Command management method for ac microgrid frequency regulation system

Abstract

The application discloses an instruction management method for an alternating current micro-grid frequency regulation system, relates to the technical field of alternating current micro-grid instruction management, and comprises the following steps: monitoring a frequency regulation instruction, extracting metadata of the frequency regulation instruction, and constructing a directed acyclic graph according to a logical relationship between the instructions; then, the number of characters and the memory occupation value of the directed acyclic graph are extracted, and a current timestamp is obtained; the application realizes dynamic and lightweight real-time integrity protection; the application discloses a method for abandoning a static key or a fixed hash strategy, and creatively converting the dynamic change characteristics of the instruction chain into a verification key in real time through a nonlinear operation rule. Any slight tampering with the instruction chain will immediately cause the chain characteristics to change, and then cause the key to be unmatched in the periodic check, so that the real-time protection of the instruction flow is realized, the security mechanism and the business flow are synchronously evolved, the protection granularity is fine, and the real-time performance is high.

Description

Technical Field

[0001] This invention belongs to the field of AC microgrid command management technology, specifically a command management method for AC microgrid frequency regulation systems. Background Technology

[0002] As an autonomous power system integrating distributed energy resources, energy storage, and loads, the frequency stability of a microgrid is crucial for ensuring power quality and safe operation. In advanced microgrids using intelligent agents such as deep reinforcement learning (DRL) for automated frequency regulation, the system needs to continuously receive, generate, and execute a large number of regulation commands. These commands constitute a dynamic, interconnected, and continuously evolving command flow, the completeness and accuracy of which directly affect the stability and security of the entire power grid.

[0003] In existing technologies, the management and security protection of such command streams largely rely on traditional information security methods, such as static key encryption, fixed hash verification, or audit logs based on fixed rules. However, these methods have significant shortcomings when dealing with the specific scenario of microgrid frequency regulation: First, traditional encryption and verification methods are usually static or periodically updated, making it difficult to adapt to the dynamic characteristics of continuous and rapid evolution of command streams, and failing to achieve fine-grained, real-time integrity verification. Second, when command sources are diverse (such as autonomous decision-making by DRL agents, confirmation by manual intervention, and emergency manual operations) and the logical relationships are complex, existing methods lack the ability to finely characterize and trace the derivative relationships and sources between commands. Once an anomaly or tampering occurs, it is difficult to quickly and accurately locate the source of the problem. In addition, during long-term system operation, normal policy updates or large-scale injection of legitimate commands may cause drastic changes in the characteristics of the command chain, and traditional static verification mechanisms are prone to generating false alarms, affecting the normal operation of the system.

[0004] Therefore, there is an urgent need for a command management method that can deeply integrate microgrid frequency regulation business logic, adapt to the dynamic and correlated characteristics of command flow, and achieve accurate anomaly detection and rapid recovery, so as to improve the overall security, reliability and operation and maintenance efficiency of the system. Summary of the Invention

[0005] This invention aims to solve at least one of the technical problems existing in the prior art;

[0006] Therefore, this invention proposes a command management method for AC microgrid frequency regulation systems, comprising:

[0007] Construction steps: Monitor frequency adjustment commands, extract their metadata, and construct a directed acyclic chain based on the logical relationships between commands;

[0008] Feature extraction steps: Extract the number of characters and memory usage of the directed acyclic chain, and obtain the current timestamp;

[0009] Dynamic calculation steps: Based on the number of characters, memory usage value and timestamp, a final number is obtained by calculation according to the preset dynamic calculation rules, and the first and second positions of the three are determined according to the number;

[0010] Key generation steps: Determine the order of the number of characters, memory usage value, and the remaining two in the timestamp based on the last digit of the final number, concatenate the three in order, and generate a standard key group based on the last digit and store it in association with the chain;

[0011] Verification and recovery steps: Periodically repeat the feature extraction step to the key generation step, generate a real-time key set and compare it with the standard key set. If they are inconsistent, it is determined that the chain has been tampered with, and chain recovery and alarm are performed.

[0012] Furthermore, the metadata includes instruction content, generation source identifier, and operator identifier; wherein, the generation source identifier is a unique identifier for the deep reinforcement learning agent.

