A virtual power plant data security protection system and method

By using trusted execution modules built into edge devices and dynamic access control tokens in the virtual power plant system, the problems of plaintext data leakage, node spoofing access, and data misalignment caused by packet loss in multi-path transmission are solved, thus achieving secure and reliable data transmission and accurate calculation.

CN123122841APending Publication Date: 2026-11-17泉州耀飏能源有限公司
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
CN202611295612.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-25
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

In existing virtual power plant systems, centralized processing by a single node can easily lead to plaintext leakage, node spoofing and message replay attacks, and data misalignment due to packet loss in multi-path transmission, affecting the accuracy of power calculation.

Method used

By embedding a trusted execution module in the edge device, the real-time physical operation data is algebraically split by using a dynamic access control token to derive a homogeneous modular group confusion factor, generating a main data packet payload and an accompanying data packet payload, which are then routed to the main operation domain and the cancellation domain for independent accumulation calculation. Finally, the decision domain performs algebraic merging and modulo dimensionality reduction operations to ensure that the data is dispersed and concealed during transmission and processing.

Benefits of technology

It ensures the security and reliability of virtual power plant data during transmission and processing, prevents plaintext privacy leaks, prevents unauthorized nodes from forging identities and accessing the network and replay attacks, and ensures the integrity and accuracy of total effective power data.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123122841A_ABST
    Figure CN123122841A_ABST
Patent Text Reader

Abstract

This invention relates to the field of data security technology and discloses a data security protection system and method for a virtual power plant. The system includes an edge terminal containing edge devices, a routing terminal containing a routing gateway, and a cloud terminal containing a decision domain, a main computation domain, and a countermeasure domain. The decision domain sends a dynamic access control token to the edge devices. The edge devices use the token to derive a homogeneous modular group confusion factor to perform algebraic splitting on the physical operation data, generating a main data packet payload and an accompanying data packet payload. These payloads are then sent via the routing gateway to the main computation domain and the countermeasure domain, respectively, to perform modular algebraic accumulation calculations and output two sets of partial sums. The decision domain merges the two sets of partial sums and performs a modulo-based dimensionality reduction operation to offset the confusion factor. This invention prevents single-point attack risks by splitting the physical operation data and distributing and concealing it in the transmission link and cloud nodes. It also solves the plaintext privacy leakage risk during the centralized calculation of massive data in a virtual power plant, ensuring the statistical security of power data.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of data security technology, specifically to a virtual power plant data security protection system and method. Background Technology

[0002] Virtual power plants face security challenges during data transmission and processing. Patent document CN116756772A discloses a blockchain-based hybrid encryption protection method; patent document CN119299147B discloses a method for generating dynamic keys by combining equipment information; and patent document CN118171293A proposes a mechanism for multi-level protection by splitting data streams. However, existing technologies rely on centralized nodes and traditional key systems. The original physical operating data is centrally decrypted and calculated on a single cloud node. When a single node is compromised or a single message is intercepted, attackers can directly reconstruct the actual power values. The existing architecture cannot distribute and conceal data across transmission links and cloud nodes, posing a risk of plaintext privacy breaches.

[0003] Existing network security protection mechanisms mostly remain at the software level. A patent document with publication number CN121637535A proposes a method for generating rate limiting mechanisms based on API call anomalies. However, conventional access authentication mechanisms do not tightly bind core data processing logic with hardware entity attributes and precise time dimensions. Unauthorized nodes can easily forge identities to access the network, and external attackers can easily launch replay attacks using intercepted historical data packets, resulting in insufficient security for edge computing nodes in the underlying data processing process.

[0004] When a multipath transmission architecture is introduced into a system, complex industrial network environments can lead to asymmetric packet loss or single-path blockage. Existing technologies lack synchronous filtering and alignment mechanisms for multiple concurrent data packets, failing to accurately identify and remove isolated packets caused by network fluctuations. This results in misaligned data packets directly participating in cloud aggregation calculations. Misaligned data packets disrupt the numerical relationships in the multipath collaborative calculation process, leading to significant deviations in the final restored total effective power data, making it difficult to guarantee the completeness and accuracy of the statistical results. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a virtual power plant data security protection system and method, which solves the problems of plaintext leakage caused by centralized processing of a single node, node spoofing and message replay attacks, and data misalignment caused by packet loss in multi-path transmission, which undermines the accuracy of power calculation.

