A power response and distributed optimization based storage and charging collaborative control method and system

Abstract

The present application relates to the technical field of cooperative control, and particularly relates to a kind of based on power response and distributed optimization's storage and filling cooperative control method and system.The method comprises the following steps: obtaining microgrid operation data;According to microgrid operation data, flexible load slice is carried out, and flexible load data is obtained;Flexible load data is evolved in heterogeneous communication timing, and heterogeneous communication data is obtained;According to heterogeneous communication data, virtual synchronous machine control is carried out, and synchronization compensation data is obtained;According to synchronization compensation data, cooperative optimization is carried out, and cooperative control data is obtained.The present application is jointly modeled by responding to the characteristics of flexible load and heterogeneous communication timing, and combined with virtual synchronous machine compensation mechanism and distributed cooperative optimization, realizes the time sequence coordination regulation between energy storage system and charging load, thereby effectively reduces the power over-compensation and control oscillation problem caused by equipment response difference, improves the power regulation stability and storage and filling cooperative control efficiency of microgrid in communication environment.

Description

Technical Field

[0001] This invention relates to the field of collaborative control technology, and in particular to a collaborative control method and system for energy storage and charging based on power response and distributed optimization. Background Technology

[0002] With the rapid development of new energy power generation and electric vehicles, a large number of distributed photovoltaic, battery energy storage devices, and electric vehicle charging piles are being connected to park-level or regional-level microgrid systems. This is causing the microgrid's operating structure to gradually shift from the traditional unidirectional power supply mode to a multi-source, multi-load, and multi-communication link operating mode. In such systems, energy storage devices are typically used for rapid power regulation, while electric vehicle charging loads have a certain degree of flexible regulation capability. The two have strong synergistic potential in peak shaving, frequency regulation, and load balancing.

[0003] In existing technologies, the coordinated control of energy storage systems and charging loads is typically achieved through centralized scheduling or simple demand response strategies. For example, charging power is uniformly adjusted based on load forecasts, or short-term power compensation is performed using energy storage systems. However, in actual microgrid operation, different devices often access the control system through various communication links, such as industrial Ethernet, wireless communication networks, or fieldbuses. These links exhibit significant differences in communication latency, data jitter, and control cycles. Furthermore, the control response speeds of different devices also vary considerably. For instance, energy storage systems can complete power adjustment in milliseconds or hundreds of milliseconds, while charging loads often require several seconds or even longer to adjust their power.

[0004] Due to differences in communication latency and varying device response rhythms, when the system executes unified scheduling commands, some devices may have already completed power regulation, while others may still be waiting or responding slowly, creating a control response gap in a short period. This response mismatch can easily lead to instantaneous power overcompensation or secondary power fluctuations in the system, thereby causing grid frequency disturbances or even control oscillations. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention proposes a power response and distributed optimization-based energy storage and charging coordinated control method and system, thereby resolving at least one of the aforementioned technical problems.

[0006] This application provides a power response and distributed optimization-based energy storage and charging coordinated control method, the method comprising:

[0007] S1. Obtain microgrid operation data; perform flexible load slicing based on microgrid operation data to obtain flexible load data;

[0008] S2. Perform heterogeneous communication time-series evolution on the flexible load data to obtain heterogeneous communication data;

[0009] S3. Perform virtual synchronization machine control based on heterogeneous communication data to obtain synchronization compensation data;

[0010] S4. Perform collaborative optimization based on the synchronization compensation data to obtain collaborative control data.

[0011] This invention utilizes flexible load slicing to process microgrid operating data, transforming dispersed load behavior into adjustable structured load units, thereby enhancing the system's ability to identify and regulate dynamic load characteristics. Through heterogeneous communication timing evolution analysis, it identifies time differences and response rhythms in control signal propagation under different communication links, providing a more realistic description of the communication environment. Virtual synchronous machine control, through inertia and damping compensation, dynamically buffers power fluctuations caused by communication timing mismatches, thereby improving system frequency stability and short-term power support capabilities. Distributed collaborative optimization enables unified coordination and scheduling of various energy storage and charging loads, allowing different regulation resources to achieve orderly relay and coordinated response at both the time and power levels. This invention effectively reduces the impact of communication delay differences and asynchronous load responses on grid stability, enhances the microgrid's regulation capabilities and control robustness in complex operating environments, and improves the collaborative utilization efficiency of energy storage resources and flexible loads.

[0012] Optionally, the flexible load slice includes:

[0013] Time envelope data is obtained by extracting the time envelope from the microgrid operation data.

[0014] Discretize and slice the time envelope data to obtain load slice data;

[0015] The load slice data is subjected to similarity measurement to obtain load similarity data;

[0016] Slice and aggregate the load similarity data to obtain slice aggregated data;

[0017] Flexible spatial reconstruction is performed based on the sliced ​​aggregated data to obtain flexible load data.

[0018] This invention extracts the time envelope of microgrid operation data and performs discretization slicing, transforming continuously changing load curves into structured time segments, thus effectively representing the power variation characteristics of the load at different time stages. Through similarity measurement and slice aggregation, load segments with similar operating patterns are categorized and integrated, enabling dispersed load behaviors to form a stable pattern structure in the time dimension. Flexible spatial reconstruction maps the time aggregation results to a set of flexible loads with adjustment capabilities, allowing the system to more accurately identify adjustable load resources and their spatial distribution characteristics. This not only improves the accuracy of flexible load identification but also provides a load structure foundation for energy storage and charging coordinated control.

[0019] Optionally, flexible spatial reconstruction includes:

[0020] Based on the sliced ​​aggregated data, flexibility node mapping is performed to obtain load node data;

[0021] Node associations are constructed based on load node data to obtain load association data;

[0022] Flexible spatial structures are generated based on load correlation data to obtain flexible structure data;

[0023] The flexible load data is obtained by integrating the adjustment capabilities based on the flexible structure data.

[0024] This invention maps sliced ​​aggregated data to flexible nodes and establishes relationships between these nodes, enabling previously dispersed load fragments to form a structurally characteristic load network at the spatial level, thereby revealing the collaborative regulation relationships between different flexible loads. Through the generation and integration of flexible spatial structures and regulation capabilities, the adjustable power range, response capability, and collaborative characteristics of each node can be uniformly represented, transforming flexible load resources from discrete units into a structured load set with overall regulation capabilities. This invention improves the systematic nature of flexible load identification and the accuracy of regulation capability assessment, providing a spatial regulation foundation for coordinated storage and charging control.

[0025] Optionally, S2 includes:

[0026] Communication link identification is performed on the flexible load data to obtain communication link data;

[0027] Based on the communication link data, communication delay features are extracted to obtain delay feature data;

[0028] Communication timing propagation is performed based on communication delay characteristic data to obtain timing propagation data;

[0029] The time-series propagation data is reconstructed by time-series folding to obtain heterogeneous communication data.

[0030] This invention identifies communication links and extracts communication delay features from flexible load data, accurately representing the signal transmission characteristics of different devices in heterogeneous communication environments, thereby revealing the time differences in the propagation of control commands by various load nodes. Through communication timing propagation analysis and timing folding reconstruction, dispersed communication propagation trajectories are mapped to the same time structure, forming heterogeneous communication data that reflects the rhythm and propagation patterns of communication response. This invention not only improves the ability to identify communication timing differences but also provides a more accurate timing basis for virtual synchronous machine compensation control, thereby enhancing the system's coordinated adjustment capability in communication environments.

[0031] Optionally, time-series folding reconstruction includes:

[0032] The propagation time reference is extracted from the time-series propagation data to obtain the time reference data.

[0033] Based on the time reference data, a propagation time normalization mapping is performed to obtain time-normalized data;

[0034] The propagation trajectory is folded based on the time-normalized data to obtain folded trajectory data.

[0035] Heterogeneous propagation structure is reconstructed based on folded trajectory data to obtain heterogeneous communication data.

[0036] This invention extracts the propagation time reference and performs time normalization mapping on the time-series propagation data, unifying the different signal propagation times under different communication links to the same time reference frame, thereby eliminating the impact of link delay differences on time-series analysis. Through propagation trajectory folding and heterogeneous propagation structure reconstruction, the dispersed communication propagation paths are transformed into a unified propagation model with structural relationships, enabling the communication response rhythms of various devices to be expressed within a unified time-series structure. This invention can more accurately reflect the time-series propagation characteristics in heterogeneous communication environments, providing a communication time-series basis for control compensation.

[0037] Optionally, S3 includes:

[0038] Based on heterogeneous communication data, communication timing mismatch is identified to obtain timing mismatch data;

[0039] The response window interval is calibrated based on the time-series mismatch data to obtain the window interval data;

[0040] Virtual inertia is dynamically mapped based on the data in the empty window interval to obtain inertia compensation data;

[0041] Virtual damping collaborative shaping is performed based on inertia compensation data to obtain damping compensation data;

[0042] Synchronous phase angle traction control is performed based on damping compensation data and local power deviation data to obtain phase angle compensation data;

[0043] Synchronous compensation data is obtained by combining inertia compensation data, damping compensation data, and phase angle compensation data.