[0013] Furthermore, the dynamic calculation rules include: selecting addition, subtraction, division, or multiplication operations to obtain the final number based on the remainder of the maximum value among the number of characters, memory usage, and timestamp.

[0014] Furthermore, in the key generation step, if the units digit is an odd number, the remaining two are arranged in their original order; if it is an even number, they are arranged in a specific order of timestamp, memory usage value, and number of characters.

[0015] Furthermore, in the verification and recovery step, performing chain recovery includes: retrieving the original version of the directed acyclic chain from the backup storage to overwrite the current chain, and recording the relevant agent identifier or operator identifier as a warning object.

[0016] Furthermore, the construction steps include: classifying instructions according to the instruction generation source, and constructing an independent directed acyclic chain for each type of instruction.

[0017] Furthermore, the method also includes: extracting and fusing the feature values ​​of various instruction chains to obtain the number of fused characters, the weighted memory usage value, and the fused timestamp, and performing the dynamic calculation steps and subsequent steps based on the fused features.

[0018] Furthermore, in the key generation step, characters are selected as key characters from the total string after concatenating all classification chains.

[0019] Furthermore, the method also includes: saving historical snapshots of the directed acyclic chain; in the dynamic calculation step, calculating a confidence index based on the difference between the current chain characteristics and the historical snapshot characteristics, and dynamically adjusting the calculation parameters based on the index.

[0020] Furthermore, the verification and recovery steps include: if the confidence index is lower than the threshold, only an alarm is issued; if the confidence index is normal but the key group does not match, it is determined to be malicious tampering, and the historical snapshot with the highest confidence index is selected for recovery.

[0021] Compared with the prior art, the beneficial effects of the present invention are:

[0022] This application achieves dynamic, lightweight, real-time integrity protection. It abandons static keys or fixed hash strategies, creatively transforming the dynamically changing characteristics of the instruction chain itself into a verification key in real time through non-linear computation rules. Any minor alteration to the instruction chain will immediately cause changes in chain characteristics, leading to key mismatches during periodic verification. This achieves dynamic, real-time protection of the instruction stream, with the security mechanism evolving synchronously with the business flow, providing fine-grained protection and strong real-time performance.

[0023] Meanwhile, this application enhances anomaly tracing and precise location capabilities by constructing a fine-grained directed acyclic instruction relationship chain and recording the generation source and operator information of each instruction, providing a clear tracing path for all instructions. Further classification chain construction and differentiated verification mechanisms enable rapid location of specific instruction categories and associated agents or operators upon detection of tampering, greatly improving the efficiency of security incident analysis and response, and facilitating precise control. Attached Figure Description

[0024] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0025] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0026] Example 1:

[0027] Please see Figure 1 This application provides a command management method for an AC microgrid frequency regulation system;

[0028] As an embodiment of this application, it specifically includes:

[0029] Construction steps: Monitor frequency adjustment commands, extract their metadata, and construct a directed acyclic chain based on the logical relationships between commands;

[0030] Feature extraction steps: Extract the number of characters and memory usage of the directed acyclic chain, and obtain the current timestamp;

[0031] Dynamic calculation steps: Based on the number of characters, memory usage value and timestamp, a final number is obtained by calculation according to the preset dynamic calculation rules, and the first and second positions of the three are determined according to the number;

[0032] Key generation steps: Determine the order of the number of characters, memory usage value, and the remaining two in the timestamp based on the last digit of the final number, concatenate the three in order, and generate a standard key group based on the last digit and store it in association with the chain;

[0033] Verification and recovery steps: Periodically repeat the feature extraction step to the key generation step, generate a real-time key set and compare it with the standard key set. If they are inconsistent, it is determined that the chain has been tampered with, and chain recovery and alarm are performed.

[0034] Example 2:

[0035] As a second embodiment of this application,

[0036] This embodiment provides a command management method for an AC microgrid frequency regulation system. This method monitors and extracts features from each command issued by the regulation device throughout its entire lifecycle, constructs a traceable command relationship chain, and dynamically generates an anti-tampering verification key based on this chain, thereby achieving integrity authentication and anomaly recovery of the command stream. Specifically, it includes the following methods:

[0037] Step S1: Instruction monitoring and construction of directed acyclic chains;

[0038] The system monitors in real time every frequency regulation command received or issued by the AC microgrid frequency regulator; for each command, it automatically extracts and records the following core metadata:

[0039] Command content: that is, the specific parameters and actions of the frequency adjustment command itself;

[0040] Source ID: A unique identifier for the deep reinforcement learning agent that generated this instruction;

[0041] Operator ID: If the instruction is confirmed or issued manually, the operator's ID is recorded; otherwise, this field is empty.

