist in the system.

[0038] To eliminate potential misjudgments caused by noise from single-point measurements, the system can also employ a moving average method to smooth the deviations obtained from multiple consecutive sampling periods before use. When the absolute value of the voltage deviation of any energy storage unit exceeds a preset fluctuation threshold (e.g., one volt) or the absolute value of the frequency deviation exceeds a preset fluctuation threshold (e.g., 0.2 Hz), the system determines that there is a "sign of local fluctuation." Furthermore, the system can further calculate the standard deviation of the voltage deviations of all units. If this standard deviation shows an increasing trend over time, it also indicates that system fluctuations are spreading, requiring intervention.

[0039] Step S80: Adjust the information interaction frequency between each energy storage unit according to the deviation amount, and generate an adjustment record. The adjustment record includes: the timestamp of the adjustment, the voltage deviation and frequency deviation of each energy storage unit that triggered the adjustment, the information interaction frequency value of each energy storage unit before the adjustment, and the information interaction frequency value of each energy storage unit after the adjustment. The adjustment record is saved to the distributed storage node.

[0040] Once step S70 determines that there are signs of local fluctuations, the system immediately activates the fault-tolerant recovery mechanism. The core operation of this mechanism is to dynamically adjust the information exchange frequency between energy storage units based on the severity of the fluctuations. Under normal stable operation, to conserve communication resources, energy storage units typically exchange status information at a low frequency, such as once per second, i.e., an information exchange frequency of one hertz. When fluctuation signs are detected, the central coordination node generates tiered frequency adjustment commands based on the magnitude of the deviation: for units with small deviations that only slightly exceed the threshold, the information exchange frequency is increased to two hertz; for units with large deviations that significantly exceed the threshold, it is increased to four hertz; and for severely fluctuating units with continuously increasing deviations, it can be further increased to five hertz or even higher. The purpose of increasing the information exchange frequency is to shorten the communication delay and control loop response time between units, enabling all units to learn about the status changes of neighboring units more quickly, thereby synchronously adjusting their respective power outputs and collaboratively suppressing the spread of fluctuations.

[0041] After frequency adjustment is completed, the system must generate a structured adjustment record to ensure subsequent traceability and analysis. The adjustment record should include at least the following fields: a precise timestamp of the adjustment (in milliseconds of Coordinated Universal Time); snapshots of the voltage and frequency deviations of each unit that triggered the adjustment; the frequency of information exchange between units before the adjustment; the frequency of information exchange between units after the adjustment; an identifier of the fluctuation type (e.g., voltage drop fluctuation or frequency oscillation fluctuation); and the source of the adjustment instruction (e.g., a central coordinating node or a distributed consensus result). These fields are serialized according to a predefined encoding format (e.g., JSON or Protocol Buffers).

[0042] To ensure the persistence, reliability, and immutability of records, the system writes adjustment records to distributed storage nodes. These nodes use a consistent hashing algorithm to shard the data, and each adjustment record is replicated to at least three different storage nodes for redundancy. The storage structure employs an append-only, immutable log format. Each new record is appended to the end of the log and linked to the hash value of the previous record, preventing subsequent tampering of historical records. After successful storage, the system returns a globally unique identifier for the adjustment record for subsequent querying and analysis.

[0043] It is important to emphasize that adjusting the information interaction frequency is not a one-time action, but a continuous closed-loop adaptive process. After the first adjustment, the system continues to run the monitoring loop in step S70 to recalculate the deviation. If the deviation gradually decreases and falls back within the preset fluctuation threshold, it indicates that the adjustment is effective, and the system can gradually restore the information interaction frequency to a normal level. If the deviation continues to increase or the fluctuation does not ease, the system further increases the information interaction frequency, and may even trigger higher-level intervention measures, such as temporarily disconnecting the unit with the most severe fluctuation. Each frequency adjustment generates a corresponding adjustment record, which is stored in the distributed nodes in the manner described above. By continuously repeating the "monitoring-judgment-adjustment-recording" loop, the system can adapt to the dynamic changes in the microgrid state in real time, ensuring that the black start process remains under control.

[0044] Therefore, step S80 may further include steps S801 to 806: Step S801: Using each energy storage unit as a node, the normalized value of the deviation as the node feature, and the reciprocal of the difference in voltage response phase difference and frequency response rate of change between units as the edge weight, a dynamic information interaction graph is constructed.

[0045] Based on the determination of fluctuation indicators and the triggering of fault-tolerant recovery mechanisms according to the deviation, the system can further employ graph theory-based methods to quantitatively solve for the optimal information interaction frequency adjustment vector, rather than simply relying on hierarchical empirical values. Specifically, each energy storage unit is treated as a node in the graph. The deviation of each unit, calculated at the moment of calculation, is processed by min-max normalization and used as the characteristic value of that node. Simultaneously, the reciprocal of the absolute value of the voltage response phase difference and the absolute value of the difference in the frequency response rate of change between any two units is used as the edge weight connecting these two nodes. The physical meaning of this edge weight is that the smaller the phase difference and the closer the frequency rate of change of two units, the more similar their electrical behaviors, resulting in a larger edge weight and a higher coupling strength in subsequent Laplace analysis. The dynamic information interaction graph constructed in this way fully depicts the state correlation and electrical coupling degree between the units under the current fluctuation scenario.

[0046] Step S802: Calculate the algebraic connectivity of the Laplacian matrix of the dynamic information interaction graph. The algebraic connectivity characterizes the system's cooperative ability to suppress bias under the current information interaction frequency configuration.

[0047] Furthermore, the Laplace matrix of the dynamic information interaction graph is calculated, and the second smallest eigenvalue of the Laplace matrix, namely the algebraic connectivity, is obtained. Algebraic connectivity is a key indicator in graph theory for measuring the strength of graph connectivity. In this embodiment, it quantitatively characterizes the ability of the entire energy storage system to collaboratively suppress deviations through information exchange between units under the current information interaction frequency configuration. The higher the algebraic connectivity, the greater the synchronization potential of the information interaction topology, and the faster the units can reach a consistent control response; conversely, the lower the algebraic connectivity, the more difficult it is for the system to form effective collaborative suppression even if information exchange exists. The system pre-sets a target connectivity, which reflects the desired level of collaborative suppression capability.

[0048] Step S803: Based on the difference between the algebraic connectivity and the preset target connectivity, solve for the information interaction frequency adjustment vector that maximizes the algebraic connectivity increment. Each component in the adjustment vector corresponds to a communication frequency adjustment coefficient between an energy storage unit and its neighboring units.

