A multi-virtual power plant-oriented aggregation regulation method and system

Abstract

The application discloses a kind of multi-virtual power plant-oriented aggregation regulation and control method and system, it is related to smart grid regulation and control technical field, including: when communication is normal, trust anchor synchronizes link health abstract and pre-sets safe operation envelope and autonomous rule;Immediately switch autonomous mode power setting after detecting link disconnection;Power deviation and event are recorded during autonomy and hash trajectory is generated;After link recovery, first two-way identity authentication is completed with quantum key handshake, then upload fingerprint set and sparse difference and receive the latest global constraint;Aggregation arranger reconstructs system model based on constraint relaxation matrix and compensation weight vector, generates optimized power instruction, and replaces autonomous instruction with slip coefficient smoothing, recursively allocates standby capacity, and finally realizes each virtual power plant from autonomous smooth transition to unified scheduling;Method considers flexibility, integrity, fairness and safety, significantly improves distributed energy regulation and control reliability and recovery efficiency.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of smart grid regulation, in particular to an aggregation regulation method and system for multiple virtual power plants. BACKGROUND

[0002] With the large-scale access of flexible resources such as photovoltaic, wind power, distributed energy storage and electric vehicles to the distribution network, the operators begin to package and aggregate the dispersed distributed energy through the virtual power plant platform, so that it has the ability to participate in active power dispatch, auxiliary services and multi-layer power market transactions. The existing technology generally adopts the mode of cloud centralized optimization + edge rapid execution: in the case of good communication link, the aggregation editor can collect the operation state and market quotation reported by each virtual power plant in real time, generate unified power instructions through hierarchical and zoned power flow calculation and economic clearing algorithm, and then send them to each trusted anchor or edge controller through edge-cloud link, and finally act on specific inverters, energy storage systems and adjustable loads.

[0003] Therefore, the industry generally configures TLS or VPN for encrypted transmission, and detects the survival of the link through the heartbeat package; but the logs need to be uploaded to the cloud or blockchain in full to ensure traceability, resulting in a large amount of log data. Therefore, when large-scale demonstration is carried out, it is found that there are situations such as power failure of cellular network base stations, misoperation of optical fiber trunks, routing blockage and network attacks on distributed energy. Once the communication quality is poor or even interrupted, the current technology will have a major vulnerability, that is, the central cannot obtain the real-time situation of the trusted anchor point, and cannot perform global security constraints; when the link is restored, the cloud side needs to pull and compare a large amount of logs to know, at this time, the system may be in a state of loss of control or in a semi-loss of control state for a period of time.

[0004] In the scenario of multiple virtual power plants accessing the power distribution network in parallel and relying on cloud-based centralized optimization, once a long-term or sudden long-term interruption occurs, or the controllable bandwidth is drastically reduced, the central aggregation orchestrator loses the ability to judge the true operating boundary of each virtual power plant and the deployment right, and the one-time overall control of the safe operating boundary will also fail. At this time, each virtual power plant will still maximize its own interests and use the previously established rules to continuously diverge in the vector frequency domain, thereby generating mutually overlapping power flow reverse transmission, node overvoltage, and primary frequency support collapse phenomena in a very short time in the time and space region; after a long time of operation, batch tripping behavior occurs due to the continuous action of the circuit breaker protection device, and a contiguous power outage affecting adjacent partitions occurs; after the communication signal is restored, only the current log can be compared with the previous batch of logs to determine the power generation, load, and power supply conditions in the same work cycle, and the use side and power supply side conditions cannot be simultaneously associated and consistent, and the fault judgment of one or both aspects can only be realized through a combination of multiple offline checking devices and manual checking.

[0005] Therefore, how can all virtual power plants maintain their own safe boundaries and run autonomously in a state of lost link connection? How to achieve data integrity checking with minimal data volume change and restore the global optimal constraint model at the fastest speed after communication is disconnected? These are the core problems that limit the reliability and scalability of the multi-virtual power plant aggregation control system. SUMMARY

[0006] (I) Technical problems solved

[0007] To overcome the problems in the prior art, the present application provides a multi-virtual power plant aggregation control method and system. When the communication is normal, the trust anchor synchronizes the link health summary and presets the safe operation envelope and autonomous rules; when the link loss is detected, the autonomous mode is switched to generate power settings; the power deviation, events, and hash trajectory are recorded to complete the recording work; if the link is restored, the fingerprint set and sparse difference are uploaded to the application server after identity authentication using quantum key handshake, and the latest global constraint is accepted; in this process, the aggregation orchestrator reconstructs the system model to generate optimal power instructions based on the constraint relaxation matrix and compensation weight vector, and finally uses the slip coefficient to replace the autonomous instructions into the standby capacity according to the formula, thereby completing the smooth transition of each virtual power plant from autonomy to unified scheduling; which can greatly improve the reliability and recovery efficiency of distributed energy regulation, and solve the technical problems mentioned in the background art.

[0008] (II) Technical solutions

[0009] To achieve the above object, the present application is realized by the following technical solutions:

[0010] A multi-virtual power plant-oriented aggregation regulation method, comprising,

[0011] When the communication is normal, a trust anchor is configured at each virtual power plant gateway, a link health summary is synchronized to the aggregation orchestrator, and a safe operation envelope and autonomous scheduling rules are preloaded to establish a decision boundary;

[0012] When the trust anchor detects a link interruption, it immediately switches to autonomous mode, calls the safe operation envelope and autonomous scheduling rules to calculate the distributed energy power setting, and ensures that the resources operate under unified constraints;

[0013] During autonomous operation, the trust anchor continuously records power deviation and event information, generates an accumulated hash root using an encryption hash, and stores it locally to form a complete and continuous verifiable operation trajectory;

[0014] After the link is restored, the trust anchor first completes two-way identity verification through quantum key handshake, then uploads the fingerprint quadruple and the current state and receives the latest global constraints issued by the aggregation orchestrator;

[0015] According to the uploaded power deviation, the aggregation orchestrator reconstructs the system constraints, generates an optimized power instruction, and replaces the autonomous power setting with a decreasing sliding coefficient to make each virtual power plant restore unified scheduling.

[0016] Finally, the trust anchor collects communication indicators such as message packet loss rate, round-trip time, and signal reliability in real time, linearly normalizes them according to the predetermined weight coefficients to obtain the link health degree, and inputs the content obtained by concatenating the link health degree and the safe operation envelope matrix into an anti-collision hash function to obtain the link health summary. The aggregation orchestrator is sent to realize version registration and consistency check.

[0017] In addition, when the link health degree is higher than the threshold, the aggregation orchestrator issues global power flow constraints to the trust anchor, the trust anchor constructs an operation safety envelope matrix containing node voltage, node power and system frequency boundary according to the constraint conditions, generates inequality autonomous scheduling rules based on the target minimum reference power offset and the aforementioned envelope matrix, and generates rule hash and returns it to the aggregation orchestrator.

[0018] In addition, when the link health degree is higher than the threshold, the aggregation orchestrator issues global power flow constraints to the trust anchor, the trust anchor constructs an operation safety envelope matrix containing node voltage, node power and system frequency boundary according to the constraint conditions, generates inequality autonomous scheduling rules based on the target minimum reference power offset and the aforementioned envelope matrix, and generates rule hash and returns it to the aggregation orchestrator.

[0019] Further, the trust anchor adopts a rolling time window to predict the load in autonomous mode, performs quadratic optimization on the power decision vector with the goal of minimizing economic bias and the shortest calculation time, and forces the calculation time not to exceed the tolerable time limit to output the autonomous power setting vector.