[0006] To achieve the above objectives, the present invention provides the following technical solution: The first aspect of this invention provides a virtual power plant data security protection system, comprising: The decision domain is used to send dynamic access control tokens to edge devices; An edge device is used to acquire real-time physical operation data, derive a homogeneous module group confusion factor using the dynamic access control token, and perform algebraic splitting calculation on the real-time physical operation data using the homogeneous module group confusion factor to generate the main data packet payload and the accompanying data packet payload. A routing gateway is used to route the main data packet payload to the main computing domain and the accompanying data packet payload to the offset domain; The main operation domain is used to perform modular algebraic accumulation calculation on the main data packet payload and output the first part of the sum; The cancellation domain is used to perform a modular algebraic accumulation calculation on the accompanying data packet payload and output the second part of the sum; The decision domain is also used to obtain the sum of the first part and the sum of the second part, perform algebraic merging and modulo dimensionality reduction operations on the sum of the first part and the sum of the second part, and output the total effective power data of the virtual power plant.

[0007] Preferably, the decision domain receives the access handshake request initiated by the edge device and performs certificate authentication. After successful authentication, it allocates access control parameters that include timestamp boundaries and permission matrices. The timestamp boundaries consist of an effective start time and an expiration end time. The decision domain concatenates the hardware device identifier, the effective start time, the expiration deadline, and the permission matrix to form a basic token payload. It then uses the decision domain's private key to perform a digital signature calculation on the basic token payload to generate a token digest. Finally, it merges the basic token payload and the token digest to form the dynamic access control token.

[0008] Preferably, the edge device has a built-in trusted execution module; The decision domain encapsulates the dynamic access control token into a secure communication data packet and inserts a hardware penetration identifier into the header of the secure communication data packet. After the underlying network daemon in the edge device identifies the hardware penetration identifier, it triggers a security monitor call instruction in the kernel layer of the edge device's built-in operating system to pass through the secure communication data packet to the physical isolation memory area dedicated to the trusted execution module. The trusted execution module then extracts the dynamic access control token and stores it in an independent hardware security register.

[0009] Preferably, the edge device acquires the real-time physical operation data fed back by the underlying physical sensors via the underlying bus driver within the timestamp boundary; The trusted execution module extracts the dynamic access control token, the hardware device identifier, and the high-precision timestamp corresponding to the current business action, and concatenates them to construct the underlying obfuscation seed. It then uses an irreversible hash function to perform a hash operation on the underlying obfuscation seed and performs a modulo-dimension reduction operation on the resulting discrete hash number to derive the homogeneous modulus group obfuscation factor.

[0010] Preferably, the trusted execution module extracts the hardware device identifier and concatenates it with the high-precision timestamp, and uses the irreversible hash function to generate a sequence identifier; The trusted execution module performs an addition operation on the real-time physical operation data and the homogeneous module group confusion factor, and performs a modulo operation on the summation result using the homogeneous module constant to generate the main data packet payload; The trusted execution module uses the homogeneous modulus constant to subtract the homogeneous modulus group confusion factor and performs a modulo operation on the difference result using the homogeneous modulus constant to generate the accompanying data packet payload; The trusted execution module concatenates the main data packet payload with the corresponding sequence identifier to obtain the first concatenated data, and concatenates the accompanying data packet payload with the corresponding sequence identifier to obtain the second concatenated data. It then uses the device private key to perform signature calculations on the first concatenated data and the second concatenated data, and finally merges and encapsulates them to output the first tamper-proof data packet and the second tamper-proof data packet.

[0011] Preferably, the edge device inserts a main routing signature into the first tamper-proof data packet to form a network transmission main data packet, and inserts an accompanying routing signature into the second tamper-proof data packet to form a network transmission accompanying data packet; After receiving the network transmission master data packet carrying the master routing signature, the routing gateway forwards and delivers the network transmission master data packet to the master computing domain. After receiving the network transmission companion data packet carrying the companion routing signature, the gateway forwards the network transmission companion data packet to the offset domain.