[0042] The system parses the logical relationships between the above elements; each new instruction may be derived from the execution result of the previous instruction or belong to a specific DRL agent decision sequence; the system constructs these relationships into a directed acyclic chain; in this chain, nodes represent instructions and their metadata, and directed edges represent the derivation, response, or attribution relationships between instructions; this chain is persistently stored in the storage medium using a specific data structure.

[0043] Step S2: Chain feature extraction and timestamp acquisition;

[0044] The current directed acyclic chain constructed in step S1 is subjected to feature analysis. The specific analysis method is as follows:

[0045] First, analyze the string representation of the chain and count the total number of all characters it contains, denoted as the number of characters CN;

[0046] Secondly, evaluate the storage space occupied by the chain data structure in memory (in bytes), denoted as the chain storage value CS;

[0047] At the same time, obtain the time point when the directed acyclic chain was generated or last updated; convert the time point into a number string in the format of "month day hour minute" (for example, May 23, 14:09 is converted to 5231409), and record this value as the timestamp TS;

[0048] Step S3: Dynamic calculation and first-order sorting;

[0049] Each digit of the character count CN, chain storage value CS, and timestamp TS obtained in step S2 is concatenated in its original order to form a temporary number sequence.

[0050] The maximum number MAX is defined as the largest of the three values ​​CN, CS, and TS, i.e., MAX = max(CN,CS,TS).

[0051] The calculation operation identifier R is: R = MAX mod 4, where mod represents the modulo operation;

[0052] Based on the value of R, perform binary operations on the three values ​​CN, CS, and TS to obtain a final number FN. The operation rules are as follows:

[0053] If R=0 or R=1, then FN=CN+CS+TS.

[0054] If R=2, then FN=|MAXmin(CN,CS,TS)|, which means subtracting the minimum value from the maximum value.

[0055] If R=3, then FN=floor(MAX / min(CN,CS,TS)), which means dividing the maximum value by the minimum value and rounding down.

[0056] If R=4, then FN=CN×CS×TS.

[0057] After obtaining FN, take the middle digit of its entirety as the median M. If the number of digits is even, take the leftmost digit from the middle two. The value of M determines the order of CN, CS, and TS in subsequent concatenation:

[0058] If M is odd, then the first character is the number of characters CN.

[0059] If M is even and not 0, then the first bit is the linked storage value CS.

[0060] If M equals 0, then the first digit is the timestamp TS.

[0061] Step S4: Subsequent sorting and key generation based on selected numbers

[0062] Obtain the units digit of the final number FN in step S3 and denote it as the selected number SN.

[0063] Based on the parity of SN, determine the order of the remaining two values ​​among CN, CS, and TS:

[0064] If SN is odd, the remaining two values ​​are placed in the original order after the first value determined in step S3 (i.e., the original order of CN, CS, and TS after removing the first value).

[0065] If SN is even, the remaining two values ​​are placed in a specific order, after removing the first and last characters, according to the timestamp TS, chain value CS, and number of characters CN.

[0066] After steps S3 and S4, the three values ​​CN, CS, and TS are arranged in a definite order and concatenated to form a numerical string NS.

[0067] Next, using the value of SN as an index, the corresponding character is retrieved from a predefined character mapping table (e.g., 0->A, 1->B, ..., 9->J), and this character is marked as the key value KV.

[0068] At the same time, using the value of SN as the position index, select the SN-th character (or the first character if SN=0) from the string representation of the current directed acyclic chain, and mark this character as the key character KC.

[0069] Combine the key value KV with the key character KC to form a standard key group SKG=(KV,KC), and store it in association with the current directed acyclic chain.

[0070] Step S5: Periodic verification and anomaly recovery

[0071] The system is set to a fixed period (e.g., every 5 minutes or after receiving 10 new instructions) to repeatedly execute all the processes from steps S2 to S4 for the currently active directed acyclic chains in the system to generate a real-time key group (RKG).