[0049] Based on the difference between the current algebraic connectivity and the preset target connectivity, the system constructs an optimization problem: finding an information interaction frequency adjustment vector that maximizes the increment of algebraic connectivity in the adjusted dynamic information interaction graph. Each component of the adjustment vector corresponds to an adjustment coefficient for the communication frequency between an energy storage unit and its neighboring units. This coefficient can be positive (indicating that the communication frequency needs to be increased) or negative (indicating that the communication frequency can be appropriately reduced). This optimization problem can be solved using semidefinite programming or gradient-based iterative algorithms.

[0050] Step S804: Update the information interaction frequency of each energy storage unit according to the adjustment vector, and jointly encode the deviation distribution before adjustment, the adjustment vector, the algebraic connectivity after adjustment, and the timestamp of this adjustment into a structured adjustment record.

[0051] After obtaining the adjustment vector, the system synchronously updates the information interaction frequency between the corresponding energy storage unit and its neighboring units according to the values ​​of each component. After the update is completed, the system jointly encodes the deviation distribution before adjustment, the obtained adjustment vector, the algebraic connectivity recalculated after adjustment, and the precise timestamp of this adjustment into a structured adjustment record.

[0052] Step S805: After compressing the adjustment record using a columnar compression algorithm, it is appended to the immutable log file of the distributed storage node. The immutable log file is stored in shards according to the time span, and each shard contains a Bloom filter. The bit array of the Bloom filter is constructed based on the globally incrementing sequence number of all adjustment records in the shard.

[0053] In the specific implementation, a columnar compression algorithm is used to compress the adjustment record to reduce storage space usage, and it is then appended to the immutable log file of the distributed storage node. The immutable log file is stored in shards according to time spans, for example, every ten minutes or every thousand records constitutes a shard. An independent Bloom filter is built within each shard. The bit array of the Bloom filter is constructed based on the globally incrementing sequence number of all adjustment records within the shard, thereby enabling subsequent fast retrieval based on sequence number ranges or the existence of specific features without traversing the entire log file.

[0054] The steps may further include: serializing the pre-adjustment deviation distribution into a one-dimensional feature vector, quantizing the adjustment vector into a fixed-width integer increment sequence, quantizing the algebraic connectivity metric into a single-precision floating-point number, and converting the timestamp into a millisecond-level integer in Coordinated Universal Time (UTC) to obtain serialized data; aggregating the serialized data into a binary payload according to a predetermined field order, and calculating the cyclic redundancy check (CRC) code or hash value of the binary payload as an integrity digest; concatenating the binary payload with the integrity digest, and appending the concatenation result with the globally incrementing sequence number of the current adjustment record to form a structured adjustment record.

[0055] In its implementation, to ensure the consistency, integrity, and efficient parsing of adjustment records within the distributed storage system, the system performs a rigorous serialization encoding and verification process on the record content: The pre-adjustment deviation distribution (i.e., the set of voltage and frequency deviation values ​​for each unit) is expanded into a one-dimensional feature vector according to the fixed order of the unit identifiers; each component in the solved information interaction frequency adjustment vector is quantized into a fixed-width integer increment sequence, for example, each adjustment coefficient is represented by an 8-bit signed integer, ranging from -128 to +127, with each unit corresponding to a frequency change step of 0.5 percent; the algebraic connectivity measure is converted into a single-precision floating-point number, occupying four bytes; and the current adjustment timestamp is converted into a millisecond-level integer starting from 00:00 UTC, occupying eight bytes. After processing the above data in this format, serialized data suitable for aggregation is obtained.

[0056] Furthermore, the system concatenates these binary data sequentially according to a predetermined field order (e.g., first storing the timestamp, then the vector dimension length, the one-dimensional feature vector of the deviation, the integer sequence of the adjustment vector, and the algebraic connectivity) to form a continuous binary payload.

[0057] To prevent bit errors or malicious tampering during transmission or storage, the system calculates a cyclic redundancy check (CRC32) or cryptographic hash (SHA-256) as an integrity digest for the binary payload. This integrity digest is appended to the end of the binary payload, and then a globally incrementing sequence number (e.g., a monotonically increasing 64-bit integer starting from zero) is appended to the beginning of the composite data block, forming a complete, self-describing structured adjustment record. This globally incrementing sequence number not only guarantees the total order relationship between records but also provides an indexing basis for subsequent Bloom filter construction and sequence-based range queries. After this structured adjustment record is written to the immutable log file of the distributed storage node, any subsequent read operation can verify the record's integrity and tamper-proof status by recalculating the payload's integrity digest and comparing it with the stored digest. This provides a solid foundation for tracing and auditing the microgrid black start process and reliably generating coordinated control sequences.

[0058] Step S90: Based on the adjustment record, a coordinated control sequence is generated by rule extraction or sequence pattern mining method. The coordinated control sequence is used to control the power output of each energy storage unit to achieve stable power distribution and suppression of system fluctuations during the black start process of the microgrid.

[0059] Through continuous operation across multiple monitoring cycles, the system accumulates a large number of adjustment records. These records comprehensively document the entire process of system fluctuations, from their emergence to the implementation of fault-tolerant measures, and then to the mitigation or deterioration of the fluctuations. This includes the distribution of deviations at each trigger point, the executed information interaction frequency switching commands, and the actual effects of the adjustments. Based on these historical adjustment records, the system generates coordinated control sequences using rule extraction or data mining methods.

[0060] A coordinated control sequence is essentially an ordered set of control instructions. Each instruction includes at least: the desired execution time or relative delay; a list of target energy storage units; the control type, such as power increase, power decrease, maintaining current power, switching information exchange frequency, adding or removing units; and specific control parameters, such as the magnitude of power adjustment and the target information exchange frequency value. Furthermore, some instructions may include triggering conditions, such as "repeatedly execute this instruction when the voltage deviation of a unit exceeds 0.8 volts again."

[0061] There are various methods for generating coordinated control sequences. For example, frequently occurring patterns can be extracted from adjustment records, and patterns such as "when the deviation exceeds a certain value and lasts for more than a certain number of seconds, execute the frequency increase command" can be transformed into conditions. The action rules are then combined into a sequence according to time order, or a sequence pattern mining algorithm is used to extract typical control flows from a large number of records to form a standardized coordinated control template. After generating the coordinated control sequence, the central coordination node distributes it to the local controller of each energy storage unit. Each unit executes the instructions in the sequence independently or collaboratively, thereby achieving closed-loop control of power output.