[0020] Further, the trust anchor concatenates the power bias vector, the alarm event vector, and the manual intervention vector into a unified sequence vector with a serialization period as the time granularity, accumulates to form a sequence matrix according to the segment window length, generates a segment hash, records the segment start and end time, and then sends it to the root chain maintenance thread.

[0021] In addition, the root chain maintenance thread will make the hash value of the current segment according to the recursive hash and add it to the global trajectory root of the last round, that is, the new root value of this round, and save the current segment in a sliding window manner. The segments that exceed this window only store their hash values in the cold archive index. When the mapping period arrives, all segments in the cold archive are re-hashed and the compressed root value is calculated.

[0022] In addition, the trust anchor and the aggregation orchestrator complete quantum random bit exchange using the BB84 protocol, obtain a quantum session key after information coordination and privacy amplification, and use the key to encrypt and decrypt the fingerprint quadruple composed of the rule hash, cumulative hash root, global trajectory root, and compressed root, and complete bidirectional verification of the fingerprint quadruple.

[0023] In addition, after the fingerprint quadruple verification passes, the trust anchor only generates an index set for the part of the power difference greater than the perception threshold value and sends it after encryption; while the aggregation orchestrator completes the above vector combination reconstruction and exponential decay coefficient determination, and sends the autonomous power setting vector and the centralized optimization power vector to the respective main control objects.

[0024] Finally, the aggregation orchestrator calculates the constraint relaxation matrix based on the power bias cumulative vector uploaded by the trust anchor and the line sensitivity matrix, and embeds it together with the compensation weight vector into a secure economic integrated optimization model containing economic cost items and line relaxation penalty items to obtain the new optimization power vector of each virtual power plant.

[0025] In addition to the above, the aggregation orchestrator allocates system-level standby power to each microgrid according to the cumulative power bias norm recursive allocation factor, forms an instruction snapshot containing the optimization power vector, the constraint relaxation matrix, the recursive allocation factor, and the sliding coefficient, encrypts it using public key cryptography, and sends it to each microgrid. Together with the original secure operating envelope matrix and the rule hash, it is written into the trust anchor mapping area to update the snapshot version information.

[0026] A multi-virtual power plant-oriented aggregation regulation system comprises:

[0027] A synchronization preset module, when communication is normal, configures a trust anchor at each virtual power plant gateway, synchronizes a link health abstract to an aggregation orchestrator, and preloads a safe operation envelope and autonomous scheduling rules to establish a decision boundary;

[0028] An out-of-contact autonomous module, when the trust anchor detects a link interruption, immediately switches to an autonomous mode, calls the safe operation envelope and autonomous scheduling rules to calculate a distributed energy power setting, and ensures that resources are operated according to unified constraints;

[0029] A deviation recording module, during autonomous operation, the trust anchor continuously records power deviation and event information, generates an accumulated hash root by using an encryption hash, and locally stores the accumulated hash root to form a complete and continuous verifiable operation track;

[0030] A safe reconnection module, after the link is restored, the trust anchor first completes bidirectional identity authentication through a quantum key handshake, then uploads a fingerprint four-tuple and a current state, and receives the latest global constraints issued by the aggregation orchestrator;

[0031] A unified re-regulation module, the aggregation orchestrator reconstructs system constraints according to uploaded power deviation, generates an optimized power instruction, and replaces the autonomous power setting with a decreasing slip coefficient to make each virtual power plant restore unified scheduling.

[0032] (Three) beneficial effects

[0033] The application provides a multi-virtual power plant-oriented aggregation regulation method and system, which has the following beneficial effects:

[0034] The above is the continuous trust of the central anchor, which completes link health synchronization, safe operation envelope matrix loading and rule hash consolidation under normal communication, uses autonomous power setting vectors to maintain unified boundary operation at the moment of out-of-contact, records complete on-site changes by using a track hash root, and finally realizes safe channel reconstruction through a quantum session key, thereby forming a complete closed and tough chain to prevent distributed resources from being out of control due to communication abnormalities.

[0035] By using the combination of a sparse difference set and a four-fingerprint, the edge side only needs to report a small amount of data to the aggregation orchestrator, so that the aggregation orchestrator can reconstruct a complete power deviation vector, and can pass through multiple historical tracks at a time, completely without relying on the process of massive log return, reducing bandwidth occupation and also greatly reducing the time of grid connection, and improving the continuity of scheduling.

[0036] The difference-driven constraint relaxation matrix is used to map the accumulated power difference in the autonomous period to the safety margin of the line, the compensation weight vector is used to convert the loss in the autonomous period to the weight of the objective function, and both are embedded into the safety and economy integrated optimization model, so that all information including physical information and economic information is calculated once in centralized scheduling, and the synchronous results of safety checking and income compensation are obtained in the solving process, highlighting the cross-layer collaborative technology innovation.

[0037] According to the disconnection time, the exponential slip decay coefficient is adaptively adjusted, the autonomous and centralized weights are adjusted, and the recursive distribution factor is intelligently tilted according to the accumulated deviation size, so as to realize the functions of inhibiting power step and voltage surge in the recovery process, and preferentially supporting the virtual power plant with the largest risk, thereby effectively increasing the dynamic stability margin of the power distribution network.

[0038] The optimization power instruction, the constraint relaxation matrix, the recursive distribution factor, the slip coefficient and the existing safety operation envelope matrix are packaged into a five-tuple, and the optimization power instruction and the like are transmitted in the mapping area once. Even if the next period is disconnected again, it can directly fall back to the current autonomous mode, and the parameter version is consistent, eliminating the cross-period symbol drift and interface mismatch problem, forming a self-increasing resilient closed loop.

[0039] The optimization power instruction, the constraint relaxation matrix, the recursive distribution factor, the slip coefficient and the existing safety operation envelope matrix are packaged into a five-tuple, and the optimization power instruction and the like are transmitted in the mapping area once. Even if the next period is disconnected again, it can directly fall back to the current autonomous mode, and the parameter version is consistent, eliminating the cross-period symbol drift and interface mismatch problem, forming a self-increasing resilient closed loop. BRIEF DESCRIPTION OF DRAWINGS

[0040] Figure 1 The application is a multi-virtual power plant aggregation control method flowchart;

[0041] Figure 2 The application is a multi-virtual power plant aggregation control system structure diagram. DETAILED DESCRIPTION

[0042] The technical solutions in the embodiments of the application will be described clearly and completely below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, not all the embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor are within the protection scope of the application.

[0043] Please refer to Figure 1This invention provides a method for aggregated control of multiple virtual power plants, comprising:

[0044] In the multi-virtual power plant aggregation and control system, the central aggregation orchestrator serves as the link to complete high-frequency interaction with each virtual power plant gateway through the edge-cloud link, thereby realizing the real-time update of the status of multiple virtual power plants and the generation of global optimization instructions.

[0045] However, uncertainty and potential intrusion make large-scale scheduling prone to resilience issues. When a link has a problem or is intruded upon, the scheduling will lack the control capability for collaborative optimization. At this time, the local constraints will be released, resulting in reverse power flow, frequency fluctuations, or even protection tripping.

[0046] The first step is to use a window with normal communication to enable the trust anchor side to complete link quality assessment, envelope loading, and rule solidification, so that all virtual power plants have the same trusted decision semantics and output verifiable health summaries.