[0044] This invention identifies communication timing mismatches in heterogeneous communication data and calibrates response window intervals, accurately pinpointing the time gap between control command propagation and actual device response, thus providing clear timing for control compensation. Through dynamic virtual inertia mapping and virtual damping shaping, power fluctuations generated during the response window phase are dynamically buffered and stabilized. Synchronous phase angle traction control is implemented in conjunction with local power deviations, ensuring coordinated responses from various regulatory resources in terms of time and phase angle. This effectively suppresses power oscillations caused by communication delay differences, improving the stability and control robustness of the microgrid during the coordinated regulation of energy storage and charging.

[0045] Optionally, the response window interval calibration includes:

[0046] Response time offset is analyzed based on timing mismatch data to obtain response offset data;

[0047] The adjustment status is determined based on the response offset data to obtain the response status data.

[0048] Power gap detection is performed based on the response status data to obtain gap identification data;

[0049] By aggregating continuous time intervals based on the gap identification data, the gap interval data is obtained.

[0050] This invention analyzes the response time offset of timing mismatch data to accurately represent the time difference between the arrival of control commands and the actual adjustment of the equipment, thus providing a quantitative basis for timing analysis of the adjustment process. By determining the adjustment status and detecting power gaps, it can identify power deviations that have not yet been compensated during the adjustment process. Furthermore, by aggregating continuous time intervals, it extracts stable intervals for these power gaps, forming response window interval data. This allows for precise location of adjustment gaps in the control response, providing clear time boundaries for compensation control, thereby improving control stability and response coordination during the coordinated adjustment of energy storage and charging.

[0051] Optionally, virtual damping co-shaping includes:

[0052] Based on the inertia compensation data, the inertia action range is analyzed to obtain the inertia range data;

[0053] Dynamic damping base value matching is performed based on inertia interval data to obtain damping base value data;

[0054] By backfilling and superimposing risk corrections on the damping baseline data, damping correction data is obtained.

[0055] The damping correction data is smoothed and shaped to obtain the damping trajectory data;

[0056] The damping trajectory data is encapsulated into parameters to obtain damping compensation data.

[0057] This invention analyzes the inertia effect range of inertia compensation data to identify key segments where virtual inertia affects the system's dynamic response at different time stages, thus providing a timing basis for damping adjustment. Through dynamic damping base value matching and risk correction for superposition, the risk of power superposition caused by the simultaneous response of multiple adjustment resources can be suppressed in advance. Through damping trajectory smoothing and parameter encapsulation processing, a continuous and stable control trajectory is formed during the damping adjustment process. This effectively suppresses power oscillations and control oscillations during the adjustment process, improving the dynamic stability and adjustment smoothness of the energy storage-charging coordinated control system.

[0058] Optionally, S4 includes:

[0059] Based on the synchronous compensation data, compensation responsibility is stratified to obtain stratified responsibility data;

[0060] Based on the hierarchical responsibility data, a time-series relay window is constructed to obtain relay window data;

[0061] Based on the relay window data, suppression constraints are generated by backfilling and superimposing suppression constraints to obtain suppression constraint data;

[0062] Local agent optimization is performed based on suppression constraint data and equipment adjustment boundary data to obtain local cooperative strategy data;

[0063] Based on the local collaborative strategy data, cross-node collaborative correction is performed to obtain global collaborative allocation data;

[0064] Compensation and exit shaping are performed based on the global collaborative allocation data to obtain collaborative control data.

[0065] This invention employs a compensation responsibility hierarchy and a time-series relay window construction on synchronous compensation data to allocate the regulation tasks of different energy storage units and flexible loads according to time sequence, enabling each regulation resource to form an orderly relay response relationship at different stages. Through the generation of backfill superposition suppression constraints and local proxy optimization processing, the power overcompensation problem caused by simultaneous backfilling of multiple devices can be effectively avoided. Through cross-node consistency correction and compensation exit shaping, the regulation strategies of each node remain coordinated at the global level, thereby improving the scheduling stability and system operational reliability during the energy storage-charging coordinated control process.

[0066] Optionally, this application also provides a power response and distributed optimization-based energy storage and charging coordinated control system for executing the power response and distributed optimization-based energy storage and charging coordinated control method described above. The power response and distributed optimization-based energy storage and charging coordinated control system includes:

[0067] The flexible load slicing module is used to acquire microgrid operation data; and to perform flexible load slicing based on the microgrid operation data to obtain flexible load data.

[0068] The heterogeneous communication timing evolution module is used to perform heterogeneous communication timing evolution on flexible load data to obtain heterogeneous communication data.

[0069] The virtual synchronizer control module is used to control the virtual synchronizer based on heterogeneous communication data to obtain synchronization compensation data.

[0070] The collaborative optimization module is used to perform collaborative optimization based on the synchronization compensation data to obtain collaborative control data.

[0071] The purpose of this invention is to improve the system's accuracy in identifying adjustable load resources by performing flexible load slicing processing on microgrid operating data, breaking down continuously changing load curves into load segments with similar behavioral characteristics, and forming a flexible load set with adjustment capabilities through spatial reconstruction. By performing heterogeneous communication timing evolution analysis on the flexible load data, the delay differences and propagation patterns of different devices under various communication link environments are identified, providing a more realistic communication response model for control strategies. Virtual synchronous machine control, through inertia compensation, damping shaping, and phase angle traction, dynamically buffers power fluctuations caused by communication timing mismatches, thereby improving the system's frequency stability under short-term disturbance conditions. Distributed collaborative optimization is used to hierarchically schedule compensation responsibilities, and combined with timing relay windows and superimposed suppression constraints, to achieve orderly relay regulation between energy storage and charging loads. This effectively reduces power oscillation problems caused by communication delay differences and asynchronous device responses, improving the collaborative control capability and operational stability of the microgrid in complex operating environments. Attached Figure Description

[0072] Other features, objects, and advantages of this application will become more apparent from the following detailed description of the non-limiting embodiments, taken with reference to the accompanying drawings:

[0073] Figure 1 A flowchart illustrating the steps of a power response and distributed optimization-based energy storage and charging coordinated control method according to one embodiment is shown.

[0074] Figure 2 A flowchart illustrating the steps of a flexible load slicing method according to an embodiment is shown.

[0075] Figure 3 A flowchart illustrating the steps of a heterogeneous communication timing evolution method according to one embodiment is shown.

[0076] Figure 4 A flowchart illustrating the steps of a heterogeneous communication timing evolution method according to one embodiment is shown.

[0077] Figure 5 A flowchart illustrating the steps of a collaborative optimization method according to one embodiment is shown.

[0078] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0079] The technical method of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0080] Furthermore, the accompanying drawings are merely illustrative of the invention and are not necessarily drawn to scale. Functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor methods and / or microcontroller methods.

[0081] It should be understood that although the terms "first," "second," etc., may be used herein to describe various units, these units should not be limited by these terms. These terms are used merely to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, a first unit may be referred to as a second unit, and similarly, a second unit may be referred to as a first unit. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0082] Please see Figures 1 to 5 This application provides a power response and distributed optimization-based energy storage and charging coordinated control method, the method comprising:

[0083] S1. Obtain microgrid operation data; perform flexible load slicing based on microgrid operation data to obtain flexible load data;

[0084] In one embodiment, the system collects operational data from energy storage devices, charging piles, adjustable air conditioning groups, and distributed photovoltaic inverters within the park through a microgrid energy management system. This operational data includes node power, equipment operating status, adjustment limits, communication node identifiers, and sampling timestamps. The system continuously samples the power of each device at a preset sampling period (e.g., 1 second or 5 seconds) to form a load power time series. The system extracts fluctuation boundaries from this time series and uses a sliding time window to statistically analyze the maximum and minimum power changes within each time period. After obtaining the power fluctuation boundaries, the system divides the continuous power series into time slices according to a preset time length (e.g., 30 seconds or 60 seconds), discretizing the load change process into multiple time segments. The system performs similarity analysis on the average power change trend, power fluctuation range, and adjustable power range of each time segment, and aggregates the time segments based on a preset similarity threshold to form multiple load structure clusters with similar adjustment characteristics. Each load structure cluster corresponds to a uniformly adjustable flexible load unit, thereby generating flexible load data.

[0085] S2. Perform heterogeneous communication time-series evolution on the flexible load data to obtain heterogeneous communication data;

[0086] In one embodiment, the system identifies the corresponding communication link type of the flexible load node based on its device communication interface information, including Ethernet links, 4G wireless links, and industrial bus links, and establishes the communication topology between the nodes. The system periodically sends test control commands and records device response times, statistically analyzes the transmission characteristics of each communication link, and extracts features such as average communication delay, maximum communication delay, and delay fluctuation range as communication delay characteristics. After obtaining the communication delay characteristics, the system constructs the control command propagation path based on the communication topology and records the arrival time sequence of commands between nodes to form a communication propagation trajectory. Using the node with the shortest propagation time as the time base, the system performs a unified time base conversion on the propagation times of different links and maps each propagation trajectory onto the same time axis for organization, thereby forming a communication propagation structure. Through the above processing, heterogeneous communication data that can represent the communication rhythm and link propagation differences of each device is obtained.