[0062] The key value of the coordinated control sequence lies not only in its role as a response record for the current fluctuation scenario but also in its ability to serve as a pre-set control strategy for similar future fluctuation scenarios. When the system detects similar deviation patterns and fluctuation signs again, it can directly invoke the pre-stored coordinated control sequence without needing to iterate and optimize the frequency adjustment again, thereby significantly shortening the response time and improving the recovery efficiency and operational stability of the black start process. Through the complete closed loop of steps S50 to S90, the system achieves end-to-end intelligent control, from state monitoring, deviation quantification, fluctuation prediction, adaptive adjustment of information interaction frequency, persistent storage of adjustment records, to the generation and reuse of the coordinated control sequence. This effectively solves the system fluctuation and local instability problems caused by poor information interaction and power distribution imbalance during the black start process of multiple energy storage units.

[0063] It should be noted that this application is not only applicable to conventional microgrid black-start scenarios, but can also be widely extended to various application environments such as grid-connected microgrids with renewable energy access, stand-alone microgrids, and industrial park-level energy storage clusters. In scenarios with high renewable energy penetration, the intermittent output of photovoltaic and wind power can exacerbate the state differences between energy storage units. This application can still achieve reasonable power allocation through distributed computing and adaptive adjustment of compensation coefficients. In energy storage systems in large industrial parks or commercial complexes, multiple distribution areas and energy storage units coexist with complex load characteristics. This application utilizes load difference detection and dynamic adjustment mechanisms for information interaction frequency to effectively suppress local fluctuations and ensure stable system operation. In remote areas or island stand-alone microgrids, where communication conditions are limited and units are distributed widely, this application adopts a lightweight distributed synchronous update and fault-tolerant recovery mechanism that does not rely on high-speed centralized communication, yet can still generate reliable coordinated control sequences. Therefore, this technical solution has broad scenario adaptability and portability, and is not limited to a specific microgrid configuration or operating conditions.

[0064] Based on the first embodiment of this application, in the second embodiment of this application, the content that is the same as or similar to that in Embodiment 1 above can be referred to the above description, and will not be repeated hereafter. Based on this, please refer to... Figure 2 Step S20 includes steps S201 to S204: Step S201: Based on the initial state record, a synchronous update mechanism is used to perform consistency calibration on the data of each energy storage unit to obtain the calibrated difference value matrix.

[0065] Since the location information and capacity status data of each energy storage unit are collected independently through a distributed sensor network, the data from different units may be inconsistent in time due to acquisition time, communication delay, or local clock deviation. If this inconsistency is used directly for difference calculation, it will introduce errors and affect the accuracy of the compensation coefficient.

[0066] To address this, this embodiment introduces a synchronization update mechanism to calibrate the data consistency of each unit. Specifically, within a unified time window, all energy storage units broadcast their status data to adjacent nodes, and each node only uses the data received within that time window for computation. Specifically, the system uses a fixed time period (e.g., 1 second) as the synchronization clock. At the beginning of each clock cycle, the central coordination node or distributed consensus node broadcasts a synchronization signal. Upon receiving the signal, each energy storage unit immediately latches its current location information, capacity status, and load data, and then packages and sends this data. Since the data of all units corresponds to the same synchronization moment, the impact of time deviation is eliminated.

[0067] Furthermore, each computing node (which can be an independent computing server or a master node within an energy storage unit) collects data from all units at the synchronization moment, arranging them according to unit identifiers to construct a two-dimensional data structure. The first dimension represents different energy storage units, and the second dimension represents various parameters of each unit, including location coordinates, capacity percentage, and load power. Based on this, the location difference and capacity difference values ​​between every two units are calculated, and all paired difference values ​​are filled into a symmetric matrix, which is the calibrated difference value matrix. The element in the i-th row and j-th column of the difference value matrix represents the location difference (e.g., Euclidean distance, in kilometers) and capacity difference (e.g., the absolute value of the capacity percentage difference, in %) between energy storage unit i and energy storage unit j. Since the location difference and capacity difference have different dimensions, they need to be stored separately or weighted and fused later. The calibrated difference value matrix ensures that all difference calculations are based on data from the same time base, avoiding misjudgments caused by asynchronous acquisition.

[0068] Step S202: Based on the difference value matrix, perform a weighted analysis on the location and capacity differences between each energy storage unit to obtain the comprehensive deviation value of each energy storage unit.

[0069] It should be understood that the purpose of weighted analysis is to address the different degrees to which location differences and capacity differences affect system stability, and these differences need to be reflected by weighting coefficients. These weighting coefficients can be predetermined based on engineering experience or system design requirements. For example, in a microgrid black-start scenario, capacity status directly determines the unit's ability to provide power support, so the capacity weight can be set relatively high (e.g., 0.6); while location differences affect line losses and response speed, so their weight can be relatively low (e.g., 0.4).

[0070] In the specific implementation, for each pair of units (i,j) in the difference value matrix, their location difference d_ij (km) and capacity difference c_ij (percentage) are extracted. The location deviation contribution value is obtained by multiplying d_ij by a preset location weight w_pos; the capacity deviation contribution value is obtained by multiplying c_ij by a preset capacity weight w_cap. The two are then added together to obtain the paired comprehensive deviation value between units (i,j). For each energy storage unit i, its paired comprehensive deviation value is accumulated or averaged with the comprehensive deviation values ​​of all other units j (j≠i) to obtain the comprehensive deviation value of that unit. The larger the comprehensive deviation value, the greater the overall difference between that unit and other units in the system, requiring greater compensation in subsequent power allocation.

[0071] For example, the calculated paired comprehensive deviations between the three units A, B, and C are 14.83, 11.66, and 8.83, respectively. These are then summed to obtain the comprehensive deviation value for each unit: the comprehensive deviation for A is 14.83(AB) + 11.66(AC) = 26.49, the comprehensive deviation for B is 14.83 + 8.83 = 23.66, and the comprehensive deviation for C is 11.66 + 8.83 = 20.49. The comprehensive deviation reflects the degree of "isolation" or "imbalance" of each unit within the entire system and is the core basis for generating the compensation coefficient.

[0072] The difference matrix is ​​processed based on a preset threshold range to obtain the adjusted deviation weights.

[0073] In some cases, the calculated overall deviation value may be too widely distributed, or the deviation value of some units may significantly exceed the normal range. In such cases, directly using the original deviation value to generate the compensation coefficient will lead to overly drastic power adjustment or insufficient response. To solve this problem, this embodiment introduces a preset threshold range to perform secondary processing on the difference value matrix to obtain the adjusted deviation weight.