[0047] The real-time scheduling of virtual power plants is based on cloud-based algorithms. However, the cloud cannot guarantee that the link will never lose packets. If the link fails, it will be too late to push the data down to the edge. Furthermore, there is no unified system between the central and edge, and data drift will occur once the data is transmitted back.

[0048] Therefore, it is necessary to synchronize the secure operation envelope and autonomous scheduling rules to the trust anchor that has already been deployed when the link is healthy in advance, and continuously detect the quality of the link at the edge through quantitative indicators, so as to achieve smooth switching and accurate recovery at the edge at any time.

[0049] Link health assessment, envelope loading, and rule solidification are interdependent: envelope synchronization can only be triggered when the link health reaches a threshold; autonomous scheduling rules can only be formulated based on the data stream status downloaded in the previous step after envelope synchronization is completed; and the calculation logic of the latency prediction term in the link health summary calculation is influenced by the rules.

[0050] Step 101: When communication is normal, the trust anchor collects packet loss rate, round-trip delay, and signal reliability in real time to generate link health. After synchronizing with the aggregation orchestrator, it constructs a safe operation envelope matrix based on the power flow boundary and solidifies the autonomous scheduling rules and rule hashes, providing a unified and verifiable decision baseline for future autonomous takeover.

[0051] When communication is normal, the trust anchor will first perform a fine-grained quality check on the link: three evaluation dimensions, namely message packet loss rate, round-trip latency, and signal reliability; thus quantifying the link health in three dimensions. Mapping to a unique scalar allows the center and edge sides to form the same metric coordinate system, enabling joint evaluation. The expression is as follows:

[0052]

[0053] Link Health Degree , value range , the larger the value, the more stable the link; weight coefficient satisfies , set according to the business level at initialization;

[0054] Link Packet Loss Rate , value range , indicating the message loss probability per unit time; link delay , value range , the round-trip time of the message; maximum allowed delay , determined by industry standards, for normalization; signal reliability , value range , indicating the proportion of effective communication;

[0055] The same index makes the central and edge have the same threshold for healthy identification, providing a basic condition for subsequent synchronous triggering. When the link health degree is greater than the set threshold , the current global flow constraint is issued by the aggregation orchestrator, and the trust anchor generates a safe operating envelope matrix encapsulating voltage, power and frequency boundaries. The formula is as follows:

[0056]

[0057] Among them, this parameter setting includes: ① node number , representing the number of distributed energy sources included in the virtual power plant; ② voltage upper and lower limits , determined according to the capacity of equipment in the distribution network; ③ power upper and lower limits , respectively determined by the equipment nameplate and safety margin; ④ frequency upper and lower limits , as specified in the current grid operation rules; ⑤ unit matrix , dimension , refers to mapping the edge corresponding to the scalar to each node; ⑥ the size of the safe operating envelope matrix refers to the size of the three-dimensional boundary configured for each energy node.

[0058] Matrix implementation of envelope calculation and one-key index function, convenient for subsequent autonomous scheduling to directly call, after synchronization, the current link health degree of the trust anchor and the generated safe operating envelope matrix are compressed by hash to obtain a link health summary of a certain length, and written back to the aggregation orchestrator. The link health summary is generated as follows:

[0059]

[0060] wherein: is an anti-collision secure hash function; denotes concatenation; unfolds the matrix vectorization by column; link health digest is fixed-length, used for later identity verification and consistency verification of data correctness.

[0061] By associating the quality indicators with the envelope and recording them, the health degree is proved regardless of how the envelope changes in the future, and it is traceable and tamper-proof.

[0062] Step 102: After the health degree and the envelope are verified at the same time, the trust anchor generates autonomous scheduling rules based on the minimum benchmark difference target, and writes the formed hash cross-verification to the persistent mapping area. After the verification is completed, the envelope, power benchmark and hash are realized in a three-way linkage through the means of efficiency safety period and minimum power amount strategy, and the required decision basis template is provided for edge computing at any time.

[0063] After receiving the confirmation of the aggregation arranger, the trust anchor uses the safe running envelope matrix and the current distributed energy benchmark power vector obtained by itself to automatically generate autonomous scheduling rules, and to impose constraints on the local optimization target after disconnection. The core constraint formula is:

[0064]

[0065] wherein: power decision vector dimension current autonomous optimization power setting to be solved, wherein is the number of distributed energy nodes in the virtual power plant, each node corresponding to three types of control quantities, active, reactive and frequency;

[0066] benchmark power vector , obtained by present state sampling, is the central benchmark power instruction received and solidified by the trust anchor in the normal communication stage, recording the target output of each node in centralized scheduling, which is the reference center of the minimum deviation in the autonomous period;

[0067] two-norm measures the deviation degree; is the right end unit vector; inequality indicates that all nodes are simultaneously constrained by the envelope;

[0068] Autonomous scheduling rules are established with the minimum baseline deviation target, taking into account user comfort and safety boundaries. When the terminal is out of contact, flexible tracking is carried out. Once the autonomous scheduling rules are fixed, the trust anchor will complete the timestamp of the rule generation and the link health summary of the last round After rehashing, the rule hash is formed and compared with the record on the aggregation orchestrator side:

[0069]

[0070] If the comparison is consistent, the aggregation orchestrator will return a verification pass identification, otherwise retransmission will be triggered until the hash is aligned.

[0071] Two-way verification enables the central and edge to have the same cognitive perspective of the envelope and rules, providing a synchronization basis for autonomous out-of-contact. After successful verification, the trust anchor writes the triple into the local persistent mapping area, and has its own life cycle When entering the out-of-contact state, the mapping can only be used within the life cycle . Beyond a certain time, the weight will be reduced to a safe bottom level to prevent decision drift caused by long-term offline. The use of triple properties binds the mapping envelope, radix power, and hash value to each other, making the mapping envelope exist for a certain period of time and be referenced, and preventing the misuse of rules that have been invalidated after multiple changes due to unstable links.

[0072] It should be noted that when the trust anchor generates or updates certain key data (such as rule hash, link health summary, cumulative hash root, etc.) locally on the gateway, it will send the data to the cloud-side aggregation orchestrator through the communication link. The aggregation orchestrator saves a copy of the same data in its own database.

[0073] After sending, the aggregation orchestrator will perform consistency check on the received value and the saved value field by field and bit by bit. Only when the two records are completely consistent, the aggregation orchestrator will return the verification pass confirmation information. If inconsistencies are found, it is determined that the data has been tampered with or the version is out of sync during transmission, and retransmission or rollback needs to be triggered.

[0074] Through the health measurement and envelope mapping in step 101, and the rule solidification and summary verification method in step 102, a single chain is formed between the central aggregation orchestrator and the edge trust anchor: quality assessment safety envelope autonomous rule two-way hash. This single chain tightly binds communication quality, operating boundaries, and scheduling targets. In the absence of contact, the edge side can still make some simple decisions, and in the presence of contact, the identity can be checked again and the difference can be calculated.

[0075] Through step one, the full-chain closure from link health assessment to autonomous rule solidification is realized, and the multi-virtual power plant aggregation control method also has the following three basic conditions: one is to compress the constraints of the whole global power flow to each edge node through the link health degree to form a quantitative criterion, and finally to the communication quality; two is to use the safe operation envelope matrix to compress the constraints of the whole global power flow to each edge node; three is to use the rule hash to generate an algorithm that verifies the semantic closed loop. At the same time, under this condition, the trust anchor can always obtain unified, reliable and control timeliness judgment parameters in the case of normal communication, so as to have a basis for implementing disconnection switching, deviation backtracking and resynchronization in the future.