[0087] S3. Perform virtual synchronization machine control based on heterogeneous communication data to obtain synchronization compensation data;

[0088] In one embodiment, the system analyzes the response timing differences of various devices based on heterogeneous communication data. By comparing the control command sending time with the actual device response time, it calculates the time offset of the device response. When the time offset exceeds a preset threshold (e.g., 500 milliseconds), the system determines that there is a response gap in the current communication process and records the start and end times of this gap. After identifying the gap, the system dynamically adjusts the inertia parameters of the virtual synchronizer based on the duration of the gap. Specifically, it uses a preset base inertia parameter as the initial value and amplifies the inertia accordingly based on the increase ratio of the gap duration, so that the system still has stable inertia support capability during the communication response lag phase. , This is the current equivalent inertia parameter of the virtual synchronizer, representing its ability to provide inertial support for system frequency or power fluctuations within the current control cycle. These are the basic inertia parameters of the virtual synchronizer, which are the standard inertia values ​​preset under normal system operation and used as the initial reference values ​​for dynamic inertia adjustment. This is the inertia adjustment factor or inertia amplification ratio factor, representing the degree of influence of communication response lag on inertia gain. This factor can be preset based on system operating experience or simulation results, with values ​​ranging from 0.2 to 1.5. The response window duration is defined as the time interval between the issuance of a control command and the generation of an effective regulatory response by the equipment. The system matches the corresponding virtual damping coefficient based on the updated inertia parameters and performs smooth transition processing on the damping change process. After completing the adjustment of the inertia and damping parameters, the system calculates the phase angle adjustment based on the power deviation of each node and drives the energy storage device or adjustable load for power compensation via the phase angle adjustment command. The inertia compensation parameters, damping compensation parameters, and phase angle adjustment parameters are encapsulated to generate synchronous compensation data.

[0089] S4. Perform collaborative optimization based on the synchronization compensation data to obtain collaborative control data.

[0090] In one embodiment, the system assigns compensation responsibilities to participating devices based on synchronous compensation data. Fast-response energy storage devices are classified as a fast-response layer, while adjustable loads such as charging piles are classified as a slow-response layer. A phased adjustment time-series relay window is constructed based on the response time characteristics of each device. For example, an initial short time interval is allocated to energy storage devices for fast power compensation, while subsequent time intervals are allocated to charging pile loads for continuous power adjustment. The system generates a power compensation overlay suppression constraint to limit multiple devices from simultaneously performing power compensation operations within the same time window, thereby reducing power fluctuations caused by overcompensation. For instance, based on the expected response time, target compensation power, and duration of each device in the synchronous compensation data, the system extrapolates the power change process of each device on a unified time axis. When the system detects that multiple devices may simultaneously perform power compensation within the same time window, and the expected total compensation power after overlay may exceed the current target power gap or exceed a preset safety margin, the system generates a power compensation overlay suppression constraint. For example, the system can limit the number of slow-response devices started in the early part of the window, or limit the maximum output power of energy storage devices during the compensation phase. The system constructs a proxy optimization model locally at each node, aiming to reduce node power deviation while considering equipment adjustment costs. It solves the model by incorporating the equipment power adjustment range, resulting in a local coordinated adjustment strategy for each node. The inputs to the proxy optimization model mainly include the target power compensation amount from the synchronous compensation data, equipment response time parameters, equipment power adjustment range, the current node's actual power, and the system-allocated time window information. The intermediate structure of the model can be a constrained optimization process, searching or adjusting the node power change trajectory based on equipment adjustment boundary conditions and time window constraints to ensure it meets system compensation requirements without violating equipment operating limitations. The model output is the target power adjustment value for each node within each control time slice, along with the corresponding start and pullback times, thus forming the node's local coordinated adjustment strategy. The system performs consistency correction on the adjustment results / local coordinated adjustment strategies of each node, correcting the adjustment amounts of nodes with deviations to ensure the overall power adjustment meets the system power balance requirements, thereby generating coordinated control data.

[0091] Optionally, the flexible load slice includes:

[0092] S11. Extract the time envelope from the microgrid operation data to obtain the time envelope data;

[0093] In one embodiment, the system acquires the active power sequence, reactive power sequence, and corresponding timestamps of each load node within a continuous scheduling cycle, and performs a sliding window scan on the power sequence according to a preset window length. For each sliding window, the system calculates the maximum and minimum power values ​​within the window and records them as the upper and lower envelope points, respectively, thus obtaining the corresponding upper and lower envelope point sets. After obtaining the envelope point sets, the system connects adjacent envelope points in chronological order, using piecewise linear connections or smooth interpolation to generate continuous curves, forming the upper and lower envelope curves. The system uniformly records the obtained upper and lower envelope curves, the corresponding window center time, and the load node identifiers to construct time envelope data.

[0094] S12. Discretize and slice the time envelope data to obtain load slice data;

[0095] In one embodiment, the system divides the time envelope data into equal-length segments according to a preset slice duration, for example, dividing a continuous running curve into multiple time segments lasting 30 seconds or 60 seconds. For each time slice, the system calculates the average value of the upper envelope curve and the average value of the lower envelope curve within that segment, and calculates the difference between the upper and lower envelopes. The system extracts the power change slope based on the changing trend of the envelope curve within that time slice and calculates the dispersion of power fluctuations. When the system detects that a time slice crosses the load start-up and stop times, it re-divides the time slice using the start-up and stop times as the boundaries, ensuring that each time slice contains load data for only a single operating state. The load start-up and stop times refer to the time points when load equipment transitions from a stopped state to an operating state, or from an operating state to a stopped state. In microgrids or park energy management systems, load equipment typically has clearly defined operating state information, such as start-up, running, standby, or stopped states. When equipment transitions from a stopped or standby state to a normal operating state, the corresponding time point can be defined as the load start-up time; when equipment transitions from an operating state to a stopped or disconnected state, it corresponds to the load stop time. This information can usually be identified through equipment operating state data, power change records, or control command execution records. For example, when equipment power rapidly rises from near zero to a stable operating range, it can be determined as the start-up time; when power drops from a stable operating level to near zero, it can be determined as the stop time. The system records the corresponding start time, end time, characteristic description information, and the identifier of the associated load node for each time slice, thereby generating load slice data.

[0096] S13. Perform similarity measurement on the load slice data to obtain load similarity data;

[0097] In one embodiment, the system performs similarity analysis on the features of each load slice in the load slice data. Each load slice corresponds to a set of feature description information, which includes at least average power level, upper and lower envelope difference, power change trend, power fluctuation degree, and duration. The system compares the differences between two load slices in terms of average power level, upper and lower envelope difference, power change trend, power fluctuation degree, and duration, and determines whether the differences between each feature are within a preset allowable range. When the differences of most features are within the allowable range, the system determines that the overall difference between the two slices is small; when multiple key features deviate significantly at the same time, the system determines that the overall difference is large. The system can summarize the differences of multiple features by counting the number of features that meet the similarity conditions or by comparing the difference levels of key features, thereby obtaining the overall difference evaluation result. For example, the key feature difference level refers to the result of the system classifying the difference degree of each feature dimension, which is used to represent the strength of the influence of the feature on the difference in load behavior. The system sets difference judgment intervals for each feature, for example, based on historical data or experience thresholds, dividing feature differences into multiple levels such as "low difference", "medium difference", and "high difference". When the difference between two load slices on a certain feature is small and within the allowable range, the system marks them as low difference level; when the difference is close to the upper limit of the threshold or shows a significant change, it is marked as medium difference level; when the difference significantly exceeds the threshold or shows different operating modes, it is marked as high difference level. The system can convert the difference level of each feature into a corresponding numerical score, assigning a small value to low difference, a medium value to medium difference, and a large value to high difference. The system summarizes the difference scores of all features to obtain a comprehensive difference score. After completing the difference degree calculation, the system compares the result with the preset similarity judgment threshold. When the difference degree between two slices is lower than the threshold, they are judged to belong to similar load behavior patterns; when the difference degree is higher than the threshold, they are judged to be different load patterns. The system records the corresponding slice number, similarity evaluation result, main difference feature identifier (when comparing each feature item by item, the system will identify the feature dimension with the largest difference degree or the feature dimension exceeding the preset difference threshold, and record these features as main difference features), and similarity relationship label, thereby generating load similarity data.

[0098] S14. Perform slice aggregation on the load similarity data to obtain slice aggregated data;

[0099] In one embodiment, the system constructs a slice association graph using load slices as nodes and similarity relationships between slices as connections, and aggregates nodes according to a preset similarity threshold. When the similarity between multiple slices at adjacent time points is continuously higher than the threshold, the system prioritizes merging them into the same cluster to form a set of load segments with continuous operating characteristics. For slices with high similarity but time intervals, the system retains the association weight between them and the cluster. After completing the initial aggregation, the system checks the size and stability of each cluster. For isolated clusters containing only a few slices and with low feature stability, the system re-merges them into the closest main cluster according to feature similarity. Each cluster corresponds to a type of load pattern with similar power fluctuation patterns and regulation characteristics, thus forming slice aggregated data.