[0074] The preset threshold range includes a lower threshold and an upper threshold. For example, the normal range for the comprehensive deviation value is set to [5, 30]. When the paired comprehensive deviation value or the accumulated comprehensive deviation value of a unit is lower than the lower threshold, it indicates that the difference between the unit and other units is minimal, and its adjustment weight can be reduced. When it is higher than the upper threshold, it indicates that the unit deviates significantly from the group, and its adjustment weight needs to be significantly increased. Specifically, each element in the difference value matrix (or the comprehensive deviation value of each unit) is traversed to determine whether it falls within the preset threshold range. For values ​​falling within the range, they remain unchanged. For values ​​lower than the lower limit, they are forcibly set to the lower limit value or multiplied by a reduction factor less than 1. For values ​​higher than the upper limit, they are forcibly set to the upper limit value or multiplied by an amplification factor greater than 1. After the above mapping process, the resulting new deviation weight distribution is smoother, avoiding excessive impact of extreme values ​​on power allocation.

[0075] Alternatively, piecewise linear transformations or sigmoid functions can be used for nonlinear mapping, ensuring that the adjusted deviation weights reflect the trend of the original deviations without exceeding a controllable range. The specific form of the adjusted deviation weights remains the same as the structure of the difference value matrix, but the numerical range is compressed or stretched to a reasonable interval.

[0076] Step S203: Based on the comprehensive deviation value, compensation coefficient data is generated using the normalized ratio method.

[0077] It should be understood that the compensation coefficient is a dimensionless value used to quantify the proportion of power that each unit should be compensated or attenuated in power distribution. There are various ways to generate it, but the core principle remains the same: the larger the overall deviation value of a unit, the higher its compensation coefficient should be, indicating that it needs to accept more power adjustments or higher priority coordination control.

[0078] In this embodiment, the normalized ratio method is used to generate the compensation coefficient. Specifically, the comprehensive deviation values ​​of all units are summed to obtain the total deviation value. For each unit i, its comprehensive deviation value is divided by the total deviation value to obtain the compensation coefficient of that unit. The sum of the compensation coefficients of all units is 1. Units with larger compensation coefficients will be allocated a smaller initial power ratio in subsequent power allocation (because their output needs to be reduced to balance the system), or they will be given a higher priority to receive adjustment commands.

[0079] For example, the comprehensive deviation values ​​of A, B, and C are 26.49, 23.66, and 20.49, respectively, with a total deviation value of 70.64. Then, the compensation coefficient of A is approximately 0.375 (26.49 / 70.64), B is approximately 0.335, and C is approximately 0.290.

[0080] As one implementation, step S202 includes: Determine the location weight and capacity weight; The location deviation value is calculated based on the location weight and the location difference between any two energy storage units, and the capacity deviation value is calculated based on the capacity weight and the capacity difference between the two energy storage units; The position deviation value and the capacity deviation value are summed to obtain the paired comprehensive deviation value between the two energy storage units; The overall deviation value of each energy storage unit is obtained by summing the paired comprehensive deviation values ​​between each energy storage unit and all other energy storage units.

[0081] In practice, location weight and capacity weight are two preset coefficients used to adjust the relative importance of location and capacity differences in the overall deviation calculation. The weight values ​​need to be determined based on the actual physical characteristics of the microgrid black start scenario. Location differences mainly affect the electrical distance and power transmission loss between energy storage units. The farther the location, the more voltage drop margin and communication delay compensation need to be reserved in coordinated control. Capacity differences affect the ability of the units to provide power support. The greater the capacity deviation, the more unbalanced the overall energy distribution of the system.

[0082] The location and capacity differences of energy storage units i and j are read from the difference value matrix. The deviation values ​​are obtained by weighting them separately using location and capacity weights. Since the two have different dimensions, direct weighting may lead to a dominant result. To ensure comparability, the difference values ​​are first normalized (e.g., divided by their respective maximum values ​​to the range of 0 - 1) before weighting. Finally, each pair of units (i, j) corresponds to one location deviation value and one capacity deviation value. The location and capacity deviation values ​​of the same pair of units (i, j) are directly added to obtain the pairwise comprehensive deviation value between the units. After obtaining the comprehensive deviation value for each energy storage unit, the comprehensive deviation values ​​of all units are added together to obtain the total comprehensive deviation value.

[0083] For example, in a microgrid with five energy storage units, their calculated comprehensive deviation values ​​are 10, 20, 15, 25, and 30, respectively, totaling 100. The compensation coefficients are then 0.10, 0.20, 0.15, 0.25, and 0.30, respectively. The initial power allocation ratio corresponding to the unit with the highest compensation coefficient (0.30) will be set to the minimum to avoid system instability caused by abnormal conditions in that unit. This generation method has clear physical meaning and sound mathematical properties, ensuring that the compensation coefficients faithfully reflect the degree of difference between units.

[0084] In one implementation, step S40 includes steps S401 to S402: Step S401: Determine whether the load difference value is greater than a preset load difference threshold.

[0085] In practice, the preset load difference threshold is pre-set based on the system's safety margin and equipment adjustment capabilities. For example, it may be set to 10% of the rated power of a single energy storage unit, or an absolute power value such as 5 kW. When the deviation between the actual load and the ideal allocation is within range, the system is in an acceptable unbalanced state and requires no intervention. Once the deviation exceeds the threshold, it indicates a serious overload or underload situation for that unit, which, if left untreated, may lead to localized overheating, voltage drops, or frequency fluctuations. The judgment process can be performed independently for each unit, or it can assess whether the maximum load difference value among all units exceeds the threshold. If the load difference value of at least one unit exceeds the threshold, the system determines that a power share redistribution process needs to be initiated.

[0086] Step S402: If the load difference value is greater than the preset load difference threshold, the power share is redistributed through iterative optimization to obtain the adjusted power allocation ratio, and the balanced power output configuration is determined based on the adjusted power allocation ratio.

[0087] When the load difference value is determined to be greater than the preset load difference threshold, this embodiment redistributes the power share through iterative optimization. In the black-start scenario of multiple energy storage units, the primary optimization objective is to converge the load difference values ​​of all units to within the threshold, while avoiding new fluctuations during the adjustment process. This embodiment uses an iterative optimization algorithm to approximate this objective. The current power allocation ratio of each unit is used as the initial solution. In each iteration, based on the current load difference distribution, the direction and step size of the power adjustment for each unit are calculated: for units with positive load differences (overload), their power share is reduced; for units with negative load differences (light load), their power share is increased. The adjustment step size can be a fixed step size (e.g., 1% adjustment each time) or an adaptive step size (dynamically changing according to the magnitude of the deviation).