[0072] Compare the real-time key set RKG with the pre-stored standard key set SKG:

[0073] If RKG and SKG are exactly the same, it is determined that the directed acyclic chain has not been tampered with, and the process ends.

[0074] If RKG and SKG are different, then the directed acyclic chain is determined to have been illegally tampered with.

[0075] When tampering is detected, the system immediately performs a recovery operation:

[0076] 1. Retrieve the original version of the directed acyclic chain from the backup storage and overwrite the current anomalous chain.

[0077] 2. Record and mark the main DRL agent IDs or operator identifiers involved in generating this abnormal chain, mark them as warning objects, and trigger an alarm notification to the system administrator.

[0078] Through the above embodiments, this invention achieves dynamic and lightweight integrity protection for the microgrid frequency regulation command stream. This method transforms the inherent characteristics of the command chain (number of characters, memory usage, time) into a verification key using deterministic but non-linear rules. Any minor alteration to the command chain will result in a key mismatch in subsequent periodic verifications, thereby triggering alarm and recovery mechanisms in a timely manner, ensuring the security and reliability of the frequency regulation system's command management.

[0079] Example 3:

[0080] A differentiated chain construction and verification method based on instruction source classification;

[0081] Based on the core technical ideas of Embodiment 1, this embodiment further refines the construction logic of the directed acyclic chain and introduces a differentiated processing mechanism to enhance the system's control over instructions from different sources and improve verification accuracy.

[0082] Step S1: Classification monitoring and chain structure initialization;

[0083] When monitoring frequency adjustment commands, the system first automatically classifies them according to their source:

[0084] Type A (Autonomous Agent Instructions): Directly generated by the DRL agent based on the policy model, without human intervention. Its metadata includes: instruction content, agent ID, and operator identifier is empty.

[0085] Type B (Operator Confirmation / Correction Instruction): This instruction is proposed by the DRL agent and issued after operator confirmation or correction. Its metadata includes: instruction content, agent ID, and operator identifier.

[0086] Type C (Operator Emergency Command): Initiated entirely manually by the operator for emergency intervention. Its metadata includes: command content, agent ID (empty or marked "Manual"), and operator identifier.

[0087] The system initializes an independent directed acyclic chain for each type of instruction. Nodes in the chain record complete metadata for the instructions of that type. For type B instructions, nodes are associated with both the agent ID and the operator ID, forming a bidirectional affiliation; for type C instructions, nodes are associated only with the operator ID.

[0088] Step S2: Classification Feature Fusion and Time Window Integration

[0089] For each independent classification chain (chain A, chain B, chain C), execute step S2 of Example 1 to obtain their respective CN. A CS A TS A CN B CS B TS B CN C CS C TS C .

[0090] Introducing Fusion Timestamp (FTS): Taking three timestamps (TS) A TS B TS C The largest value represents the time of the most recent instruction activity.

[0091] Introducing the Weighted Chain Value (WCS): Based on the security weights of instruction types (e.g., Class A weight 0.3, Class B weight 0.5, Class C weight 0.2), the weighted total value is calculated as: WCS = 0.3 * CS A +0.5*CS B +0.2*CS C .

[0092] In this embodiment, the core triplet used for subsequent calculations is defined as:

[0093] FCN=CN (Fused Character Count) A +CN B +CN C ;

[0094] Weighted Chained Store (WCS) (as described above);

[0095] Fusion timestamp FTS (as described above);

[0096] Step S3: Dynamic calculation and ranking based on classification features

[0097] FCN, WCS, and FTS are treated as new triples and substituted into the formula in step S3 of Example 1 for calculation. A new maximum number MAX' = max(FCN, WCS, FTS) is defined.

[0098] Calculate R' = MAX' mod 4, and based on the value of R', select the corresponding operation rules (addition, subtraction, division, multiplication) to calculate FCN, WCS, and FTS to obtain the final number FN'.

[0099] Obtain the median M' of FN', and determine the first position of FCN, WCS, and FTS in the sorting based on the parity (odd, even, non-zero, or zero) of M'.

[0100] Step S4: Cross-chain key generation and verification;

[0101] Obtain the units digit of FN' as the selected number SN';

[0102] Based on the parity of SN', determine the order of the remaining two values ​​(the rule is the same as step S4 in Example 1).