[0100] S15. Perform flexible spatial reconstruction based on the sliced ​​aggregated data to obtain flexible load data.

[0101] In one embodiment, the system maps each aggregation cluster in the sliced ​​aggregation data to a corresponding flexible node, where each flexible node represents a set of loads with similar adjustment characteristics. The system analyzes the spatial relationships between flexible nodes by combining the electrical connection relationships and control structure information of the load equipment. These spatial relationships include at least the feeder, access bus, geographical region, and control channel connection. If two flexible nodes are located on the same feeder or share the same control branch, and their adjustable time intervals overlap, the system increases the spatial correlation between them and establishes a corresponding spatial connection relationship. After completing the spatial correlation construction, the system integrates parameters such as the adjustable power range, minimum adjustable duration, response speed, and adjustment priority of each node based on the operating characteristics of the loads contained in each flexible node, forming a spatial structure unit. The system outputs flexible load data, which simultaneously includes load aggregation results, spatial distribution relationships, and corresponding adjustment capability parameters.

[0102] Optionally, flexible spatial reconstruction includes:

[0103] Based on the sliced ​​aggregated data, flexibility node mapping is performed to obtain load node data;

[0104] In one embodiment, the system reads the cluster number, the set of slices within the cluster, the corresponding device number, and the power characteristic information of each slice from the slice aggregation data, and constructs flexible nodes using the cluster as the basic mapping unit. For slice data from multiple devices within the same cluster, the system performs secondary partitioning based on the electrical connection relationship of the devices. The partitioning criteria include the feeder to which the device belongs, the node access location, and the control interface affiliation. When multiple devices are located on the same feeder and share the same control interface, the system maps them to the same flexible node; when devices belong to different feeders or have inconsistent control interfaces, they are split into multiple sub-nodes. After completing the node mapping, the system performs statistical analysis on the slice data contained in each node, extracting parameters such as the node's average power level, the maximum achievable adjustable power range, the minimum holding power, and the response time, and records the spatial identification information of the node in conjunction with the device access location, thereby forming load node data, so that each node has a description of its spatial location and adjustable capability.

[0105] Node associations are constructed based on load node data to obtain load association data;

[0106] In one embodiment, the system uses each load node in the load node data as the analysis object to construct the correlation between nodes. The system identifies the electrical adjacency between nodes based on their electrical access information, such as whether they are connected to the same feeder, the same bus branch, or share the same control link, and assigns basic correlation weights accordingly. The system combines historical operating data to compare and analyze the power change trends between nodes, calculates the consistency of power changes between two nodes within the same time period and the time overlap ratio of load slices, and adjusts the basic correlation weights. When the power change trends between two nodes have high consistency and their operating time overlap ratio exceeds a preset threshold, the system establishes a correlation between the two nodes and records the corresponding correlation strength, correlation type, and main influencing period. Load correlation data is formed through the above processing.

[0107] Flexible spatial structures are generated based on load correlation data to obtain flexible structure data;

[0108] In one embodiment, the system constructs a load node association graph based on load association data and uses the association strength between nodes as the weight of the connection relationship to divide the node set into connected structures. When the association strength between two nodes is higher than a preset threshold, the system prioritizes classifying them into the same flexible spatial unit to form a node set. For boundary nodes that have high association with multiple flexible spatial units simultaneously, the system determines the node's assignment based on the dominant feeder it belongs to and the magnitude of the association strength, and assigns the node to the spatial unit with the highest association degree. After completing the spatial unit division, the system performs statistical analysis on the structural characteristics of each flexible spatial unit, including the number of nodes within the unit, total power scale, average response speed, and the degree of association between nodes. For structural units with a small number of nodes or a low degree of internal association, the system merges or removes them based on their association with adjacent units. The system obtains flexible structure data.

[0109] The flexible load data is obtained by integrating the adjustment capabilities based on the flexible structure data.

[0110] In one embodiment, the system integrates the adjustment capabilities of each flexible spatial unit in the flexible structure data. The system statistically summarizes the adjustment parameters of each load node within the unit, including the adjustable power range, adjustable power range, response time, sustainable adjustment duration, and recovery time, thereby obtaining the overall adjustment capability range of the spatial unit. When nodes with different response speeds exist simultaneously within the same flexible spatial unit, the system uses a hierarchical integration approach, recording the capabilities of nodes with faster response speeds separately from those with slower response speeds to avoid distortion of the adjustment capability description caused by direct merging. The system combines the adjustment priority of each node, the load interruptibility level (configured by maintenance personnel when the equipment is connected to the system, or automatically assigned based on historical operating data and business attributes), and equipment operating constraints (from pre-stored local equipment technical specifications, preset equipment monitoring systems, or pre-stored local equipment operating databases) to represent the adjustment capability of spatial units, forming a set of adjustment parameters including adjustment capacity range (the range of adjustable power up and down), response time, continuous support capability (the maximum duration the system can maintain the adjustment state under rated operating conditions), and call priority order (arranging each load node or flexible unit according to adjustment priority from high to low). The system generates flexible load data. The adjustment priority can be derived from pre-established load management rules or equipment registration information. In actual energy management systems or demand response systems, different types of loads have different degrees of impact on production or services. Therefore, adjustment priorities are usually configured for equipment when it is connected to the system. For example, higher adjustment priorities can be set for equipment with less impact on production or with greater adjustment flexibility; lower adjustment priorities are set for critical production equipment or critical infrastructure. This priority information can be stored in the device registry, load management policy library, or user-side configuration parameters, and can be read by the system during the adjustment capability integration process for sorting or invocation control.

[0111] Optionally, S2 includes:

[0112] S21. Identify communication links in the flexible load data to obtain communication link data;

[0113] In one embodiment, the system reads information such as load node number, control terminal number, access switch identifier, communication interface type, and upper controller address from the flexible load data, and identifies the communication access method of each load node in conjunction with device registration information and network topology table. The system determines the link category of a node based on the communication interface type: when a node accesses via industrial Ethernet, it is marked as a wired link; when a node accesses via 4G, 5G, or WiFi communication modules, it is marked as a wireless link; when a node accesses via RS485, CAN bus, or power line carrier, it is marked as a fieldbus link. After link type identification, the system determines the communication path of each node based on network topology information and records the uplink channel, downlink channel, and number of relay nodes. When a load node is configured with a primary link and a backup link, the system establishes a primary link identifier and a backup link identifier respectively, and records the link switching conditions. The system generates communication link data, including node number, link type, communication path structure, link level, and primary / backup link status.

[0114] S22. Extract communication delay features based on communication link data to obtain delay feature data;

[0115] In one embodiment, the system continuously sends multiple sets of standard control messages to each flexible load node within a preset observation period, and records the message sending time, node reception confirmation time, and device execution feedback time. Based on this time information, the system statistically analyzes the time elapsed for a single communication transmission, the round-trip time of the control message, and the time required for device execution response, thereby forming basic communication latency information. The system uses communication link type as the classification unit and performs statistical analysis on the communication records of each node to obtain characteristics such as average latency, maximum latency, minimum latency, and latency fluctuation range, and evaluates the latency stability and jitter level of the link accordingly. For links experiencing communication congestion within the observation period (the system identifies congestion by comparing the actual communication latency within the observation period with the normal average latency level of the link. When the communication latency is significantly higher than the normal average level of the link and fluctuates significantly in multiple consecutive sampling periods, the system determines that there is communication congestion during that time period), the system selects the latency value at a higher proportion as a representative latency indicator. When a node is configured with a primary link and a backup link, the system extracts the latency characteristics of the two links respectively and records the additional latency generated during link switching. The system organizes information such as node number, link type, average latency, latency fluctuation level, jitter level, and link switching latency (the additional communication time overhead incurred when the system switches to the backup link when the primary link fails or its quality degrades, i.e., the additional latency generated during link switching) to generate latency feature data.

[0116] S23. Based on the communication delay characteristic data, perform communication timing propagation to obtain timing propagation data;

[0117] In one embodiment, the system uses the time when the control center issues the scheduling command as the propagation starting point and calculates the propagation process of the control command in the communication network based on communication delay characteristic data. The system combines the average communication delay and fluctuation range of each node's link to predict the arrival time of the control command from the control center to each flexible load node. When a node is connected via a single-hop link, the system adds the command sending time to the transmission delay of that link to obtain the estimated arrival time; when a node is connected via a multi-hop link, the system sequentially accumulates the transmission delays of each link according to the network topology, taking into account the processing time of relay devices, to obtain the complete propagation time. After obtaining the estimated arrival time, the system combines the device's local control cycle to match and calculate the command arrival time with the device's execution trigger time, thereby determining the propagation start time, estimated arrival time, execution alignment time, and propagation end time of each node. When a node is configured with a primary link and a backup link, the system generates propagation path records corresponding to the two links respectively. The system forms time-series propagation data.

[0118] S24. Perform time-series folding and reconstruction on the time-series propagation data to obtain heterogeneous communication data.