[0076] When the communication is normal, the trust anchor has completed the first step: the link health degree , the safe operation envelope matrix , the reference power vector and the rule hash are solidified to the local mapping area, and each group of mappings is given a life cycle ; However, the power communication network is relatively fragile and cannot withstand the risk of sudden link interruption. For example, due to power grid feeder damage, cell site power failure or malicious human attacks, the generation side and the aggregation arranger will be disconnected. If the global scheduling results issued by the central optimizer cannot be obtained at this time, the distributed energy side output will change with weather conditions and load transient state, and without making unified and implementable decisions, each edge value will take independent action, resulting in reverse power flow, frequency swing, voltage out-of-limit and other problems.

[0077] Step two, when the link is broken, the trust anchor switches to autonomous mode and uses the safe operation envelope matrix and the reference power vector to construct the autonomous power setting vector , and at the same time, the power deviation vector needs to be monitored and encrypted cache needs to be implemented to establish a solid data foundation for constraint reconstruction and difference value calculation after the communication is restored.

[0078] The central-edge disconnection caused by link disconnection lacks global view in local decision-making, which may cause a lot of resource competition in the local area; if the edge side only locks the current output, it ignores the dynamic risk caused by the change of renewable load and renewable power generation, and needs an online autonomous optimization framework that can inherit the safe operation envelope matrix Meanwhile, the security envelope is updated according to the actual data, and the solution process of the optimization problem is put on the edge side, and the acceptable optimization time limit is set , which can respond to the solution in time and will not wait endlessly to ensure the safe and economic operation during the disconnection.

[0079] When the link monitoring thread detects that the health degree is lower than the threshold and no central response is received for several consecutive sampling periods, the autonomous mode is started, which consists of two parts: the autonomous optimization main line, which is responsible for outputting the autonomous power setting vector ; the second is the deviation tracking main line, which constantly calculates the power deviation vector , and saves this vector and the time corresponding to the vector as part of a ciphertext sequence as a preliminary preparation for the third step of operation trajectory recording.

[0080] Step 201, when the link health degree is continuously less than the threshold and no central response is received, the trust anchor triggers the disconnection marker and locks the latest triple; when the rule hash matches, the autonomous optimization model is started, and the autonomous power setting vector is output within the specified time, and all resources are still within the unified safety boundary.

[0081] Once the monitoring thread does not receive the central heartbeat packet and the link health degree for several consecutive periods , the trust anchor will immediately set the disconnection marker and lock the last valid triple in the mapping area ; and it will also check the rule hash whether it is consistent, if the rule hash is not consistent, the power instruction will be reduced to the bottom proportion , and it will stop after the rule hash is consistent.

[0082]

[0083] Where: the link health degree , the value , represents the quantitative value of communication quality; the threshold , the value , represents the disconnection trigger threshold; the number of consecutive periods , a positive integer, anti-shake count; the disconnection marker , a binary variable, indicating the current link state; the bottom proportion : the value , the minimum safe output coefficient of the device.

[0084] Based on the strict disconnection marker The hash consistency check ignores occasional jitter when determining disconnection, ensuring that the edge has obtained a reliable global boundary slice set when determining the global boundary slice of the updated edge;

[0085] When the target is disconnected And the hash is consistent, the trust anchor performs the rolling interval Autonomous optimization, taking the minimum power deviation as the objective function, balancing power deviation and optimization time, and achieving the optimal solution of both:

[0086]

[0087]

[0088] Among them: power decision vector , dimension , autonomous optimization to be solved power setting, corresponding to the active, reactive and frequency control of each distributed energy node; reference power vector , dimension , the value issued by the central before disconnection, the centralized scheduling instruction fixed by the trust anchor during normal communication, as the reference center for the minimum deviation in the autonomous stage;

[0089] Autonomous cost function , scalar, autonomous optimization target; weight coefficient , meet So that economy and real-time are balanced; calculation time Is a real number greater than or equal to 0, the actual calculation time consumed by the edge device in the autonomous optimization within a rolling window, measured by the trust anchor timer; safe operation envelope matrix , is a diagonal block matrix constructed by aggregating orchestrator using node voltage, power and frequency boundaries, that is, mapping the physical safety interval to the normalized matrix under the unified standard; optimization time limit , positive value, the longest tolerable optimization time under disconnection, determined by the system according to the demand of real-time control cycle; prediction window , that is, the trend part used to extract the future load curve, which is positive; power upper and lower limit vector , given by the device nameplate.

[0090] Economy and timeliness are written into the objective function to avoid overkill of the edge, and through hard limit At any time, you can get a suitable autonomous power setting vector within the controllable time delay .

[0091] Step 202: Within the autonomous period, the trust anchor calculates the power deviation vector in real time, and checks the sensitivity of the power deviation vector according to the decreasing deviation value; and hashes the power deviation, time and offline state, and updates the cumulative hash root every time the process is rolled over, and gradually forms a verifiable power deviation chain based on limited storage.

[0092] One to be autonomous power setting vector After calculation, the power deviation vector is calculated in real time , detect whether it exceeds the deviation threshold, if The deviation threshold is exceeded , adjust the part exceeding the amount according to the sensitivity coefficient , recalculate until :

[0093]

[0094] Among them: the power deviation vector , dimension , real-time deviation dimension; infinite norm , refers to the maximum value of the vector (i.e. the maximum absolute value of each component); deviation threshold , is a positive real number, used to define the maximum acceptable single node deviation; sensitivity coefficient , is a positive real number, indicating the step size; iteration index is a non-negative integer.

[0095] Through the power deviation vector Dynamic reduction makes the power output in autonomous mode not too far from the user's expectation value, and also does not exceed the safety boundary range, which can balance good experience and high stability.

[0096] After deviation checking, the trust anchor hashes the triple And write it into the ring buffer, and update the deviation cumulative hash root as the reference item of step three:

[0097]

[0098] Among them: timestamp , real number, sampling time, deviation signature , fixed length hash sequence; , the current Round cumulative hash root value, The cumulative value of the previous round, the ring buffer, the capacity Old data is overwritten according to the first-in, first-out principle. Use the cumulative hash root ​Rolling update saves log for next communication recovery, prevents log too large to block recovery.

[0099] Step two, call the out-of-contact judgment-mapping area-rolling autonomous optimization-power deviation tracking-encrypted cache string into a single chain, through autonomous power setting vector Parallel generation of deviation signature Realize the safe operation envelope matrix Accurate landing in the out-of-contact state, and provide the minimum and trusted data anchor point for the operation trajectory record of step three and the quantum key resynchronization of step four.

[0100] Step two, in the scenario of communication out-of-contact, two parallel main lines are used to ensure uninterrupted scheduling, that is, using the safe operation envelope matrix And the benchmark power vector Output autonomous power setting vector The autonomous optimization main line is both economical and can meet the online real-time calculation requirements; at the same time, it closely tracks the power deviation vector , establishes a deviation chain and uses the sensitivity value in the deviation chain, the cached error information and the encrypted hash to build a verifiable deviation chain, so that subsequent different clearing work can be smoothly carried out. The same mapping area parameters are used in the two main lines, and they run simultaneously driven by the out-of-contact mark , that is, it ensures that the entire process scheduling can produce unified constraints, while obtaining the corresponding correction ability. When entering step three, the trust anchor uses the accumulated hash root And the existing power deviation vector in the operation log to expand the operation log, so that the entire control path does not exist in the form of breakpoint connection to complete the closed-loop operation process.