[0012] Preferably, the main operation domain and the offset domain each perform public key validity verification; The main operation domain extracts the verification mark by aggregating the sequence marks in the message to construct a main sequence mark set, which is then uploaded separately to the decision domain. The offset domain extracts the verification mark by aggregating the sequence marks in the message to construct an accompanying sequence mark set, which is then uploaded to the decision domain. The decision domain performs a set intersection operation to extract common sequence identifier elements that exist simultaneously in the main sequence identifier set and the accompanying sequence identifier set, constructs an alignment identifier set, and sends the alignment identifier set back to the main operation domain and the cancellation domain.

[0013] Preferably, the main operation domain performs the modular algebraic summation calculation on the main data packet payload contained in the alignment identifier set, outputs the first partial sum, and counts the total number of sequence identifiers contained in the effective intersection as the first effective node number.

[0014] Preferably, the offsetting domain performs the modulo algebraic summation calculation on the accompanying data packet payload contained in the alignment identifier set, outputs the second partial sum, and counts the total number of sequence identifiers contained therein as the second effective node number; The decision domain performs a numerical consistency comparison between the number of the first effective nodes and the number of the second effective nodes. If the number of the first effective nodes and the number of the second effective nodes are consistent and greater than the minimum effective participating node threshold, the decision domain performs an algebraic addition operation on the first part and the second part, and performs the modulus reduction operation on the isomorphic modulus constant again to output the total effective power data of the virtual power plant.

[0015] Preferably, the edge device encapsulates an internally embedded hardware device identifier in the access handshake request and initiates the access handshake request.

[0016] Preferably, the effective start time is set to the time when the server system completes the current authentication in the decision domain, and the expiration deadline is determined by adding the effective start time to the token lifecycle set by the system.

[0017] Preferably, the isomorphic modulus constant is determined by combining the product of the maximum safety boundary value that the corresponding data type of the underlying physical sensor can accommodate and the maximum number of concurrent access nodes.

[0018] Preferably, the main operation domain and the offset domain are physically isolated at the same level and a hard blocking strategy is implemented for horizontal data content interaction.

[0019] Preferably, the main operation domain determines that the sequence identifier in the temporary record that is outside the alignment identifier set belongs to an isolated message and performs physical-level data erasure on the main data packet payload corresponding to the isolated message.

[0020] A second aspect of the present invention provides a method for data security protection of a virtual power plant, comprising the following steps: Send dynamic access control tokens to edge devices; Acquire real-time physical operation data, derive homogeneous module group obfuscation factor using the dynamic access control token, and perform algebraic splitting calculation on the real-time physical operation data using the homogeneous module group obfuscation factor to generate main data packet payload and accompanying data packet payload. The main data packet payload is routed and distributed to the main computation domain, and the accompanying data packet payload is routed and distributed to the offset domain; For the main data packet payload, perform modular algebraic accumulation calculation and output the first part sum; at the same time, perform modular algebraic accumulation calculation for the accompanying data packet payload and output the second part sum. Extract the first part and the second part, perform algebraic merging and modulo dimensionality reduction operations on the first part and the second part, and output the total effective power data of the virtual power plant.

[0021] This invention provides a virtual power plant data security protection system and method, which has the following beneficial effects: This invention utilizes edge devices to perform algebraic decomposition calculations on real-time physical operation data using a homogeneous module group confusion factor to generate main data packet payloads and accompanying data packet payloads. These payloads are then sent via a routing gateway to the main computation domain and the offset domain for independent accumulation calculations. Finally, the decision domain performs algebraic merging and modulo dimensionality reduction operations to offset the homogeneous module group confusion factor. The overall system architecture ensures that the original physical operation data is dispersed and concealed in both the network transmission link and the cloud processing node. Intercepting a single transmission message or breaching a single cloud computation domain cannot restore the true power value. This solves the risk of plaintext privacy leakage during the collection, transmission, and centralized calculation of massive amounts of virtual power plant data, ensuring the security and reliability of total effective power data statistics.