[0103] The concatenation results in a categorized and merged numerical string NS'.

[0104] Using SN' as the index, retrieve the key value KV' from the predefined mapping table;

[0105] Using SN' as the index, select the SN'th character from the total string obtained by concatenating the string representations of the three chains A, B, and C in order, and use it as the key character KC';

[0106] The standard key group SKG' = (KV', KC') is formed.

[0107] Step S5: Periodic classification verification and source location;

[0108] Repeat steps S2 to S4 periodically to generate a real-time key group RKG';

[0109] When RKG' and SKG' do not match, the system not only determines that tampering has occurred, but also initiates source tracing analysis:

[0110] 1. Calculate the real-time key group (RKG) for chains A, B, and C separately. A RKG B RKG C The differences between the key sets and their respective historical standard key sets.

[0111] 2. The chain with the greatest discrepancy is identified as the primary target chain for tampering. The system prioritizes restoring this chain from the backup.

[0112] 3. Record all associated IDs (agent IDs and / or operator identifiers) corresponding to this chain as high-priority alert targets.

[0113] This method can more accurately pinpoint the source of security incidents and improve operational efficiency.

[0114] Example 4:

[0115] An incremental verification and dynamic weight adjustment method based on historical chain snapshots;

[0116] This embodiment focuses on long-term operating microgrid systems. By introducing historical chain snapshots and dynamic weights, the verification mechanism is made adaptable in the time dimension and has stronger anti-disturbance capabilities.

[0117] Step S1: Chain versioning and snapshot management;

[0118] The system not only maintains the current directed acyclic chain, but also periodically (e.g., hourly) or based on key events (e.g., primary controller switching), generates a read-only snapshot of the current directed acyclic chain and timestamps it. TS The currently active chain and historical snapshots together constitute a chain version sequence.

[0119] Step S2: Incremental feature extraction and confidence calculation;

[0120] Each time a key needs to be generated, the system selects the currently active chain and the top N historical snapshots (e.g., N=3, i.e., the three most recent snapshots) as the processing objects;

[0121] Calculate the characteristics of the current chain: CN u CS u TS u ;

[0122] For the selected i-th historical snapshot, its characteristics are also calculated: CN Spi CS Spi TS Spi ;

[0123] Calculate the Euclidean distance D between the current chain and each historical snapshot feature. i (As a measure of difference):

[0124] Di=sqrt((CN) u -CN Spi ) 2 +(CS u -CS Spi ) 2 +(TS) u -TS Spi ) 2 )

[0125] Define the confidence index CI for the current feature: CI = 1 / (1 + avg(D1, D2, ..., D...)) N When the current chain is highly similar to a historical snapshot, CI approaches 1; when the difference is significant, CI approaches 0.

[0126] Step S3: Dynamic weight calculation and sorting

[0127] Introduce a dynamic adjustment factor DF: DF = floor(CI * 10), which maps the confidence index to integers from 0 to 10;

[0128] The number of characters in the current chain, CN. u The current chain store value CS u The dynamic adjustment factor DF is used as the triple in this embodiment;

[0129] Define MAX'' = max(CN) u CS u, DF);

[0130] Calculate R'' = (MAX'' + DF) mod 4; This formula introduces DF, so that the generation of the operation identifier R'' depends not only on the feature value, but also on the confidence of the current chain relative to the history, thereby dynamically changing the operation rules;

[0131] Based on the value of R'', for CN u CS u Perform operations on DF to obtain the final number FN'';

[0132] Obtain the median M'' of FN'', and determine the ordering rule for the first digit of the triples based on the value of M'' (as before);

[0133] Step S4: Key generation based on historical references;

[0134] Obtain the units digit of FN'' as the selected number SN'';

[0135] The order of the last two digits is determined by the parity of SN'';

[0136] Concatenate the strings to form the numerical string NS'';

[0137] Retrieve the key value KV'' using SN'' as the index;

[0138] Using (SN''+DF) mod (total number of characters in the current chain) as the index, select the key character KC'' from the current chain string. The introduction of DF allows the selection position of the key character to dynamically shift with the confidence level;

[0139] The standard key group SKG'' = (KV'', KC'') is constructed and stored together with the current CI value;

[0140] Step S5: Adaptive periodic verification and recovery;

[0141] When executed periodically, the system first calculates the current real-time confidence index (CI). u ;

[0142] If CI u If the value is below a preset threshold (e.g., 0.3), the system assumes that the chain may have undergone a drastic but not necessarily malicious, normal evolution (e.g., a large-scale policy update). In this case, only an alert is issued but the system does not immediately restore the chain, prompting the administrator to review the information.