[0101] In step two, the out-of-contact has output the autonomous power setting vector , and continuously calculates the power deviation vector ; and uses the deviation signature And the first layer of data fingerprints are generated locally using the accumulated hash root The longer the out-of-contact time, the greater the distributed energy output curve changes, the more the load fluctuations, and the more frequent the on-site alarm events. Simply recording the power deviation vector Cannot completely restore the autonomous process, and using a fragmented log as a constraint after restoring communication is very easy to cause power settlement errors and frequency verification errors due to inconsistent timing.

[0102] Step three, in the process of autonomous operation, high-time-resolution sampling is performed, the multi-modal data stream generated is serialized and unified, and the operation trajectory root And index them locally using sliding windows, guaranteeing data integrity but saving storage.

[0103] Autonomous phase data types are complex and diverse, including continuous power deviation vectors , discrete distributed energy alarm events , and enumerated manual intervention markers , etc. If they are placed separately, multi-channel synchronization is required when communication is restored, which will waste bandwidth and make it difficult to extract the order of decompression. Therefore, to reduce complexity, events of different modalities must be encoded on the same time axis, and the sequence must be fixed by a hash chain. The autonomous phase is also running in real time, and new logs are being generated continuously. Local storage is limited, and blindly expanding will squeeze the controller's operation space. Therefore, we must consider using sliding indexing + segmented root methods to ensure the traceability of integrity, and also consider the storage limit, which cannot increase the number of log entries indefinitely.

[0104] Step three is divided into two tightly coupled main lines: the trajectory encoding main line is used to merge different events into a segment vector at the same time granularity , and calculate the segment hash ; the root chain maintenance main line is used to roll up the continuous segment hash according to the order to generate the global trajectory root , and use a double queue management method to ensure the replayability of the recent window . The two main lines update once every time to update the cumulative hash root of step two, ensuring homologous synchronization; if the trust anchor in this step detects the corresponding life cycle threshold , it will automatically trigger history segment compression, thereby avoiding the storage burden caused by a large amount of redundant historical data.

[0105] Step 301 uses trust anchors to fix the serialization period, and uses the sequence accumulation method to accumulate the segment sliding window vector to the sequence matrix on the power deviation, alarm event, and manual intervention vectorization splicing sequence, and generates a segment hash for the global trajectory root maintenance thread to input multiple modal events with consistent granularity.

[0106] To solve the problem of unifying continuous, discrete, and enumerated events, in each serialization period , the trust anchor is constructed as a sequence of sequence vectors :

[0107]

[0108] Among them: continuous power deviation vector , dimension ​, derived from the result of the rolling optimization in step two; discrete alarm vector , dimension , each bit represents whether a type of event is triggered, alarm type number , positive integer, alarm type defined by the system;

[0109] artificially marked vector , dimension , encode the intervention type of the operator; sequence vector , dimension , at the same time form a multi-mode parallel; node number , positive integer, the number of distributed energy sources in the virtual power plant; artificial marking number , positive integer, artificial intervention action category; serialization period , positive real number, value range milliseconds to seconds; vector concatenation symbol , indicating column connection.

[0110] Vectorization operation puts data of different modalities into the same structure, which can make the hash function invariant to the input structure, avoiding the problem of hash offset caused by different input data formats, and can simplify the analysis logic in the recovery phase.

[0111] After serialization, the trust anchor accumulates all sequence vectors within the segment window length , , together to form the sequence matrix , and then apply the hash function to obtain the segment hash :

[0112]

[0113] wherein: segment hash , fixed-length byte string; segment start and end markers , record the time range, for the root chain maintenance main line reference; segment window length , positive real number, typical value number of seconds to tens of seconds; sequence matrix , size ;

[0114] The fragmentation strategy balances granularity and volume: sequence granularity is determined by the serialization period , and segment volume is determined by the segment window length . Two-stage segmentation can use either fast positioning or fine playback according to demand in the recovery period.

[0115] Step 302, the root chain maintenance thread uses the recursive hash method to hash splice the fragments into the global trajectory root, and uses a sliding window to maintain active fragments, cold archives old fragment hashes, after the fragment mapping period, cold hashes the new compressed root and the cumulative hash root, maintains historical traceability and controllable storage.

[0116] Generate a new fragment hash Then, the global trajectory root of the trust anchor is updated according to the recursion :

[0117]

[0118] At the same time, the new fragment hash is pushed into the active fragment queue together with the time mark; if the queue length exceeds the window , the oldest fragment is stored in the cold archive index and the plaintext is deleted, and the hash value is retained .

[0119] Wherein: the global trajectory root , the root value of the nth update, corresponding to the last period; the active fragment queue, capacity , maintaining the latest window for playback; the cold archive index, only saving the hash reference of the archived fragment. The recursive root chain ensures that any modification of a single fragment will affect the global trajectory root

[0120] , and the sliding window allows the trust anchor to achieve fine-grained playback of the latest fragments even with limited storage space, achieving good resource utilization while meeting security requirements. After the update is completed, the latest global trajectory root of the root chain maintenance main thread

[0121] is written to the cumulative hash root of step two, and they are kept consistent; check the difference between the autonomous mapping life cycle and the current running time . If the difference

[0122] (the early warning threshold), the pre-expiration compression process will be started. All cold archived fragments are hashed again from the active queue, and the compressed root

[0123] is output and written to the mapping expiration reminder mark area, which is used to complete the verification and establish the compressed root in step four when the quantum key is resynchronized.

[0124] ​

[0125] wherein: life cycle threshold , positive real number, mapping zone effective duration; running duration , positive real number, start from the loss of the cumulative; early warning threshold , positive real number, for compression trigger; compression root , fixed length hash, for one-time verification, for the archive segment hash;

[0126] Life cycle gate is data compression and mapping update in one and simultaneously active protection of storage space, providing a single checkpoint for step 4, without being extracted by too much cold data in the quantum key recovery phase, long resynchronization delay.

[0127] When the communication is lost, step three is to encapsulate the three key events in the serialization cycle with high resolution vector, and form a hash segment with segment window , constantly cycle into global trajectory root , update the global cumulative hash root , update the global cumulative hash root export. This structure can support both instant playback and long-term compression of history, and the linear structure makes the large amount of heterogeneous data information generated by the autonomous node process occupy a fixed size of space to keep verifiable records, and does not need to involve any quantum key handshake to use the data for very simple and fast consistency check after completing the quantum key handshake.

[0128] At the same time, the sliding window mechanism ensures that the latest segment can be taken at any time to realize fault playback or operation and maintenance analysis; cold archive compression root can effectively prevent the historical expansion problem from occupying the edge side resources. Using the safe operation envelope matrix generated in step one and mapped into the running cycle , and based on the power deviation signature generated in step two, the trajectory chain generated in this step realizes the combination of physical security, economic operation and data credibility, providing a practical and verifiable big data underlying platform for subsequent quantum key resynchronization, and providing a complete, continuous and traceable autonomous operation scenario before the multi-virtual power plant recovery centralized scheduling.

[0129] The first running cycle from this stage, the cumulative online global trajectory root and compression root , and synchronizes with the cumulative hash root, gradually recovers through on-site repair or network self-healing, and the edge detects rapid link health improvement and packet loss rate reduction, and the three-way handshake can also be completely run.