[0022] This invention incorporates a hardware-level trusted execution module within the edge device to process dynamic access control tokens. It also constructs an underlying obfuscation seed by combining the hardware device identifier with a high-precision timestamp corresponding to the current business action. This seed performs irreversible hash operations to derive homogeneous module group obfuscation factors. The security mechanism binds data obfuscation logic to hardware entity attributes and time dimensions, thus resolving the risk of unauthorized nodes forging identities to access the network. Simultaneously, it prevents external attackers from launching replay attacks using intercepted historical data packets, thereby improving the security of edge computing nodes in the underlying data processing process.

[0023] This invention constructs a set by extracting the sequence identifiers that have passed the verification of the main operation domain and the cancellation domain, performs an intersection operation to generate an alignment identifier set, and explicitly performs modular algebraic accumulation calculation only on the main data packet payload and the accompanying data packet payload contained in the alignment identifier set. The synchronous filtering mechanism can accurately identify and remove isolated messages caused by asymmetric packet loss or single-path blocking during network transmission, eliminate the damage of data packet misalignment to the subsequent homogeneous modulus group confusion factor offsetting process, and ensure that the final restored output of the virtual power plant total effective power data is complete and accurate. Attached Figure Description

[0024] Figure 1 This is a block diagram of the overall architecture of the virtual power plant data security protection system of the present invention; Figure 2 This is the main flowchart of the virtual power plant data security protection method of the present invention; Figure 3 This is a flowchart of the dynamic access control token generation and underlying injection steps of the present invention; Figure 4 This is a flowchart of the underlying data algebraic decomposition and anti-counterfeiting message encapsulation steps of the present invention; Figure 5 Flowchart of the network packet feature appending, targeted distribution, and lateral blocking steps of this invention; Figure 6 This is a flowchart of the encrypted message verification, state synchronization, and orphaned message removal steps of the present invention. Figure 7 This is a flowchart of the isomorphic modular algebra accumulation calculation steps within the isolated memory space of the present invention; Figure 8 This is a flowchart of the steps for symmetry verification of calculation results, data merging, and total power restoration in this invention; Figure 9 This is a comparison chart of the actual data and the split load values ​​of this invention. Figure 10 This diagram illustrates the impact of the scale of concurrent access nodes on computation time in this invention. Detailed Implementation

[0025] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. 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] See Figure 1 This invention provides a virtual power plant data security protection system, including an edge terminal, a routing terminal, and a cloud terminal.

[0027] The edge end includes multiple distributed access edge devices. The edge devices have built-in hardware-level trusted execution modules. The edge devices are connected to the underlying physical sensors through a bus to obtain real-time physical operation data. The trusted execution modules provide isolated memory and encrypted computing resources independent of the conventional operating system.

[0028] The routing end includes a routing gateway, which is configured with a bidirectional independent communication link. The routing gateway is used to perform targeted traffic splitting and network isolation distribution of data packets based on data characteristics.

[0029] The cloud is logically divided into a decision domain, a main computing domain, and an offset domain.

[0030] The decision domain has the highest system-wide access and data synthesis permissions, and is internally configured with access control credential generation logic. The main computation domain and the offset domain are physically isolated at the same level and each has its own independent access authentication key. In terms of permission division, the main computation domain only has the permission to receive the main data packet payload, and the offset domain only has the permission to receive the accompanying data packet payload; furthermore, any horizontal data content interaction between the two is prohibited.