[0143] If CI u If the generated real-time key group RKG'' does not match SKG'', it is determined to be a high probability of malicious tampering.

[0144] During recovery, the system selects the current chain CI. u The historical snapshot with the highest value is restored because it represents the most recent trusted state;

[0145] Record the abrupt change in CI values ​​before and after the tampering occurs, and mark all IDs active near that time point as warning targets;

[0146] This embodiment reduces the probability of false alarms caused by normal, drastic but legitimate changes in the instruction stream by introducing historical reference and confidence mechanisms, while enhancing the detection sensitivity to covert and gradual tampering.

[0147] Of course, this application can also incorporate the above embodiments.

[0148] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. A command management method for an AC microgrid frequency regulation system, characterized in that, include: Construction steps: Monitor frequency adjustment commands, extract their metadata, and construct a directed acyclic chain based on the logical relationships between commands; Feature extraction steps: Extract the number of characters and memory usage of the directed acyclic chain, and obtain the current timestamp; Dynamic calculation steps: Based on the number of characters, memory usage value and timestamp, a final number is obtained by calculation according to the preset dynamic calculation rules, and the first and second positions of the three are determined according to the number; Key generation steps: Determine the order of the number of characters, memory usage value, and the remaining two in the timestamp based on the last digit of the final number, concatenate the three in order, and generate a standard key group based on the last digit and store it in association with the chain; Verification and recovery steps: Periodically repeat the feature extraction step to the key generation step, generate a real-time key set and compare it with the standard key set. If they are inconsistent, it is determined that the chain has been tampered with, and chain recovery and alarm are performed.

2. The command management method for an AC microgrid frequency regulation system according to claim 1, characterized in that, Metadata includes instruction content, generation source identifier, and operator identifier; wherein, the generation source identifier is a unique identifier for the deep reinforcement learning agent.

3. The command management method for an AC microgrid frequency regulation system according to claim 1, characterized in that, The dynamic calculation rules include: selecting addition, subtraction, division, or multiplication operations to obtain the final number based on the remainder of the maximum value among the number of characters, memory usage, and timestamp.

4. The command management method for the AC microgrid frequency regulation system according to claim 1, characterized in that, In the key generation step, if the units digit is an odd number, the remaining two are arranged in their original order; if it is an even number, they are arranged in a specific order of timestamp, memory usage value, and number of characters.

5. The command management method for an AC microgrid frequency regulation system according to claim 1, characterized in that, In the verification and recovery step, chain recovery includes: retrieving the original version of the directed acyclic chain from the backup storage to overwrite the current chain, and recording the relevant agent identifier or operator identifier as a warning object.

6. The command management method for an AC microgrid frequency regulation system according to claim 1, characterized in that, The construction steps include: classifying instructions according to the instruction generation source, and constructing an independent directed acyclic chain for each type of instruction.

7. The instruction management method for an AC microgrid frequency regulation system according to claim 6, characterized in that, The method further includes: extracting feature values ​​from various instruction chains and fusing them to obtain the number of fused characters, weighted memory usage value and fused timestamp, and performing the dynamic calculation steps and subsequent steps based on the fused features.

8. The command management method for an AC microgrid frequency regulation system according to claim 7, characterized in that, In the key generation step, characters are selected as key characters from the total string after concatenating all classification chains.

9. The command management method for an AC microgrid frequency regulation system according to claim 1, characterized in that, The method further includes: saving historical snapshots of the directed acyclic chain; in the dynamic calculation step, calculating a confidence index based on the difference between the current chain characteristics and the historical snapshot characteristics, and dynamically adjusting the calculation parameters based on the index.

10. The command management method for an AC microgrid frequency regulation system according to claim 9, characterized in that, The verification and recovery steps include: if the confidence index is lower than the threshold, only an alarm is issued; if the confidence index is normal but the key group does not match, it is determined to be malicious tampering, and the historical snapshot with the highest confidence index is selected for recovery.

Images