[0119] In one embodiment, the system selects the earliest control command transmission time within the same scheduling cycle as a unified time reference and converts the propagation time of each node into a time offset relative to this reference time. The system segments these time offsets according to a preset time window, for example, dividing a scheduling cycle into multiple consecutive propagation stages, and grouping the propagation events of nodes whose time offsets fall within the same stage window into the same folding unit. After completing the stage division, the system performs statistical analysis on the propagation events in each folding unit, recording the number of nodes in that stage, the link type composition (the combination of communication link categories participating in the propagation of control commands within the same propagation stage, where the link types may include industrial Ethernet links, wireless communication links (such as 4G, 5G, or WiFi), and fieldbus links (such as RS485, CAN bus, or power line carrier links), etc.), the average propagation arrival time, and the dispersion of the propagation time. The system also statistically analyzes the offset of the node execution trigger time relative to the arrival time. For nodes with similar propagation times but using different communication links, the system retains the link category identifier to represent the heterogeneous characteristics of the communication structure; for nodes with significantly delayed propagation times, they are separately marked as delayed propagation branches. The system generates heterogeneous communication data, including information such as propagation stage number, stage node set, link type distribution, average propagation time, and propagation fluctuation characteristics.

[0120] Optionally, time-series folding reconstruction includes:

[0121] The propagation time reference is extracted from the time-series propagation data to obtain the time reference data.

[0122] In one embodiment, the system reads information such as the control command transmission time, the reception time of each node, the execution trigger time, and the propagation end time from the timing propagation data, and uses the time when the control center issues the first control command within the same scheduling cycle as the global propagation time reference. If multiple batches of control commands exist within a scheduling cycle, the system uses the transmission time of the main scheduling command corresponding to the power regulation task as the main reference, and uniformly converts the times of the remaining control commands into time offsets relative to this main reference. After the time reference is determined, the system aligns the propagation records of each node. When a node experiences link retransmission or primary / backup link switching during communication, the system prioritizes selecting the time of the first valid reception of the control command for time alignment. The system organizes and encapsulates information such as node number, unified reference time, control command transmission time offset, and node reception time offset to generate time reference data.

[0123] Based on the time reference data, a propagation time normalization mapping is performed to obtain time-normalized data;

[0124] In one embodiment, the system uses the reference time in the time base data as a unified time zero point, and converts the reception time, execution trigger time, and propagation end time recorded by each node into propagation times relative to the reference time. The system performs a unified scale conversion on the above relative propagation times based on the total duration of the current scheduling cycle, enabling the propagation times of each node to be expressed within the same time range. During the normalization process, the system compares the relative propagation time with the scheduling cycle length and converts it into a normalized time value within a unified proportion range. When the propagation time of a node exceeds the current scheduling cycle range, the system marks that node as a cross-cycle propagation node. For cases where there are significant differences in propagation times between different communication links, the system uses a segmented mapping method to adjust the propagation time, ensuring that both faster propagation links and links with long-tail delays can be effectively represented within a unified time scale. The system generates time-normalized data, including node number, normalized reception time, normalized execution trigger time, and cross-cycle propagation identifier.

[0125] The propagation trajectory is folded based on the time-normalized data to obtain folded trajectory data.

[0126] In one embodiment, the system segments the normalized propagation timeline, dividing the entire time range into multiple consecutive folded windows with a fixed width, for example, dividing it into several stage intervals according to a uniform time interval. The system determines the time window to which the corresponding propagation event belongs based on the normalized reception time and execution trigger time of each node, and groups propagation events falling within the same window into the same propagation folding unit. If the normalized reception time or execution trigger time of a propagation event falls within the time range of a certain window, the propagation event is assigned to the corresponding window; when the reception time and execution trigger time fall into two adjacent windows, the system prioritizes classification based on the window to which the execution trigger time belongs. The system statistically analyzes the propagation situation within each folding unit, including the number of nodes, average normalized reception time, average execution trigger time, and the dispersion of propagation time within the window. When the same node simultaneously forms propagation records through the main link and backup link, the system prioritizes retaining the propagation path that actually generates the control trigger and marks the other path as a redundant propagation branch. For propagation events with similar time positions but crossing the boundaries of adjacent windows, if their time difference is less than a preset threshold, the system assigns them to the central window according to the principle of minimum deviation. The system generates folded trajectory data to represent the propagation and aggregation of each node in a unified time structure.

[0127] Heterogeneous propagation structure is reconstructed based on folded trajectory data to obtain heterogeneous communication data.

[0128] In one embodiment, the system reconstructs the communication propagation structure in stages based on folded trajectory data. The system uses the time interval corresponding to each folded window as the propagation stage node, and the set of devices participating in propagation within that window, along with the corresponding link type combination, as stage attribute information. The system analyzes the propagation succession relationship between adjacent folded windows in chronological order. When a node in a later stage is triggered by control propagation in the previous stage, a directed connection is established between the two stage nodes, and the average propagation time increment and link changes between stages are recorded. Within each propagation stage, the system statistically analyzes the distribution of different communication link types. When wired links, wireless links, and fieldbus links coexist in the same stage, the system statistically analyzes the proportion of each link type, the average propagation time offset, and the propagation time fluctuation, respectively, to form the heterogeneous communication characteristics within the stage. The system generates heterogeneous communication data, including the propagation stage number, stage node set, link type distribution, stage average propagation time, propagation time fluctuation characteristics, and connection relationships between stages.

[0129] Optionally, S3 includes:

[0130] S31. Based on heterogeneous communication data, identify communication timing mismatch to obtain timing mismatch data;

[0131] In one embodiment, the system reads communication propagation characteristics from heterogeneous communication data, such as propagation stage number, stage node set, link type distribution, and average propagation time of each stage. Node reception time and execution trigger time are obtained by combining propagation stage time information with device operation logs. The system can calculate the expected reception time of control commands based on the average propagation time of each stage and the stage to which the corresponding node belongs, and extract the actual execution trigger time of the node from the device operation record or control log. The system uses the execution trigger time as the basis for determining timing mismatch, compares it with the target synchronization time to calculate the offset, and uses the target synchronization time set within the same scheduling cycle as the reference time. The system compares the actual execution trigger time of each node with this reference time to calculate its time offset degree. When the time offset of a node exceeds a preset threshold, the system marks the node as a timing mismatch node and records its offset direction and offset magnitude. For multiple nodes within the same propagation stage, the system statistically analyzes the trigger time differences between nodes, including the maximum time offset difference and the average time offset difference. When the system detects that the trigger time deviation between wireless link nodes and wired link nodes continues to increase, it adds a link mismatch identifier to the relevant nodes. The system generates timing mismatch data, which includes information such as node number, time offset magnitude, offset direction, propagation stage, and link mismatch type.

[0132] S32. Based on the time-series mismatch data, the response window interval is calibrated to obtain the window interval data;

[0133] In one embodiment, the system identifies window intervals in the control response process based on timing mismatch data. The system reads the actual execution trigger time of each mismatched node and compares it with the corresponding target response time to determine the time range in which the control command has been issued but the load has not yet produced an effective regulation output. The system uses the target response time as the starting point of the interval and the time when the actual power of the corresponding node begins to deviate continuously from the original operating state as the ending point of the interval, thereby forming a candidate window interval for a single node. When the candidate window intervals of multiple nodes overlap on the time axis, the system merges these overlapping parts and statistically analyzes the difference between the system's target regulation power and the actual regulation power within this time range. If the power difference continues to be higher than a preset threshold in multiple consecutive sampling periods, the system confirms that the time period is an effective response window interval. The system generates window interval data, including the interval start time, end time, duration, average power gap, and associated node set (when the candidate window intervals of multiple nodes overlap or are consecutive on the time axis, the system records the node numbers of all nodes involved in the mismatch during this time period while merging the intervals, and forms a corresponding associated node set).