[0031] See Figure 2 This invention provides a method for data security protection of a virtual power plant, comprising the following steps: S100, the decision domain receives the access handshake request initiated by the edge device and performs certificate authentication. After successful authentication, a dynamic access control token with timestamp boundaries is generated. The decision domain then transmits the dynamic access control token to the independent hardware security register of the trusted execution module of the corresponding edge device through secure communication data packets. S200, the edge device acquires real-time physical operation data, and the trusted execution module derives a homogeneous module group confusion factor in the set homogeneous module group space based on the internally fixed dynamic access control token, hardware device identifier and high-precision timestamp. The trusted execution module uses the homogeneous module group confusion factor to perform algebraic split calculation on the real-time physical operation data to generate the main data packet payload and the accompanying data packet payload. The trusted execution module uses its internally fixed device private key to independently sign and encapsulate the separated payload and the corresponding sequence identifier, and outputs the first tamper-proof data packet and the second tamper-proof data packet. S300, the edge device sends anti-tampering data packets to the upper layer through the first encrypted transmission tunnel and the second encrypted transmission tunnel. The routing gateway performs distribution control, routes the main network transmission data packet with the main routing signature to the main computing domain, and routes the accompanying network transmission data packet with the accompanying routing signature to the offset domain. S400, the main operation domain and the offset domain each perform public key validity verification on the received message. The main operation domain extracts the signature verification and forms a main sequence identifier set through the sequence identifier in the message. The offset domain extracts the signature verification and forms a companion sequence identifier set through the sequence identifier in the message. The main operation domain and the offset domain exchange identifier digest information through the network to obtain the alignment identifier set of the main sequence identifier set and the companion sequence identifier set. The data packet payload corresponding to the isolated message whose sequence identifier is not included in the alignment identifier set is discarded. S500, the main operation domain performs modular algebraic summation on the main data packet payload corresponding to the alignment identifier set in its independent memory space and outputs the first part of the sum, and the cancellation domain performs modular algebraic summation on the accompanying data packet payload corresponding to the alignment identifier set in its independent memory space and outputs the second part of the sum; S600, the decision domain sends a result query command to the main operation domain and the offset domain to obtain the first part of the output of the main operation domain and the number of the first effective nodes participating in the calculation, and to obtain the second part of the output of the offset domain and the number of the second effective nodes participating in the calculation. The decision domain compares the number of the first effective nodes with the number of the second effective nodes. Under the condition that the two values ​​are consistent, the decision domain performs algebraic merging operation and modulo dimensionality reduction operation on the first part of the sum and the second part of the sum in the homogeneous module group space to offset the confusion factor of the underlying homogeneous module group, and outputs the total effective power data of the virtual power plant.

[0032] To enable those skilled in the art to better understand the present invention, the following will elaborate on the technical details and specific implementation logic of the above-mentioned system operation and method execution process.

[0033] See Figure 3 In step S100, the system needs to establish a trust foundation with the distributed access nodes at the global management layer in the cloud, providing a legitimate basis for subsequent isolated computing. The specific implementation steps are as follows: S101, the decision domain receives the access handshake request initiated by the edge device. The edge device encapsulates its internally embedded hardware device identifier and a pre-issued device digital certificate in the request data packet. After receiving the request, the decision domain verifies the validity of the device digital certificate using the system's built-in root certificate authority public key and extracts the device public key from the certificate to verify the data signature of the access handshake request. For the certificate verification mechanism and two-way handshake protocol under public key infrastructure, those skilled in the art can use standard transport layer security protocols for implementation. The specific message exchange format and verification process are well-known technologies in the field and will not be elaborated here.

[0034] S102, after successful certificate authentication, the decision domain assigns corresponding access control parameters to the requesting edge device, including timestamp boundaries and a permission matrix. The timestamp boundaries consist of an effective start time and an expiration deadline. The effective start time is set to the server system time when the decision domain completes the current authentication, and the expiration deadline is determined by adding the token lifespan set by the system to the effective start time. The token lifespan value is set based on the frequency of specific scheduling instructions in the virtual power plant; for high-frequency power scheduling scenarios, the value range is set to 5 to 15 minutes. The permission matrix specifies the types of physical sensor data that the edge device is allowed to collect and report within the current timestamp boundary, as well as the authorized data flow to the network boundary.