[0088] In one implementation, step S402 includes: Obtain the deviation between the current power share and the ideal power configuration; The required power allocation ratio is determined based on the deviation value; With the goal of minimizing the total system fluctuation, the power share is iteratively optimized to obtain the adjusted power allocation ratio.

[0089] In practice, the current power share refers to the proportion of the power output actually undertaken by each energy storage unit to the total output power of the system before the start of this iteration, expressed as a percentage; the ideal power configuration refers to the ideal power distribution ratio that theoretically makes all units completely balanced and without any overload or light load. Under the ideal configuration, the load difference value of each unit is zero.

[0090] In practical applications, ideal power configuration is defined in several ways: for example, by allocating power according to the rated capacity ratio of each unit, or by allocating power in reverse according to the compensation coefficient of each unit. Regardless of the definition used, the deviation value is the difference between the current power share and the ideal power share.

[0091] For a specific energy storage unit i, let its current power share be P_i (%) and its ideal power share be P_i* (%), then the deviation value Δ_i = P_i - P_i*. A positive Δ_i indicates that the unit is currently carrying a power share exceeding the ideal (potentially overloaded), while a negative Δ_i indicates carrying a share below the ideal (light load). The magnitude of the deviation value determines the extent of subsequent adjustments. After obtaining the deviation values ​​for all units, the norm of the deviation vector (such as the root mean square error) can be further calculated as a quantitative indicator of the overall system imbalance. When the deviation values ​​are generally large, it indicates that the current power configuration deviates significantly from the ideal state and requires substantial adjustments; when the deviation values ​​are already small, only minor adjustments are needed for convergence.

[0092] After obtaining the deviation values ​​for each unit, the power allocation ratio that needs to be adjusted for each unit is determined based on the sign and magnitude of the deviation values. For units with positive deviation values ​​(overload), their power share should be reduced by at least the magnitude of the deviation value; for units with negative deviation values ​​(light load), their power share should be increased by at least the absolute value of the deviation value. Since the total system power must be conserved (the sum of all shares is 100%), the reduction in the share of one unit must be compensated by the increase in the share of other units. Therefore, determining the adjustment ratio needs to satisfy the overall balance constraint.

[0093] In the specific implementation, the overload values ​​of all units with positive deviation values ​​are summed to obtain the total overload value; the absolute values ​​of the light load values ​​of all units with negative deviation values ​​are summed to obtain the total light load value. Since the total overload value should equal the total light load value (otherwise it indicates that the ideal configuration definition is unreasonable), this value can be directly used as the total adjustment amount. For each overloaded unit, its reduction amount is set to its deviation value (or the deviation value multiplied by a relaxation coefficient of 0.8~1.0) to avoid adjustment overshoot; for each light load unit, its increase amount is allocated according to its capacity or available margin ratio. After such allocation, a set of power adjustment step size vectors is obtained, where each component represents the percentage point that the unit should increase or decrease in the current round. This step outputs the adjustment direction (increase / decrease) and adjustment step size, i.e., the "power allocation ratio to be adjusted".

[0094] After determining the power adjustment step size for each unit, the optimal solution is approached through multiple iterations with the goal of minimizing the total system fluctuation. The total system fluctuation can be quantified by the sum of the variance, standard deviation, or absolute deviation of the load difference values ​​of all units. During iterative optimization, the current power share is added to the current adjustment step size to obtain a temporary new power share. To ensure physical feasibility, the new share needs to be constrained between the minimum adjustable output and the maximum allowable output of the unit (e.g., not lower than 0, not higher than the share after conversion of rated capacity). Furthermore, the load difference values ​​of each unit and the total system fluctuation index are recalculated based on the new share. If the total fluctuation index decreases compared to the previous round, the adjustment is accepted, and the new share is used as the current share for the next round; if the fluctuation index increases instead, the adjustment is rejected, the step size is reduced, or the adjustment is reversed before trying again. The above steps are repeated until the convergence condition is met, such as the decrease in the total fluctuation being less than a preset minimum threshold for several consecutive rounds, or the load difference values ​​of all units being lower than the preset load difference threshold. The power share at the final convergence is the adjusted power allocation ratio.

[0095] For example, with initial power shares of 20%, 23%, 18%, 28%, and 21%, assuming an ideal power share of 20% per unit (average distribution), the deviations would be 0%, +3%, -2%, +8%, and +1%, respectively. The total overload is 12%, and the total light load is 2%, indicating an imbalance that requires adjustment of the ideal configuration definition. After redefining the ideal shares according to capacity ratios, and through multiple iterations: the first round of adjustments reduced the overloaded unit 4 (from 28% to 25%) and increased the light load unit 3 (from 18% to 21%), resulting in new shares of 20%, 23%, 21%, 25%, and 21%. The second round further fine-tuned units 2 and 5, ultimately converging to 19%, 22%, 19%, 27%, and 23%. At this point, the load difference between each unit was less than the threshold, and the total system fluctuation decreased from the initial 26% to approximately 8%, achieving the optimization goal. This iterative process embodies a progressive optimization approach guided by minimizing total fluctuation, ensuring convergence while avoiding secondary fluctuations that might be caused by a single large adjustment.

[0096] Based on the first embodiment of this application, in the third embodiment of this application, the content that is the same as or similar to that in the first embodiment described above can be referred to the above description, and will not be repeated hereafter. Based on this, please refer to... Figure 3 Step S10 may include steps S101 to S102: Step S101: Collect the location information, capacity status and load data of each energy storage unit, and perform format consistency processing on the location information, capacity status and load data to obtain a formatted dataset.

[0097] Step S102: Transmit the formatted dataset to the coordination node, summarize and detect anomalies in the formatted dataset, and obtain the initial state record of each energy storage unit.

[0098] After collecting the original location information, capacity status, and load data of each energy storage unit, this embodiment performs format consistency processing on the data because the data formats of different units may be inconsistent (such as different coordinate representations and inconsistent capacity units). All data is converted into a unified data type and unit, forming a formatted dataset. This formatted dataset is transmitted to the central coordination node via a communication network. The coordination node summarizes and merges the data and performs anomaly detection on each data item, such as checking whether the capacity is within the 0%~100% range and whether the load is non-negative. The valid data that passes the anomaly detection is finally compiled into the initial state record for each energy storage unit.