[0130] Opportunity and risk coexist in recovery: if the wrong fake control message is accepted, all the security benefits obtained in the autonomous phase will be lost; if there is no effective differential synchronization mechanism, a large amount of historical logs will also delay the speed of network access.

[0131] Step four, using quantum key distribution to ensure the authenticity of the link between the two parties and the confidentiality of the data, and achieving state reconciliation between the autonomous phase and the central side with minimal data increment, downloading the latest global constraints, and using local power instruction slip fusion to rejoin the virtual power plant in unified scheduling.

[0132] The key stream of quantum key distribution is one-time pad and cannot be cloned. If it is added to the scheduling link, any eavesdropping action will be detected immediately and suffer from the quantum bit collapse effect. However, due to the small bandwidth of the quantum channel, it cannot complete a large amount of log accumulation, so it is necessary to first use quantum key to establish a symmetric key, then use lightweight rule fingerprints, cumulative hash roots, global trajectory roots, and compressed root four-tuples for overall verification, to determine the trusted edge in advance, and finally only need to upload the difference set with the smallest difference, which can restore the central model. Therefore, this method can ensure security, maintain low bandwidth occupation, and also achieve high real-time performance.

[0133] Step 401, achieve handshake-identity-fingerprint three-ring to verify the trusted channel and ensure data integrity; and step 402 is to re-obtain constraints by reporting differences, and use slip fusion closed loop to complete the smooth splicing of local power instructions and central optimization results, both using quantum session key and fingerprint four-tuple, step 401 output through marker to determine whether step 402 is executed, maintaining seamless progression of process operation.

[0134] Step 401: After the link is restored, the trust anchor and the aggregation orchestrator complete the quantum key handshake, then use the session key to encapsulate and send the transmission rule hash, cumulative hash root, global trajectory root, and compressed root information to the central for fingerprint comparison, and return the authentication success flag to the entire cloud network fusion data center network to cloud networking system after verification.

[0135] In the process of establishing the link, the trust anchor and the aggregation orchestrator start the BB84 protocol to exchange polarization base selection and measure the quantum bits, then remove inconsistent bits to get the original key string , and then get the final quantum session key through information coordination and privacy amplification:

[0136]

[0137] wherein: privacy amplification function , original key string , performing reversible hash mapping, security parameter , is a positive real number of acceptable leakage upper bound, that is, the acceptable leakage upper limit; quantum session key , length is equal to AES-256, wherein: original key string , is a bit string measured by a quantum channel; quantum session key , binary representation, fixed bit length, used to participate in AES encryption in the following steps.

[0138] A large number of high-entropy keys are generated by quantum bits at one time, and any eavesdropping behavior is detected by error rate statistics, so as to ensure that the data of the session layer in the fingerprint reconciliation process has not been tampered with; after generating the quantum session key , it can be converted into an AES-GCM symmetric session layer, and the edge side reconciliation server sends a fingerprint quadruple to obtain a fingerprint set :

[0139]

[0140] The central side calculates the self-stored mirror image , and compares it with the stored mirror image, all of which are set to :

[0141]

[0142] wherein: fingerprint set , edge message payload, mirror image set , central archive fingerprint, authentication mark , binary quantity, determines whether to execute the subsequent process.

[0143] The quadruple fingerprint covers the envelope, power deviation, serialized trajectory, and cold archive compression, and only when any one of the chains in the fingerprint is tampered with will the fingerprint verification fail, so as to achieve the purpose of complete integrity proof with minimum upload and maximum coverage.

[0144] After authentication, the trust anchor uploads the sparse power difference component index set greater than the perception threshold to the central end, the central end reconstructs the complete deviation and issues a centralized optimized power vector, and the edge end adjusts the sliding difference decay coefficient according to the offline duration to smoothly flatten the autonomous power vector to the instructions of the centralized end.

[0145] In under the premise that in the working process, the trust anchor compares the central reference power vector with the local autonomous power setting vector to obtain a difference vector :

[0146]

[0147] In order to reduce the link load, only upload the sparse difference set greater than the threshold :

[0148]

[0149] After the corresponding components in the set are encrypted by AES-GCM, they are sent out.

[0150] Wherein: different difference vectors : dimension , edge-central power difference; threshold : positive real number, greater than zero, no inductive difference upper limit; sparse difference set , subset, refers to the uploaded entry index.

[0151] According to the perception threshold to upload the difference, the exponential compression means is adopted to make the amount of uploaded data lower than the linear growth speed, while not affecting the central side to accurately determine the change of the autonomous stage of the power flow. The sparse difference set is obtained from the central side, and the set is reconstructed into a complete difference vector , which is combined with the latest global optimization power vector to obtain a slip fusion vector , after which it is replaced by a pure global optimization power vector control, as follows:

[0152]

[0153] Wherein: synchronous time interval , actual time length of disconnection; slip decay coefficient , exponentially decreases with the synchronous time interval ;

[0154] Global optimization power vector , dimension , the latest instruction of the center; slip decay coefficient , , determines the degree of progression; exponential decay coefficient , is a positive real number, adjusts the speed of fusion; slip fusion vector , is the instruction executed in the transition period.

[0155] Exponential sliding fusion eliminates the step mutation point of autonomous instruction and centralized instruction, and suppresses voltage surge. The sliding decay coefficient is adaptively decayed according to the disconnection time. The longer the disconnection time is, the smaller the remaining autonomous weight is. The economy is considered while ensuring stability.

[0156] Step 401 outputs quantum session key and authentication mark , which is given to 402 as a trusted channel; after the completion of four-fingerprint verification, only differential vector uploading is performed, and only sparse difference set is contained , which greatly reduces the transmission bandwidth; at the same time, the exponential sliding fusion vector transitions the autonomous state to centralized optimization through the sliding decay coefficient , on the basis of which the constraint reconstruction and secondary optimization distribution of step five system are carried out, and data and power double continuity are realized.

[0157] When step four realizes link recovery, identity zero forgery, data zero omission, bandwidth zero waste, and power zero mutation need to be considered. Quantum session key ensures that the link cannot be eavesdropped, and four-fingerprint can make the central station and the edge station complete the consistency verification of this step at any time with a small amount of data; when uploading the difference vector, only the data greater than the threshold value is uploaded, and the massive log is compressed into several packages; the exponential sliding fusion realizes smooth and jump-free fusion on the physical level with adaptive weight , which ensures that the multi-virtual power plant will not cause the action of circuit breaker protection due to too low or too high voltage when reconnected to the grid. Based on the uploaded sparse difference set and the verified trajectory root, the aggregated orchestrator completes the power flow reconstruction, and the sliding decay coefficient is used as the transition period weight to optimize the model, which truly realizes smooth recovery and unified scheduling.

[0158] It is confirmed that the link is safe after quantum handshake, the differential vector has been uploaded to the edge side, and the exponential sliding fusion has been executed at the trust anchor side. These three core information fall into the hands of the aggregated orchestrator: the first type is the fingerprint four-tuple from each virtual power plant: regular hash , cumulative hash root , global trajectory root , and compressed root , which is used to build the process time chain of the entire historical autonomous period; the second type is the sparse difference set and the corresponding power difference value, which shows the gap size and their respective distribution form between the actual autonomous power and the central reference; the third type is the sliding fusion vector is running in each trust anchor side, and the sliding decay coefficient The goal of central optimization is gradually approaching.