[0035] The permission matrix is ​​presented as a two-dimensional Boolean algebra structure. The rows represent the types of physical sensor data that are allowed to be collected, including categories such as active power, reactive power, node voltage, and operating temperature. The columns represent the authorized data flows to the network boundary, which includes the main computation domain, the cancellation domain, and the gateway node. Matrix elements in the permission matrix are represented by binary values, with 1 representing authorization and 0 representing blocking. The specific mathematical expression is defined as follows: ; In the formula, Indicates assignment to the first Permission matrix for each edge device; Indicates the total number of categories of data types; Indicates the total number of nodes at the network boundary; Represents the first in the permission matrix Line 1 The cross matrix elements of the columns; when At that time, the instruction agrees to allow the edge device to collect the first... Physical sensor data and to the first Send at network boundary; when At that time, the instruction prohibits the transmission operation. Based on the two-dimensional Boolean algebra structure, the control rules clarify the precise matching relationship between the data acquisition source and the transmission endpoint.

[0036] S103, the decision domain generates a dynamic access control token based on the determined access control parameters and the hardware device identifier. Specifically, the decision domain concatenates the hardware device identifier, effective start time, expiration deadline, and permission matrix in a predetermined byte order to form the basic token payload. Then, it uses the internally securely stored decision domain private key to perform a digital signature calculation on the basic token payload to generate a token digest. Finally, the basic token payload and the token digest are merged to form a complete dynamic access control token. The mathematical expression for generating the dynamic access control token is as follows: ; ; In the formula, The decision domain is the first... A dynamic access control token generated by an edge device; For device serial number, and , This represents the total number of valid edge devices that access concurrently within the current scheduling period. Indicates the underlying token payload; Indicates the first Hardware device identifier for each edge device; Indicates the effective start time; Indicates the expiration date; Indicates that for the first Permission matrix configured for each edge device; This indicates a byte concatenation operation; This represents the private key that is stored independently within the decision domain; This represents a digital signature function performed using a private key.

[0037] The dynamic access control token generated by the above logic achieves hard binding of underlying data structures such as edge device identity information, legal operation time window, and operation permission range, ensuring that the token cannot be stripped or tampered with during network transmission or system transfer, thereby providing a secure input source with non-repudiation characteristics for subsequent steps to use the token to derive homogeneous module group confusion factors.

[0038] After generating the dynamic access control token, the system needs to securely distribute and persist the token to the protected hardware area of ​​the edge device to prevent malicious interception or tampering at the edge device's conventional operating system level. The specific implementation steps are as follows: S104, the decision domain encapsulates the generated dynamic access control token into a specific secure communication data packet. To meet the control requirements of direct hardware-level writing, a specific hardware penetration identifier is inserted into the header of the secure communication data packet according to the underlying communication protocol specifications supported by the trusted execution module built into the edge device.

[0039] After the data packet is encapsulated, the secure communication data packet with the hardware penetration identifier is sent to the corresponding edge device through the pre-established transport layer encrypted tunnel between the cloud and the edge.

[0040] S105, the network communication interface of the edge device receives the aforementioned secure communication data packet, and the underlying network daemon residing in the conventional operating system environment performs preliminary parsing of the packet header. When a hardware penetration identifier is detected, the underlying network daemon immediately suspends the application layer's decryption and reading operations on the data packet payload, and instead triggers a security monitor call instruction in the kernel layer of the edge device's built-in operating system.

[0041] The security monitor invocation instruction is a hardware-level state-switching instruction supported by the central processing unit (CPU). Execution triggers a forced suspension of the currently running process in the regular operating system, elevates the CPU's execution privileges to the highest level, and transfers system control to the dedicated security monitoring program responsible for low-level isolation verification. Through this control transfer operation, the system can bypass application-layer interception by the regular operating system and directly pass data through to the physically isolated memory region dedicated to the trusted execution module.

[0042] Executing the aforementioned call instruction changes the CPU's operating privilege level and execution state, thereby directly transmitting the complete secure communication data packet to the dedicated physically isolated memory area of ​​the Trusted Execution Module. For the underlying instruction calls and processor state switching mechanisms between the Trusted Execution Module and the conventional operating system, those skilled in the art can implement them using standard operating modes such as hardware security extension architectures. The underlying register configuration logic is well-known in the field and will not be elaborated upon here.