[0099] For example, the microgrid contains 10 energy storage units. First, sensor modules deployed on each unit collect location information (e.g., Unit 1 is located at 39.9042°N, 116.4074°E), capacity status (current capacity 80%, total capacity 100kWh), and load data (current load 20kW) once per second. After collection, the raw data undergoes format consistency processing: location information is uniformly converted to decimal latitude and longitude format, capacity status is uniformly converted to a percentage value (80%), and load data is uniformly converted to kilowatt units (20kW). This data is then encapsulated into a JSON format data packet according to a unified node communication protocol, with fields defined as {"ID":1, "Location":{"Lat":39.9042, "Lon":116.4074}, "Capacity":80, "Load":20}, along with a CRC32 checksum (e.g., 0x4A3B2C1D). If the checksum fails, the data is automatically retransmitted, ensuring data integrity and format consistency, thus obtaining a formatted dataset.

[0100] Furthermore, these formatted datasets are transmitted to the central coordination node via an AES-128 encrypted channel. Upon receiving the data, the node decrypts and parses it, aggregates and merges the data from all units, and performs anomaly detection: checking whether the location coordinates are within the microgrid's geographical boundaries, whether the capacity status is between 0% and 100%, and whether the load data is non-negative and does not exceed the unit's maximum output capacity. For example, if the capacity of unit 2 is found to be only 60%, lower than 90% of the average capacity of all units (75%) (i.e., lower than 67.5%), it is marked as an anomaly and a re-collection is triggered. After verification, it is included as valid data. After aggregation and anomaly detection, the central node generates an initial state record for each energy storage unit, such as the record for unit 1 being "ID:1, Location: 39.9042°N, 116.4074°E, Capacity Ratio: 80%, Load: 20kW", and stores it in the database. Through the above process, the entire process from data acquisition, formatting, transmission to aggregation and anomaly detection is completed, obtaining accurate and reliable initial state records for each energy storage unit.

[0101] In one implementation, step S30 includes steps S301 to S304: Step S301: Extract compensation coefficient values ​​that deviate from the preset threshold range from the compensation coefficient data as key adjustment parameters, perform data consistency verification on the key adjustment parameters, and obtain the verified compensation coefficients.

[0102] In the specific implementation, after generating the compensation coefficient data, this embodiment further extracts key adjustment parameters from it, such as the specific value of a certain unit compensation coefficient deviating from the normal range. Data consistency verification is performed on these key adjustment parameters, that is, checking whether the parameters have been tampered with or lost during transmission and storage. Verification is usually carried out by check and comparison or digital signature verification. After verification, a reliable verified compensation coefficient is obtained.

[0103] Step S302: Determine whether the verified compensation coefficient exceeds a preset threshold. If it does, generate an adjustment command based on the key adjustment parameters and send it to obtain sending status feedback.

[0104] Determine whether the compensation coefficient after verification exceeds the preset threshold range (e.g., the normal range of 0.9~1.1). If it does, it indicates that the unit is in an abnormal state. The system generates a targeted adjustment command based on the deviation value of the excess part (e.g., "adjust the compensation coefficient from 1.25 to 1.0") and sends it to the corresponding energy storage unit through the wireless communication channel. At the same time, it receives the transmission status feedback returned by the unit (e.g., confirmation of receipt or no response after timeout).

[0105] Step S303: Optimize transmission efficiency based on the transmission status feedback to obtain the adjusted interactive information stream.

[0106] Based on the sending status feedback, the system assesses the latency and packet loss of the current communication network, and optimizes the transmission efficiency by dynamically adjusting the data packet size or enabling the retransmission mechanism to ensure that the command arrives reliably, thereby obtaining the adjusted interactive information flow, which includes the corrected compensation coefficient and the response confirmation of each unit.

[0107] Step S304: Determine the initial power allocation ratio of each energy storage unit based on the adjusted interactive information flow, and calculate the load difference value of each energy storage unit in conjunction with the load data.

[0108] Based on the correction values ​​in the adjusted interactive information flow, the initial power allocation ratio of each energy storage unit is redefined, and then the load difference value of each unit is calculated in conjunction with real-time load data. The load difference value is equal to the difference between the actual load and the theoretical allocated power, and is used to determine whether a power reallocation process needs to be triggered.

[0109] This application collects the capacity, load, and location status of each energy storage unit through a distributed sensor network, and generates cross-unit compensation coefficients based on distributed differential calculations. It combines data verification and communication delay optimization to issue adjustment commands, detects system fluctuations caused by load differences in real time, and dynamically redistributes power shares using group iterative logic. Through voltage and frequency closed-loop feedback, local fluctuation fault-tolerant recovery, and adaptive adjustment using interactive strategies, a stable and coordinated control sequence is formed. Finally, by combining convergence verification of key black-start indicators and data fusion evaluation, it achieves coordinated equilibrium, fluctuation suppression, and smooth and orderly recovery of multi-area energy storage clusters during the microgrid black-start process, improving the stability and reliability of power supply restoration in isolated grid scenarios.

[0110] This application also provides an intelligent control system for a distribution area energy storage system for black start of a microgrid. The system includes: an initial state record acquisition module, a compensation coefficient generation module, an initial power allocation determination module, a power allocation ratio adjustment module, an updated state acquisition module, a state diagram construction module, a deviation determination module, an interaction frequency adjustment module, and a coordinated control generation module. The initial state record acquisition module is used to acquire the initial state records of each energy storage unit in the microgrid. The initial state records include location information, capacity status, and load data. The compensation coefficient generation module is used to perform distributed calculations on the location and capacity differences between each energy storage unit based on the initial state record, and generate compensation coefficient data. The compensation coefficient data characterizes the relative deviation of the energy storage unit in the energy storage system. The compensation coefficient data is the proportion of the comprehensive deviation value of each energy storage unit to the sum of the comprehensive deviation values ​​of all energy storage units. The comprehensive deviation value is obtained by weighting the location and capacity differences of each energy storage unit. The initial power allocation determination module is used to determine the initial power allocation ratio of each energy storage unit based on the inverse relationship between the compensation coefficient data and the power allocation share, and to determine the load difference value of each energy storage unit based on the initial power allocation ratio and the load data. A power allocation ratio adjustment module is used to adjust the initial power allocation ratio based on the load difference value, obtain the adjusted power allocation ratio, and determine the balanced power output configuration based on the adjusted power allocation ratio. The update status acquisition module is used to acquire the updated operating status of each energy storage unit according to the power output configuration. The updated operating status includes the voltage response value and frequency response value of each energy storage unit. The state diagram construction module is used to construct a time-seriesd operating state diagram by recording the voltage response value and the frequency response value in chronological order. The deviation determination module is used to execute a monitoring loop based on the running state diagram to determine the deviation between the voltage response value and the corresponding voltage response value in the initial state record, and between the frequency response value and the corresponding frequency response value in the initial state record; The interaction frequency adjustment module is used to adjust the information interaction frequency between each energy storage unit according to the deviation amount and generate an adjustment record. The adjustment record includes: the timestamp of the adjustment, the voltage deviation and frequency deviation of each energy storage unit that triggered the adjustment, the information interaction frequency value of each energy storage unit before the adjustment, and the information interaction frequency value of each energy storage unit after the adjustment. The adjustment record is saved to the distributed storage node. The coordination control generation module is used to generate a coordination control sequence based on the adjustment record through rule extraction or sequence pattern mining methods. The coordination control sequence is used to control the power output of each energy storage unit to achieve stable power distribution and suppression of system fluctuations during the black start process of the microgrid.