[0159] Step five, based on the minimum patch principle, complete the constraint reconstruction and global optimization of the system on the central side, and distribute the optimized results to each virtual power plant through recursive allocation and adaptive compensation coefficients, so that the virtual power plant can smoothly transition from the sliding fusion stage to the fully unified scheduling, and output the parameter snapshot to the next scheduling period.

[0160] The power deviation in the autonomous stage is not random noise, but reflects the fine information of the edge end about local weather, load and equipment state; if this part of information is covered by the center side, it means that the knowledge of the autonomous stage is discarded, and the control goal is deviated again; on the other hand, the deviation compensation without tide flow safety check may bring new constraint conflict points after global optimization, so it is necessary to convert the autonomous deviation into a constraint relaxation coefficient, and combine it with the safety margin of the main network and the market price weight to participate in global optimization, and realize the process of compensating for the difference and absorbing autonomous knowledge at the same time during formula operation.

[0161] Step 501 uses two technical means of constraint reconstruction and relaxation coding to convert the deviation statistics analyzed by the sparse difference set and the four-element fingerprint Into a constraint relaxation matrix and a compensation weight vector; then carry out a whole period of full amount of multi-virtual power plant safety-economic integrated optimization scheduling, and issue new instructions and pack new parameters for direct basis for subsequent period decision.

[0162] In step 501, the aggregation arranger reconstructs the power deviation vector based on the uploaded sparse difference patch with the line sensitivity matrix; generates the constraint relaxation matrix from the line sensitivity matrix; converts the economic loss in the autonomous period into the compensation weight vector, and introduces the physical relaxation and economic compensation information into the safety-economic integrated optimization at the same time.

[0163] The aggregation arranger first collects all the sparse difference sets generated by the virtual power plants (The superscript indicates the virtual power plant), and after fingerprint verification, the complete deviation vector is reconstructed:

[0164]

[0165] Among them: the standard basis vector , the unit vector with the same length as the dimension of the power decision vector, the dimension , the first bit is 1 and the rest is 0;

[0166] The reconstructed deviation vector , the dimension , the completed power difference, which means the virtual power plant The complete power difference vector after recovery via sparse patches, the dimension is consistent with the secure operation envelope matrix, and the element order corresponds to the active, reactive and frequency control of each node; sparse difference set , for the virtual power plant The uploaded power difference entry number set in the link recovery phase;

[0167] Power difference component , corresponding to the index difference value; for the virtual power plant The uploaded first The number of super-threshold power difference scalars represents the instantaneous deviation of autonomous power setting and central reference power at the node; node number , the number of distributed energy sources in the virtual power plant;

[0168] Complete the sparse patch to a complete vector, and do homogeneous extension, which is convenient for subsequent constraint reconstruction operation by matrix, then the power deviation accumulation vector :

[0169]

[0170] Multiply the line sensitivity matrix to generate the constraint relaxation matrix :

[0171]

[0172] Where: the power deviation accumulation vector , represents all the deviation amounts accumulated by the node during autonomy; the line sensitivity matrix , size , is used to map node power to line flow change; the constraint relaxation matrix , size , is used to give the given relaxation amount on each restricted line; the line entry refers to the number of all monitored lines under the virtual power plant.

[0173] Convert the accumulated power difference to line-level relaxation for dynamic adjustment of the size of the security margin during global optimization, and keep the effective input of autonomous knowledge; at the same time, the weighted sum of the market price slope vector and the autonomous period economic deviation vector of the aggregation orchestrator is taken as the compensation weight vector:

[0174]

[0175] Where: the market price slope vector , dimension , comes from the day-ahead market; the autonomous period economic deviation vector The power deviation and real-time electricity price integral are obtained; the weight coefficient , meet , according to the regulatory agreement, that is, not adjust with the marketization process; the compensation weight vector , dimension , represents the dimension size for the objective function.

[0176] Therefore, both the instantaneous market signal is considered by centralized optimization, and the marginal loss borne by each virtual power plant during the autonomous period is included as a compensation factor, ensuring economic fairness.

[0177] Step 502, the central side jointly uses the compensation weight vector and the constraint relaxation matrix in the safe and economic integration model to solve the optimization power vector of each virtual power plant, and generates an instruction snapshot containing power instructions, relaxation matrix, allocation factor and slip coefficient after allocating standby power according to the cumulative deviation size, and issues and writes it into the edge mapping area for direct calling in the next period.

[0178] The aggregation orchestrator constructs a safe-economic integration optimization model:

[0179]

[0180]

[0181] Where: the number of virtual power plants , the number of virtual power plants accessed; the power decision vector , optimization variable, dimension ;

[0182] The reference power vector , the instruction of the previous cycle; the safe operation envelope matrix , solidified by step one; the relaxation matrix , see above; a norm , indicating the total amount of line relaxation; the trade-off coefficient , a positive real number, balancing economic cost and safety relaxation; the upper and lower limits of power , equipment constraints.

[0183] The first half of the objective function minimizes the power deviation according to the economic weight, and the latter part uses a norm to punish the line relaxation, forcing the algorithm to prefer to digest the deviation rather than uncontrolled relaxation; the relaxation matrix is added to the right end in the constraint, allowing limited safety margin to be transferred. The model solution is the new power vector of each virtual power plant. In order to avoid the concentration of sudden load, the central side allocates the recursive allocation factor:

[0184]

[0185] The system-level standby power is injected into the virtual power plant with the most relaxed relaxation, improving the global margin.

[0186] wherein: recursive allocation factor , proportionality coefficient, non-negative and sum to one; cumulative power deviation norm Risk measure.

[0187] The virtual power plant with greater deviation during autonomy is regarded as a risk hotspot, and is given priority to reserve power support, embodying a differentiated resilience compensation mechanism. After optimization, the central generates an instruction snapshot :

[0188]

[0189] and is issued through a session key . After receiving the trust anchor, replace the part in the slip fusion vector, and write to the next period mapping area, together with the old triple to form a five-element snapshot, and the version number is incremented.

[0190] wherein: instruction snapshot , centralized optimization output package; slip coefficient , still decreasing to zero according to the length of disconnection; five-element snapshot, as the next period trust anchor starting baseline.

[0191] Snapshot solidification connects new instructions and old security envelopes along the transverse direction, and always maintains consistency; if the next process period is disconnected again, only the trust anchor needs to be connected to adjust the new version snapshot and proceed to autonomy, without the need for further negotiation, which is the strengthening way of closed-loop resilience.

[0192] Step 501 maps the difference patches uploaded by each virtual power plant into a constraint relaxation matrix and a compensation weight vector ; Step 502 optimizes the solution of endogenous relaxation-economic target integration , and dynamically supports risk hotspots with a recursive allocation factor ; the final instruction snapshot , together with the old security envelope matrix , forms a snapshot five-element group, and in the next period, the trust anchor can directly continue to execute link health synchronization with the snapshot as a template.

[0193] After completing step five, the last hop of the multi-virtual power plant from autonomous recovery to fully unified scheduling is achieved: with the help of matrix relaxation and economic weight coupling, all implicit power and economic information during the autonomous period will be equally included in the global model to obtain the final safe and fair optimization result; increasing the recursive allocation factor to give more standby support to the more risky virtual power plant, improving the resilience of the system; package all new and old core parameters into an instruction snapshot, which can quickly find a correct decision version baseline under any communication scenario at any time point, thus providing a good idea from basic methods to practical application for achieving the best results of all multi-virtual power plants under multi-communication conditions for a long time in the future.