[0134] S33. Perform virtual inertia dynamic mapping based on the empty window interval data to obtain inertia compensation data;

[0135] In one embodiment, the system dynamically maps the inertia parameters of a virtual synchronous machine based on the data from the empty window intervals. The virtual synchronous machine refers to a control module in a power electronic interface device (such as an energy storage converter or a distributed power inverter) that simulates the inertia and damping characteristics of a traditional synchronous generator through a control algorithm. This module sets equivalent inertia parameters in the control loop, enabling the power electronic device to exhibit a dynamic response capability similar to the rotational inertia of a synchronous machine when power changes. The virtual synchronous machine module is pre-deployed in the energy storage converter controller. Its basic structure includes an input interface, an inertia calculation unit, and a parameter output interface. The input interface receives empty window interval data, real-time power deviation, and current operating status parameters provided by the system scheduling layer. The inertia calculation unit calculates the dynamic inertia value based on the above input information. The output interface outputs updated inertia control parameters to the power control module to adjust the power response process of the energy storage device. The system reads the duration, average power deficit, and power deficit change trend of each empty window interval and determines the adjustment demand characteristics of that empty window interval accordingly. When the power gap is short but the power deficit increases rapidly, the system increases the inertia adjustment range to enhance the system's transient power support capability. When the power gap is long and the power deficit changes relatively smoothly, a smaller inertia adjustment range is used. In specific implementation, the system can classify the power gap into three types based on its duration: transient, transitional, and delayed, and match corresponding inertia levels to different types. When the power gap is short and the power deficit increases rapidly, the system defines it as a transient power gap, indicating that the system experiences a sudden power deficit in a short period of time, requiring high inertia for rapid support. When the power gap is of medium duration and the power deficit changes relatively smoothly, the system defines it as a transitional power gap, indicating that the system is in a load adjustment or resource handover phase. When the power gap is long and the power deficit changes slowly, it is defined as a delayed power gap, corresponding to situations where slow-response loads have not yet arrived or the system has a long-term power shortage. The system refines the inertia level based on the power deficit size to obtain inertia compensation parameters. The system determines the base inertia level according to the type of window interval and adjusts it based on the difference between the average power deficit and the current system's adjustment capability. When the power deficit is large, the inertia adjustment range is appropriately increased based on the base inertia level; when the power deficit is small, the inertia level is maintained or slightly adjusted. The system generates inertia compensation data, including information such as the inertia adjustment range, the effective start time, the effective duration, and the subsequent gradual recovery rate.

[0136] S34. Perform virtual damping cooperative shaping based on inertia compensation data to obtain damping compensation data;

[0137] In one embodiment, the system performs coordinated shaping of the damping parameters of the virtual synchronizer based on inertia compensation data. The system analyzes the time range of the inertia compensation data and divides it into an inertia boosting phase, an inertia maintenance phase, and an inertia decline phase. For example, the system reads the inertia compensation values ​​corresponding to each control cycle in chronological order and compares the inertia changes between adjacent control cycles. When the inertia value continuously increases over multiple consecutive cycles and the increase exceeds a preset change threshold, this time period is classified as the inertia boosting phase. When the inertia value remains stable within a certain range and the change is below the stability threshold, it is determined to be the inertia maintenance phase. When the inertia value gradually decreases in subsequent cycles and shows a continuous decline trend, it is classified as the inertia decline phase. Based on the principle of coordinated matching between inertia and damping, the system assigns corresponding damping baseline values ​​to each stage, ensuring a higher damping level for the inertia boosting phase and a gradually decreasing damping level for the inertia decline phase. After the damping baseline values ​​are determined, the system performs risk correction on the damping parameters based on the expected adjustment time of the slow-response load. When the system predicts that multiple loads may simultaneously replenish within a few future sampling periods (i.e., the system extrapolates the potential replenishment power within a few future sampling periods based on the expected arrival time of each load node in the response offset data and the number of non-arriving nodes in the response status data), the damping level is increased in advance to limit the rate of power change on the energy storage side. When the replenishment behavior is relatively dispersed, a relatively gradual damping change is maintained. The system performs rate-of-change constraint and smoothing processing on the damping parameter sequence to form damping compensation data.

[0138] S35. Based on the damping compensation data and local power deviation data, perform synchronous phase angle traction control to obtain phase angle compensation data;

[0139] In one embodiment, the system collects power deviation information of each local node in real time and, combined with the target damping change trajectory in the damping compensation data, calculates the phase angle adjustment of the virtual synchronizer within the current control cycle. When the system detects that the local power deviation continues to increase and the damping is in the increasing phase, the system appropriately reduces the phase angle advance speed; when the power deviation gradually decreases and the damping enters the stable phase, the system gradually relaxes the phase angle constraint, allowing the system synchronization rhythm to gradually return to normal. In the specific implementation process, the system generates a phase angle traction sequence according to a continuous control cycle and sets limits on the phase angle change amplitude between adjacent control cycles. The system generates phase angle compensation data, including the phase angle correction amount, the phase angle change rate (phase angle change amount between adjacent control cycles / control cycle time), the effective time segment (the control cycle range corresponding to the phase angle compensation), and the regression threshold for restoring normal operation (the power deviation threshold that allows the system to exit phase angle traction control), etc.

[0140] S36. Combine the inertia compensation data, damping compensation data, and phase angle compensation data to obtain synchronous compensation data.

[0141] In one embodiment, the system performs time alignment processing on inertia compensation data, damping compensation data, and phase angle compensation data under a unified time axis, and extracts the corresponding inertia adjustment value, damping adjustment value, and phase angle correction amount within the same control cycle. The system performs consistency checks on the three types of control parameters. When an adjustment conflict is detected between different parameters, such as when the inertia adjustment has entered the decline phase while the phase angle still maintains strong traction, the system performs coordinated correction based on whether the current window interval has ended and the power gap change. After completing the coordination processing, the system unifies the inertia parameter, damping parameter, and phase angle correction parameter to generate synchronous compensation data. This data includes at least the parameter number, target inertia value, target damping value, target phase angle correction amount, parameter effective time, and parameter exit conditions, thus forming synchronous compensation data.

[0142] Optionally, the response window interval calibration includes:

[0143] Response time offset is analyzed based on timing mismatch data to obtain response offset data;

[0144] In one embodiment, the system reads the node number, target synchronization time, actual trigger time, and corresponding mismatch amplitude and direction information from the timing mismatch data, and analyzes the response time offset of each node using the target synchronization time as a unified reference benchmark. The system determines the degree of response offset of a node based on the difference between the actual trigger time and the target synchronization time, thereby obtaining the response time offset information of each node. When the same node has multiple trigger records within a scheduling cycle, the system preferentially selects the trigger time of the first valid control response as the actual response time of that node; if no valid response is detected within multiple consecutive sampling cycles, the node is marked as a non-responding node and recorded according to a preset maximum offset range. The system classifies the node response type based on the magnitude of the response time offset: when the offset amplitude is within a preset range, it is recorded as a quasi-synchronous response; when the actual trigger time is significantly later than the target synchronization time, it is recorded as a delayed response; when the actual trigger time is significantly earlier than the target synchronization time, it is recorded as an early response. The system obtains response offset data based on the above.

[0145] The adjustment status is determined based on the response offset data to obtain the response status data.

[0146] In one embodiment, the system determines the adjustment status of a node based on response offset data and the target adjustment power and actual output power of each node. After a node triggers a response (the load node begins to execute adjustment actions and generates observable power changes after receiving a scheduling control command), the system monitors the actual power changes over several consecutive sampling periods. When the direction of the actual power change is consistent with the target adjustment direction, and the actual adjustment amount reaches a preset proportion of the target adjustment amount, the system determines that the node is in the "in-place" state. If the node has generated a trigger response, but the actual power change is lower than the preset proportion of the target adjustment amount, it is determined to be in the "partially in-place" state; if the node does not detect a valid trigger or the power change is not significant, it is determined to be in the "not in-place" state. For nodes with large power fluctuations and frequent changes in direction, the system adds a stability check, confirming the in-place status only when the power change direction remains consistent over multiple consecutive sampling periods. The system generates response status data, including node number, adjustment status, in-place proportion, and status confirmation time.

[0147] Power gap detection is performed based on the response status data to obtain gap identification data;

[0148] In one embodiment, the system compares the target adjustment power of the current scheduling task with the actual achieved adjustment power on a unified time axis to identify power gaps in the system. The system statistically analyzes the actual contribution of each node based on response status data: for nodes in the achieved state, their actual adjustment power is included in the total system adjustment based on real-time measurements; for partially achieved state nodes, it is calculated based on their achieved adjustment ratio; for nodes not in the achieved state, their corresponding target adjustment power is entirely included in the power gap. The real-time power gap value is obtained through this method. The system performs sliding statistics on the power gap within continuous sampling periods. When the gap value is consistently higher than a preset gap threshold for multiple consecutive sampling periods, and the duration exceeds the minimum duration threshold, the system determines that time period as an effective power gap interval. When the gap decreases locally for a short time but does not fall below the release threshold, it is still considered part of the same gap process. The system generates gap identification data, including gap start time, gap peak value, average gap power, and associated node set information.

[0149] By aggregating continuous time intervals based on the gap identification data, the gap interval data is obtained.

[0150] In one embodiment, the system performs continuity analysis on the effective power gap segments identified on the time axis and performs interval aggregation processing on adjacent gap segments. When the time interval between two gap segments is less than a preset gap threshold and their associated node sets have a high degree of overlap, the system determines them as the same response window process and merges them on the time axis; when the time interval between gap segments is long or the associated nodes differ greatly, they remain independent window intervals. After completing the interval aggregation, the system performs feature statistics on each window interval, including the interval start time, end time, duration, average power gap, and peak power gap, and identifies the dominant responsible node that contributes the most to the interval. The system categorizes window intervals based on their duration and power deficit intensity. First, the duration is divided into short-term and continuous intervals; for example, intervals shorter than a preset threshold are labeled as short-term gaps, while those longer are labeled as continuous gaps. Power deficit intensity is categorized into mild and severe levels; average or peak power deficits below a preset threshold are labeled as mild, and those above are labeled as severe. Based on this, the system further classifies window intervals, such as into short-term mild gaps, short-term severe gaps, and continuous gaps. The system then generates window interval data.