[0111] The intelligent control system for a distributed energy storage system for microgrid black start provided in this application adopts the intelligent control method for a distributed energy storage system for microgrid black start in the above embodiments, which can solve the technical problems of poor information interaction between units and power distribution imbalance leading to system fluctuations and local instability. Compared with the prior art, the beneficial effects of the intelligent control system for a distributed energy storage system for microgrid black start provided in this application are the same as the beneficial effects of the intelligent control method for a distributed energy storage system for microgrid black start provided in the above embodiments, and other technical features in the intelligent control system for a distributed energy storage system for microgrid black start are the same as the features disclosed in the methods of the above embodiments, and will not be repeated here.

[0112] It should be noted that, in this document, the terms "comprising," "including," and any other variations are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Specific examples have been used in this document to illustrate the principles and implementation methods of the present invention. These examples are only intended to aid in understanding the method and core ideas of the present invention. The above descriptions are merely preferred embodiments of the present invention. It should be pointed out that, due to the limitations of written expression and the objective existence of infinite specific structures, those skilled in the art can make various improvements, modifications, or variations without departing from the principles of the present invention, and can also combine the above technical features in an appropriate manner. These improvements, modifications, variations, or combinations, or the direct application of the concept and technical solution of the present invention to other situations without modification, should all be considered within the scope of protection of the present invention.

Claims

1. A smart control method for a distribution network energy storage system for black start in a microgrid, characterized in that, The method includes: Acquire the initial state records of each energy storage unit in the microgrid, the initial state records including location information, capacity status and load data; Based on the initial state record, distributed calculations are performed on the location and capacity differences between each energy storage unit to generate compensation coefficient data. The compensation coefficient data characterizes the relative deviation of each energy storage unit in the energy storage system. The compensation coefficient data is the proportion of the comprehensive deviation value of each energy storage unit to the sum of the comprehensive deviation values ​​of all energy storage units. The comprehensive deviation value is obtained by weighting the location and capacity differences of each energy storage unit. The initial power allocation ratio of each energy storage unit is determined based on the inverse relationship between the compensation coefficient data and the power allocation share, and the load difference value of each energy storage unit is determined based on the initial power allocation ratio and the load data. The initial power allocation ratio is adjusted based on the load difference value to obtain the adjusted power allocation ratio, and the balanced power output configuration is determined based on the adjusted power allocation ratio. The updated operating status of each energy storage unit is obtained according to the power output configuration, and the updated operating status includes the voltage response value and frequency response value of each energy storage unit. A time-series operating state diagram is constructed by recording the voltage response value and the frequency response value in chronological order. Based on the operating state diagram, a monitoring loop is executed to determine the deviation between the voltage response value and the corresponding voltage response value in the initial state record, and between the frequency response value and the corresponding frequency response value in the initial state record. The information interaction frequency between each energy storage unit is adjusted according to the deviation amount, and an adjustment record is generated. The adjustment record includes: the timestamp of the adjustment, the voltage deviation and frequency deviation of each energy storage unit that triggered the adjustment, the information interaction frequency value of each energy storage unit before the adjustment, and the information interaction frequency value of each energy storage unit after the adjustment. The adjustment record is saved to the distributed storage node. Based on the adjustment records, a coordinated control sequence is generated through rule extraction or sequence pattern mining methods. The coordinated control sequence is used to control the power output of each energy storage unit to achieve stable power distribution and suppression of system fluctuations during the black start process of the microgrid.

2. The intelligent control method for distribution area energy storage systems for microgrid black start as described in claim 1, characterized in that, Based on the initial state record, the distributed calculation of the location and capacity differences between each energy storage unit is performed to generate compensation coefficient data, including: Based on the initial state record, a synchronous update mechanism is used to perform consistency calibration on the data of each energy storage unit to obtain a calibrated difference value matrix. Based on the difference value matrix, a weighted analysis is performed on the location and capacity differences between each energy storage unit to obtain the comprehensive deviation value of each energy storage unit. Based on the comprehensive deviation value, compensation coefficient data are generated using the normalized ratio method.

3. The intelligent control method for distribution area energy storage systems for microgrid black start as described in claim 2, characterized in that, The step involves performing a weighted analysis of the location and capacity differences among the energy storage units based on the difference value matrix to obtain a comprehensive deviation value for each energy storage unit, including: Determine the location weight and capacity weight; The location deviation value is calculated based on the location weight and the location difference between any two energy storage units, and the capacity deviation value is calculated based on the capacity weight and the capacity difference between the two energy storage units; The position deviation value and the capacity deviation value are summed to obtain the paired comprehensive deviation value between the two energy storage units; The overall deviation value of each energy storage unit is obtained by summing the paired comprehensive deviation values ​​between each energy storage unit and all other energy storage units.

4. The intelligent control method for distribution area energy storage systems for microgrid black start as described in claim 1, characterized in that, The step of adjusting the initial power allocation ratio based on the load difference value to obtain the adjusted power allocation ratio, and determining the balanced power output configuration based on the adjusted power allocation ratio, includes: Determine whether the load difference value is greater than a preset load difference threshold; If the load difference value is greater than the preset load difference threshold, the power share is redistributed through iterative optimization to obtain the adjusted power allocation ratio, and the balanced power output configuration is determined based on the adjusted power allocation ratio.

5. The intelligent control method for distribution area energy storage systems for microgrid black start as described in claim 4, characterized in that, The step of iteratively optimizing the redistribution of power shares to obtain an adjusted power allocation ratio includes: Obtain the deviation between the current power share and the ideal power configuration; The required power allocation ratio is determined based on the deviation value; With the goal of minimizing the total system fluctuation, the power share is iteratively optimized to obtain the adjusted power allocation ratio.