[0194] Referring to Figure 2 The application provides a multi-virtual power plant-oriented aggregation regulation system, comprising,

[0195] A synchronization preset module, when communication is normal, a trust anchor is configured at each virtual power plant gateway, a link health abstract is synchronized to an aggregation orchestrator, and a safe operation envelope and autonomous scheduling rules are preloaded to establish a decision boundary;

[0196] A disconnection autonomous module, when the trust anchor detects link interruption, autonomous mode is switched immediately, distributed energy power settings are calculated by calling the safe operation envelope and autonomous scheduling rules, and resources are ensured to operate under unified constraints;

[0197] A deviation recording module, during autonomous operation, the trust anchor continuously records power deviation and event information, generates an accumulated hash root by using an encryption hash, and stores it locally to form a complete and continuous verifiable operation track;

[0198] A safe reconnection module, after link recovery, the trust anchor first completes two-way identity authentication through quantum key handshake, then uploads a fingerprint four-tuple and the current state and receives the latest global constraints issued by the aggregation orchestrator;

[0199] A unified re-regulation module, the aggregation orchestrator reconstructs system constraints according to uploaded power deviation, generates optimized power instructions, and replaces the autonomous power settings with a decreasing slip coefficient to make each virtual power plant restore unified scheduling.

[0200] Those skilled in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be realized by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are realized in hardware or software depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to realize the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0201] Those skilled in the art can clearly understand the specific working process of the system, the device and the unit described above can refer to the corresponding process in the foregoing method embodiments for description convenience and brevity, and details are not described herein.

[0202] In several embodiments provided in the present application, it should be understood that the disclosed system, device and method can be implemented in other ways. For example, the device embodiments described above are merely illustrative, for example, the division of the units is only some logical function division, and actual implementation can have another division manner, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units shown or discussed can be indirect coupling or communication connection through some interfaces, devices or units, and can be electrical, mechanical or other forms.

[0203] The units described as separate components can or can not be physically separated, and the components shown as units can or can not be physical units, that is, they can be located in one place, or can be distributed on a plurality of network units. Part or all of the units can be selected to achieve the purpose of the embodiment scheme according to actual needs.

[0204] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of changes or replacements within the technical range disclosed in the present application, which should be covered in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A multi-virtual power plant oriented aggregated regulation method, characterized in that: Comprising, When the trust anchor detects link interruption, it immediately switches to autonomous mode, calls the safe operating envelope and autonomous scheduling rules to calculate the distributed energy power setting, and ensures that resources run under unified constraints; During autonomous operation, the trust anchor continuously records power deviation and event information, generates cumulative hash roots using encryption hash, and stores them locally to form a complete and continuous verifiable operation trajectory; After link recovery, the trust anchor first completes two-way identity verification through quantum key handshake, then uploads the fingerprint four-tuple and the current state and receives the latest global constraints issued by the aggregation orchestrator; The aggregation orchestrator reconstructs the system constraints according to the uploaded power deviation, generates optimized power instructions, and replaces the autonomous power setting with a decreasing sliding coefficient to restore unified scheduling for each virtual power plant.

2. The multi-virtual power plant aggregation control method of claim 1, wherein: When the link health degree is higher than the threshold, the aggregation orchestrator issues global power flow constraints, and the trust anchor constructs a safe operating envelope matrix containing node voltage, node power and system frequency boundary according to the constraints, generates inequality autonomous scheduling rules with minimum reference power deviation as target and envelope matrix as constraint, and generates rule hash and returns to the aggregation orchestrator.

3. The multi-virtual power plant aggregation control method of claim 2, wherein: When the trust anchor detects that the link health degree is lower than the trigger threshold and there is no central response within a continuous monitoring period, it sets the disconnection flag to one and locks the last envelope matrix and autonomous rules that pass the hash check, and if the rule hash is inconsistent with the local record, it reduces the power of each distributed energy source to a minimum guaranteed proportion until the hash is consistent.

4. The multi-virtual power plant aggregation control method of claim 3, wherein: In autonomous mode, the trust anchor uses a rolling time window to predict load, performs quadratic optimization on the power decision vector with minimum economic deviation and shortest calculation time as target, and forces the calculation time not to exceed the tolerable time limit to output the autonomous power setting vector.

5. The multi-virtual power plant aggregation control method of claim 4, wherein: The trust anchor concatenates the power deviation vector, the alarm event vector and the artificial intervention vector into a unified sequence vector with a serialized period as the time granularity, accumulates the sequence matrix according to the fragment window length to form the fragment hash, records the start and end time of the fragment, and sends it to the root chain maintenance thread. ​ 6. The method of claim 5, wherein: The root chain maintenance thread concatenates the latest segment hash with the previous round global trace root by recursive hashing to form a new root value, and saves active segments in a sliding window manner, with segments beyond the window only retaining hash values written to a cold archive index, while all cold archive segments are re-hashed to generate a compressed root before expiration.

7. The method of claim 6, wherein: The trust anchor and the aggregation orchestrator complete quantum random bit exchange using the BB84 protocol, obtain a quantum session key after information coordination and privacy amplification, and then encrypt and transmit the fingerprint quadruple composed of the rule hash, cumulative hash root, global trace root, and compressed root using the corresponding key and complete bidirectional verification.

8. The method of claim 7, wherein: After the fingerprint quadruple is verified, the trust anchor only generates an index set for power difference components exceeding the perception threshold and encrypts and uploads it, and the aggregation orchestrator reconstructs the complete power bias vector and performs sliding fusion of the autonomous power setting vector and the centralized optimized power vector with an exponential decay coefficient.

9. The method of claim 8, wherein: The aggregation orchestrator generates a constraint relaxation matrix based on the power bias cumulative vector uploaded by each trust anchor and the line sensitivity matrix, embeds the matrix and the compensation weight vector into a security-economic integrated optimization model containing economic cost items and line relaxation penalty items, and solves to obtain a new optimized power vector for each virtual power plant.

10. The method of claim 9, wherein: The aggregation orchestrator calculates a recursive distribution factor based on the norm of the cumulative power bias of each virtual power plant, proportionally distributes the system-level standby power, generates an instruction snapshot containing the optimized power vector, constraint relaxation matrix, recursive distribution factor, and sliding coefficient, encrypts and issues it, and writes it together with the original security operating envelope matrix and rule hash to the trust anchor mapping area to complete snapshot version upgrade.

11. A multi-virtual power plant oriented aggregated regulation system, applying the aggregated regulation method of any one of claims 1 to 10, characterized in that: including, A synchronization preset module, when communication is normal, a trust anchor is configured at each virtual power plant gateway, a link health summary is synchronized to the aggregation orchestrator, a security operating envelope and autonomous scheduling rules are preloaded to establish a decision boundary; An out-of-contact autonomous module, when the trust anchor detects link interruption, immediately switches to autonomous mode, calculates distributed energy power settings using the security operating envelope and autonomous scheduling rules to ensure that resources operate under unified constraints; A bias recording module, during autonomous operation, the trust anchor continuously records power bias and event information, generates a cumulative hash root using encrypted hashing and stores it locally to form a complete and continuously verifiable operation trace; A security reconnection module, after link recovery, the trust anchor first completes bidirectional identity verification through quantum key handshake, then uploads the fingerprint quadruple and current state and receives the latest global constraints issued by the aggregation orchestrator; The unified complex modulation module is configured to reconstruct the system constraint according to the uploaded power deviation, generate an optimized power instruction, and replace the autonomous power setting with a descending sliding coefficient to restore the unified scheduling of each virtual power plant.

Images