[0151] Optionally, virtual damping co-shaping includes:

[0152] Based on the inertia compensation data, the inertia action range is analyzed to obtain the inertia range data;

[0153] In one embodiment, the system reads information such as the inertia adjustment magnitude, effective start time, effective duration, and inertia recovery rate from the inertia compensation data, and analyzes its effective range based on the changing trend of the inertia parameters in the continuous control cycle. The system divides the inertia change process into three stages: when the inertia value continuously increases and exceeds a preset change threshold within the continuous control cycle, it is determined to be the inertia boosting stage; when the inertia value remains stable within a certain range, it is determined to be the inertia maintenance stage; and when the inertia value gradually decreases and enters the recovery process, it is determined to be the inertia fallback stage. When the system detects that multiple gaps overlap or occur consecutively in time, it divides the inertia action process into multiple inertia sub-intervals according to time sequence, and records the start time, end time, corresponding dominant gap identifier, and related gap level information for each sub-interval. The system forms inertia interval data.

[0154] Dynamic damping base value matching is performed based on inertia interval data to obtain damping base value data;

[0155] In one embodiment, the system dynamically matches the damping base value of the virtual synchronizer based on inertia interval data. The system reads the inertia level and interval type of each inertia interval, and assigns corresponding damping base values ​​to different intervals according to a pre-established correspondence between inertia level and damping base value. When the inertia is in a higher level interval, the system matches a higher damping base value to suppress power overshoot generated under high inertia operating conditions; when the inertia is in a medium level interval, a medium damping level is matched; when the inertia enters the fallback phase, the damping base value is gradually reduced to avoid lag in system control response. When multiple inertia intervals are superimposed within the same time range, the system evaluates the damping base value based on the dominant inertia level and corresponding power gap degree of each interval, and determines the damping benchmark level. The system identifies the sub-interval with the highest inertia level and the largest power gap influence range, and takes it as the dominant inertia interval. The system refers to the damping base value corresponding to the dominant interval as the basic damping level, and at the same time, it makes appropriate corrections to the basic damping value based on the power gap degree of other sub-intervals. When the inertia levels of multiple sub-intervals differ slightly, the system can weight and adjust the damping value based on the relative size of the power gap in each interval (normalizing / activating the relative size to a value between 0 and 1 as a weight) to obtain a baseline damping value. The system generates damping baseline data, which includes information such as the interval number, the corresponding baseline damping value, and the effective time range of the damping parameters.

[0156] By backfilling and superimposing risk corrections on the damping baseline data, damping correction data is obtained.

[0157] In one embodiment, the system proactively corrects the damping baseline data based on the expected adjustment time of slow-response loads and the local power recovery trend. The system analyzes the expected arrival time of each slow-response node and predicts the potential recovery power scale within several sampling periods. When the system determines that multiple slow-response nodes will enter the adjustment state at similar times, and their expected recovery power, combined with the current energy storage compensation power, exceeds the current target power gap, it determines that there is a risk of recovery superposition and increases the damping correction magnitude in the corresponding time interval. When the predicted recovery behavior is relatively dispersed or the recovery power scale is small, the system maintains the original damping baseline value or makes only minor adjustments. The system classifies risk levels based on the difference between the sum of the expected replenishment power and the current compensation power and the target power gap. It then applies corresponding damping correction factors to different risk levels. For example, if the system predicts that only a small number of slow-response nodes will gradually arrive within several sampling periods, and the sum of the expected replenishment power and the current energy storage compensation power is still significantly lower than the target power gap, the system classifies this situation as low-risk and only slightly increases the original damping base value. If the system predicts that multiple slow-response nodes will arrive in adjacent sampling periods, and the sum of the expected replenishment power and the current energy storage compensation power is close to the target power gap, it is classified as medium-risk. In this case, the system appropriately increases the damping correction magnitude to reduce the output change rate of the energy storage units, allowing the slow-response load to gradually take over the regulation task. If the system predicts that a large number of slow-response nodes will arrive in a short period, and the sum of the expected replenishment power and the current compensation power significantly exceeds the target power gap, it is classified as high-risk. In this time period, the system significantly increases the damping correction value and limits the output slope of the energy storage units in advance to avoid power overshoot caused by multiple devices replenishing simultaneously. The system generates damping correction data, which includes the corrected damping value, the corresponding risk level, and the correction reason identifier.

[0158] The damping correction data is smoothed and shaped to obtain the damping trajectory data;

[0159] In one embodiment, the system performs smoothing and shaping on the damping sequence formed by the damping correction data. The system generates the original damping sequence in chronological order and sets a maximum rate of change limit for the damping variation amplitude between adjacent control cycles. When the damping difference between two adjacent cycles exceeds a preset threshold, the system truncates the excess portion and distributes the remaining variation to subsequent control cycles for gradual release. During the transition from the inertia increase phase to the maintenance phase, and from the maintenance phase to the fallback phase, the system uses a segmented smoothing transition method to shape the damping sequence, ensuring continuous damping variation at the boundaries of these intervals. The above processing yields damping trajectory data, including the target damping value for each control cycle, the damping variation trend, and trajectory smoothing status indicators.

[0160] The damping trajectory data is encapsulated into parameters to obtain damping compensation data.

[0161] In one embodiment, the system encapsulates damping trajectory data according to a unified control timeline. The system reads the target damping value and its trajectory for each control cycle and associates them with the corresponding inertia interval number, window interval number, and backfill risk identifier to form a structured set of damping compensation parameters. The system organizes the damping parameters according to the control cycle, ensuring that each time period corresponds to specific damping adjustment information. During encapsulation, the system generates a corresponding damping compensation data packet for each control cycle, including the control cycle number, target damping value, start time, duration, upper limit of damping change rate, and parameter release conditions. When damping compensation is in a high-risk backfill stage, the system adds a priority identifier to the data packet.

[0162] Optionally, S4 includes:

[0163] S41. Based on the synchronous compensation data, compensation responsibility is stratified to obtain stratified responsibility data;

[0164] In one embodiment, the system reads the inertia compensation value, damping compensation value, phase angle compensation value, effective time interval, and corresponding window interval number from the synchronization compensation data. Combining this with the response speed, adjustable power range, and continuous support capability of each regulating device, the system stratifies the responsibilities of the resources involved in regulation. The system identifies energy storage devices with fast response speeds and bidirectional power regulation capabilities, classifying them as the transient support layer to quickly fill power gaps in the initial control phase. Flexible loads with moderate response speeds that can quickly take over power regulation tasks are classified as the transitional takeover layer. Devices with slower response speeds but longer sustainable regulation times, such as charging pile groups or delayable loads, are classified as the steady-state regression layer. After stratification, the system records the corresponding responsibility time interval, target compensation share, and responsibility switching trigger conditions for each layer of regulating resources, thus forming stratified responsibility data.

[0165] S42. Construct a time-series relay window based on the hierarchical responsibility data to obtain the relay window data;

[0166] In one embodiment, the system constructs inter-layer adjustment relay intervals on a unified scheduling time axis based on the effective time and expected exit time of each responsibility layer in the hierarchical responsibility data. When the end time of the previous responsibility layer is later than the start time of the next responsibility layer, the system determines the overlapping portion of their times as a relay window; when there is a short time interval between them, the system adds a preset buffer time on both sides of this time interval to construct a transition relay window. After the transition relay window is determined, the system divides the adjustment responsibility within the window into stages according to the principle that the previous responsibility layer gradually reduces its adjustment share and the next responsibility layer gradually increases its adjustment share. Specifically, the previous responsibility layer undertakes the main compensation task in the first part of the window, the responsibility gradually transitions in the middle stage, and the next responsibility layer takes over the main adjustment task in the last part of the window. The system generates relay window data, which includes information such as window number, window start and end time, participating responsibility level, change process of responsibility handover ratio, and allowable power change rate.

[0167] S43. Generate suppression constraint data by backfilling and superimposing suppression constraints based on the relay window data;

[0168] In one embodiment, the system predicts and analyzes the power reduction of the preceding responsibility layer and the power increase of the subsequent responsibility layer for each relay window, and assesses the potential power superposition between the two within the window. The system compares the predicted superposition power with the current target power gap. When the expected superposition power exceeds the target gap and is higher than a preset safety margin, the system determines that there is a risk of backfilling and generates corresponding suppression constraints. The suppression constraints include limiting the adjustment behavior of different responsibility layers, such as limiting the maximum output power of the preceding responsibility layer in the latter part of the relay window, limiting the power increase rate of the subsequent responsibility layer in the first part of the window, setting a dynamic upper limit on the overall compensation power within the window, and controlling the number of multiple subsequent nodes that simultaneously initiate adjustment. When the system detects that slow-response devices are concentratedly entering the adjustment state, it also adds a priority suppression flag to enable fast-response devices such as energy storage to enter the power decline phase earlier. The system generates suppression constraint data, which includes information such as constraint object, constraint type, constraint threshold, corresponding relay window number, and constraint triggering and release conditions.