6. The intelligent control method for distribution area energy storage systems for microgrid black start as described in claim 1, characterized in that, The initial state records of each energy storage unit in the microgrid are obtained, and the initial state records include location information, capacity status, and load data, including: The location information, capacity status, and load data of each energy storage unit are collected, and the location information, capacity status, and load data are processed to ensure format consistency, thereby obtaining a formatted dataset. The formatted dataset is transmitted to the coordination node, where it is aggregated and anomaly detected to obtain the initial state record of each energy storage unit.

7. The intelligent control method for distribution area energy storage systems for microgrid black start as described in claim 1, characterized in that, The process of determining the initial power allocation ratio of each energy storage unit based on the inverse relationship between the compensation coefficient data and the power allocation share, and determining the load difference value of each energy storage unit based on the initial power allocation ratio and the load data, includes: The compensation coefficient values ​​that deviate from the preset threshold range are extracted from the compensation coefficient data as key adjustment parameters. The data consistency of the key adjustment parameters is verified to obtain the verified compensation coefficients. Determine whether the compensation coefficient after verification exceeds a preset threshold. If it does, generate an adjustment instruction based on the key adjustment parameters and send it to obtain the sending status feedback. Based on the aforementioned sending status feedback, transmission efficiency is optimized to obtain the adjusted interactive information stream; The initial power allocation ratio of each energy storage unit is determined based on the adjusted interactive information flow, and the load difference value of each energy storage unit is calculated in combination with the load data.

8. The intelligent control method for distribution area energy storage systems for microgrid black start as described in claim 1, characterized in that, The step of adjusting the information exchange frequency between each energy storage unit based on the deviation and generating an adjustment record includes: A dynamic information interaction graph is constructed using each energy storage unit as a node, the normalized value of the deviation as the node feature, and the reciprocal of the difference in voltage response phase difference and frequency response rate of change between units as the edge weight. Calculate the algebraic connectivity of the Laplacian matrix of the dynamic information interaction graph, whereby the algebraic connectivity characterizes the system's cooperative ability to suppress bias under the current information interaction frequency configuration. Based on the difference between the algebraic connectivity and the preset target connectivity, the information interaction frequency adjustment vector that maximizes the algebraic connectivity increment is solved. Each component in the adjustment vector corresponds to a communication frequency adjustment coefficient between an energy storage unit and its neighboring units. The information interaction frequency of each energy storage unit is updated according to the adjustment vector, and the deviation distribution before adjustment, the adjustment vector, the algebraic connectivity after adjustment, and the timestamp of this adjustment are jointly encoded into a structured adjustment record. After the adjustment records are compressed using a columnar compression algorithm, they are appended to the immutable log file of the distributed storage node. The immutable log file is stored in shards according to the time span, and each shard contains a Bloom filter. The bit array of the Bloom filter is constructed based on the globally incrementing sequence number of all adjustment records in the shard.

9. The intelligent control method for a distribution area energy storage system for microgrid black start as described in claim 8, characterized in that, The step of updating the information interaction frequency of each energy storage unit according to the adjustment vector, and jointly encoding the deviation distribution before adjustment, the adjustment vector, the algebraic connectivity after adjustment, and the timestamp of this adjustment into a structured adjustment record, includes: The unadjusted deviation distribution is serialized into a one-dimensional feature vector, the adjustment vector is quantized into a fixed-width integer increment sequence, the algebraic connectivity metric is quantized into a single-precision floating-point number, and the timestamp is converted into a millisecond-level integer in Coordinated Universal Time to obtain serialized data. The serialized data is aggregated into a binary payload according to a predetermined field order, and the cyclic redundancy check code or hash value of the binary payload is calculated as an integrity digest. The binary payload is concatenated with the integrity summary, and the concatenation result is appended with the globally incrementing sequence number of this adjustment record to form a structured adjustment record.

10. An intelligent control system for a distribution area energy storage system for black start in a microgrid, characterized in that, The system includes: an initial state record acquisition module, a compensation coefficient generation module, an initial power allocation determination module, a power allocation ratio adjustment module, an updated state acquisition module, a state diagram construction module, a deviation determination module, an interaction frequency adjustment module, and a coordination control generation module. The initial state record acquisition module is used to acquire the initial state records of each energy storage unit in the microgrid. The initial state records include location information, capacity status, and load data. The compensation coefficient generation module is used to perform distributed calculations on the location and capacity differences between each energy storage unit based on the initial state record, and generate compensation coefficient data. The compensation coefficient data characterizes the relative deviation of the energy storage unit in the energy storage system. The compensation coefficient data is the proportion of the comprehensive deviation value of each energy storage unit to the sum of the comprehensive deviation values ​​of all energy storage units. The comprehensive deviation value is obtained by weighting the location and capacity differences of each energy storage unit. The initial power allocation determination module is used to determine the initial power allocation ratio of each energy storage unit based on the inverse relationship between the compensation coefficient data and the power allocation share, and to determine the load difference value of each energy storage unit based on the initial power allocation ratio and the load data. A power allocation ratio adjustment module is used to adjust the initial power allocation ratio based on the load difference value, obtain the adjusted power allocation ratio, and determine the balanced power output configuration based on the adjusted power allocation ratio. The update status acquisition module is used to acquire the updated operating status of each energy storage unit according to the power output configuration. The updated operating status includes the voltage response value and frequency response value of each energy storage unit. The state diagram construction module is used to construct a time-seriesd operating state diagram by recording the voltage response value and the frequency response value in chronological order. The deviation determination module is used to execute a monitoring loop based on the running state diagram to determine the deviation between the voltage response value and the corresponding voltage response value in the initial state record, and between the frequency response value and the corresponding frequency response value in the initial state record; The interaction frequency adjustment module is used to adjust the information interaction frequency between each energy storage unit according to the deviation amount and generate an adjustment record. The adjustment record includes: the timestamp of the adjustment, the voltage deviation and frequency deviation of each energy storage unit that triggered the adjustment, the information interaction frequency value of each energy storage unit before the adjustment, and the information interaction frequency value of each energy storage unit after the adjustment. The adjustment record is saved to the distributed storage node. The coordination control generation module is used to generate a coordination control sequence based on the adjustment record through rule extraction or sequence pattern mining methods. The coordination control sequence is used to control the power output of each energy storage unit to achieve stable power distribution and suppression of system fluctuations during the black start process of the microgrid.

Images