[0169] S44. Perform local proxy optimization based on the suppression constraint data and the equipment adjustment boundary data to obtain local cooperative strategy data;

[0170] In one embodiment, the system treats energy storage units, charging pile clusters, and other flexible loads as independent local agents and establishes a local optimization task for each agent. The system loads the responsibility layer information, the relay window involved, and the corresponding suppression constraint data for each agent, while simultaneously reading the device adjustment boundary data. This boundary data includes information such as the rated power range, minimum operating power, allowable power change rate, shortest continuous operating time, and the state of charge range of the energy storage device. The system generates a local power adjustment trajectory for each agent, minimizing local compensation deviations and reducing adjustment costs while meeting window constraints and device operating boundaries. For energy storage agents, the system prioritizes short-term support capacity and constrains the state of charge; for load agents such as charging piles, it provides adjustable load reduction space without affecting the minimum service power. The system generates local collaborative strategy data, including the target power, start adjustment time, exit adjustment time, and corresponding relay window number for each agent in different time slices.

[0171] S45. Perform cross-node collaborative correction based on local collaborative strategy data to obtain global collaborative allocation data;

[0172] In one embodiment, the system aggregates the local coordination strategies generated by each local agent to the global coordination layer and performs consistency checks on a time-by-time basis according to a unified timeline. The system statistically analyzes the compensation power of each node at the same time, determines whether the total compensation meets the target power gap requirement, and simultaneously detects whether relay window constraints are violated and whether there are adjustment conflicts between different responsibility layers. When the total compensation is insufficient at a certain time, the system prioritizes selecting nodes with adjustment margins within the same responsibility layer for supplementation; when the total compensation exceeds the target requirement, it gradually reclaims some adjustment shares from subsequent responsibility layers according to their priority order. When the system detects that multiple nodes on the same feeder or within the same control area simultaneously increase power, causing local adjustment congestion, it performs time shifting or amplitude compression on the power change process of the relevant nodes to maintain consistent adjustment rhythms within the area. After the above corrections, global coordination allocation data is formed, including the unified target power change trajectory of each node, the execution satisfaction level of each responsibility layer, the completion status of relay window handover, and the overall power balance deviation of the system.

[0173] S46. Perform compensation and exit shaping based on global collaborative allocation data to obtain collaborative control data.

[0174] In one embodiment, the system performs compensation and exit shaping processing on the target power trajectory of each node based on global collaborative allocation data, focusing on adjusting the power withdrawal process at the end of the compensation process. The system identifies the responsibility layer type of each node. When a node belongs to the transient support layer, its output power is gradually reduced according to a preset fading time after the system power gap has basically converged and the subsequent responsibility layer has stably taken over the adjustment task. When a node belongs to the steady-state regression layer, its exit is appropriately delayed after the system operation has stabilized, and the power is slowly withdrawn at a smaller rate of change. During the exit shaping process, the system simultaneously checks the exit time relationship between different nodes. When a significant misalignment is detected in the exit times of multiple nodes, the system fine-tunes the exit start time of the relevant nodes to ensure that the power withdrawal process of adjacent nodes remains sequential. The system generates collaborative control data, including equipment number, control cycle, target power value, start adjustment time, fading time, fading change rate, and abnormal replacement rules, which can be directly sent to the execution layer for control.

[0175] Optionally, this application also provides a power response and distributed optimization-based energy storage and charging coordinated control system for executing the power response and distributed optimization-based energy storage and charging coordinated control method described above. The power response and distributed optimization-based energy storage and charging coordinated control system includes:

[0176] The flexible load slicing module is used to acquire microgrid operation data; and to perform flexible load slicing based on the microgrid operation data to obtain flexible load data.

[0177] The heterogeneous communication timing evolution module is used to perform heterogeneous communication timing evolution on flexible load data to obtain heterogeneous communication data.

[0178] The virtual synchronizer control module is used to control the virtual synchronizer based on heterogeneous communication data to obtain synchronization compensation data.

[0179] The collaborative optimization module is used to perform collaborative optimization based on the synchronization compensation data to obtain collaborative control data.

Claims

1. A power storage-charging coordinated control method based on power response and distributed optimization, characterized in that, The method includes: S1. Obtain microgrid operation data; extract time envelope data from microgrid operation data to obtain time envelope data; discretize and slice the time envelope data to obtain load slice data; measure the similarity of the load slice data to obtain load similarity data; aggregate the load similarity data to obtain aggregated slice data; reconstruct flexible space based on aggregated slice data to obtain flexible load data. S2. Perform heterogeneous communication time-series evolution on the flexible load data to obtain heterogeneous communication data; S3. Perform virtual synchronous machine control based on heterogeneous communication data to obtain synchronization compensation data; S4. Perform collaborative optimization based on the synchronization compensation data to obtain collaborative control data.

2. The energy storage and charging coordinated control method based on power response and distributed optimization according to claim 1, characterized in that, Flexible spatial reconstruction includes: Based on the sliced ​​aggregated data, flexibility node mapping is performed to obtain load node data; Node associations are constructed based on load node data to obtain load association data; Flexible spatial structures are generated based on load correlation data to obtain flexible structure data; The flexible load data is obtained by integrating the adjustment capabilities based on the flexible structure data.

3. The energy storage and charging coordinated control method based on power response and distributed optimization according to claim 1, characterized in that, S2 include: Communication link identification is performed on the flexible load data to obtain communication link data; Based on the communication link data, communication delay features are extracted to obtain delay feature data; Communication timing propagation is performed based on communication delay characteristic data to obtain timing propagation data; The time-series propagation data is reconstructed by time-series folding to obtain heterogeneous communication data.

4. The energy storage and charging coordinated control method based on power response and distributed optimization according to claim 3, characterized in that, Temporal folding reconstruction includes: The propagation time reference is extracted from the time-series propagation data to obtain the time reference data. Based on the time reference data, a propagation time normalization mapping is performed to obtain time-normalized data; The propagation trajectory is folded based on the time-normalized data to obtain folded trajectory data. Heterogeneous propagation structure is reconstructed based on folded trajectory data to obtain heterogeneous communication data.

5. The energy storage and charging coordinated control method based on power response and distributed optimization according to claim 1, characterized in that, S3 includes: Based on heterogeneous communication data, communication timing mismatch is identified to obtain timing mismatch data; The response window interval is calibrated based on the time-series mismatch data to obtain the window interval data; Virtual inertia is dynamically mapped based on the data in the empty window interval to obtain inertia compensation data; Virtual damping collaborative shaping is performed based on inertia compensation data to obtain damping compensation data; Synchronous phase angle traction control is performed based on damping compensation data and local power deviation data to obtain phase angle compensation data; Synchronous compensation data is obtained by combining inertia compensation data, damping compensation data, and phase angle compensation data.

6. The energy storage and charging coordinated control method based on power response and distributed optimization according to claim 5, characterized in that, Response window interval calibration includes: Response time offset is analyzed based on timing mismatch data to obtain response offset data; The adjustment status is determined based on the response offset data to obtain the response status data. Power gap detection is performed based on the response status data to obtain gap identification data; By aggregating continuous time intervals based on the gap identification data, the gap interval data is obtained.

7. The energy storage and charging coordinated control method based on power response and distributed optimization according to claim 5, characterized in that, Virtual damping synergistic shaping includes: Based on the inertia compensation data, the inertia action range is analyzed to obtain the inertia range data; Dynamic damping base value matching is performed based on inertia interval data to obtain damping base value data; By backfilling and superimposing risk corrections on the damping baseline data, damping correction data is obtained. The damping correction data is smoothed and shaped to obtain the damping trajectory data; The damping trajectory data is encapsulated into parameters to obtain damping compensation data.

8. The energy storage and charging coordinated control method based on power response and distributed optimization according to claim 1, characterized in that, S4 includes: Based on the synchronous compensation data, compensation responsibility is stratified to obtain stratified responsibility data; Based on the hierarchical responsibility data, a time-series relay window is constructed to obtain relay window data; Based on the relay window data, suppression constraints are generated by backfilling and superimposing suppression constraints to obtain suppression constraint data; Local agent optimization is performed based on suppression constraint data and equipment adjustment boundary data to obtain local cooperative strategy data; Based on the local collaborative strategy data, cross-node collaborative correction is performed to obtain global collaborative allocation data; Compensation and exit shaping are performed based on the global collaborative allocation data to obtain collaborative control data.

9. A power storage and charging coordinated control system based on power response and distributed optimization, characterized in that, For executing the energy storage and charging coordinated control method based on power response and distributed optimization as described in claim 1, the energy storage and charging coordinated control system based on power response and distributed optimization includes: The flexible load slicing module is used to acquire microgrid operation data; and to perform flexible load slicing based on the microgrid operation data to obtain flexible load data. The heterogeneous communication timing evolution module is used to perform heterogeneous communication timing evolution on flexible load data to obtain heterogeneous communication data. The virtual synchronizer control module is used to control the virtual synchronizer based on heterogeneous communication data to obtain synchronization compensation data. The collaborative optimization module is used to perform collaborative optimization based on the synchronization compensation data to obtain collaborative control data.

Images