[0043] S106, the trusted execution module receives the transparent security communication data packet in a controlled and isolated state, and performs unpacking operation in its independent execution environment to extract the dynamic access control token carried inside.

[0044] After extraction, the trusted execution module uses the decision domain public key pre-programmed into its secure read-only memory to verify the legitimacy of the signature data attached to the token. If signature verification fails, the data is deemed to have been illegally tampered with, the data packet is discarded, and the underlying anomaly event is recorded.

[0045] S107. If the signature verification passes, it indicates that the received token was legitimately issued by the cloud decision domain and has not been corrupted during the transfer phase. The trusted execution module then deconstructs and extracts important parameters such as the timestamp boundary and permission matrix from the dynamic access control token and writes them into an internally configured independent hardware security register.

[0046] The physical address space of this independent hardware security register is completely isolated from the memory mapping mechanism of a conventional operating system, preventing any program residing in an external environment from reading, copying, or overwriting the aforementioned token parameters. After the write operation is completed, the trusted execution module constructs an encrypted receipt message indicating successful injection and sends it to the decision domain via the hardware reverse call path, thus completing the physical solidification of the underlying communication trust root.

[0047] See Figure 4 After the underlying injection of the dynamic access control token is completed, step S200 enters the data obfuscation and processing stage on the edge side, which uses the physical hardware isolation mechanism to generate encryption parameters to prevent overflow. The specific implementation steps are as follows: S201: Within its authorized timestamp boundaries, the edge device continuously collects real-time physical operation data fed back by the underlying physical sensors via the underlying bus driver. At the current data reporting moment, the trusted execution module reads the pre-fixed dynamic access control token from the independent hardware security register and simultaneously obtains the local hardware device identifier and the high-precision timestamp corresponding to the current business action.

[0048] The aforementioned real-time physical operation data is directly routed from the underlying bus to the physically isolated memory of the trusted execution module during transmission, without passing through the application layer processes of the device's regular operating system. This low-level hardware-through data acquisition method physically blocks external malicious programs from intercepting the original plaintext characteristics.

[0049] S202, to prevent large numbers generated by subsequent cryptographic hash algorithms from causing numerical bit width overflow in the registers of conventional microprocessors, the system pre-sets a homogeneous modulus constant globally. The homogeneous modulus constant is determined by comprehensively considering the product of the maximum security boundary value that the corresponding data type of the underlying physical sensor can accommodate and the maximum number of concurrent access nodes.

[0050] Specifically, if real-time physical operation data is transmitted using a standard 32-bit unsigned integer format, the trusted execution module of the edge device will employ large number arithmetic logic to support an algebraic space exceeding 32 bits. When setting the homogeneous modulus constant, a global boundary constraint must be satisfied, ensuring that the sum of the extreme physical power values ​​of all concurrently accessing edge devices within the current scheduling cycle is less than this homogeneous modulus constant. Satisfying this constraint provides a theoretical basis for lossless algebraic reconstruction during subsequent cloud-based blind computing.

[0051] S203, the Trusted Execution Module extracts the dynamic access control token, hardware device identifier, and high-precision timestamp, and concatenates them according to a predetermined byte format to construct the underlying obfuscation seed. The Trusted Execution Module performs a hash operation on the underlying obfuscation seed using an irreversible hash function, and then performs a modulo-dimension reduction operation on the resulting discrete hash number using a homogeneous modulus constant, thereby deriving a homogeneous modulus group obfuscation factor that resides in the same algebraic space as the real-time physical operation data. The specific mathematical formula for calculating the homogeneous modulus group obfuscation factor is as follows: ; In the formula, Indicates the first Homogeneous module group obfuscation factor derived from the trusted execution module of an edge device; This represents an irreversible hash function; This represents a dynamic access control token embedded within the trusted execution module; This indicates the hardware device identifier of the edge device; This represents a high-precision timestamp indicating the time of the operation. This indicates a byte concatenation operation; Represents a globally defined isomorphic modulus constant; Represents the constant modulus of isomorphism Performs the remainder and modulo operation.

[0052] For irreversible hash functions, those skilled in the art can use standard secure hash algorithms for underlying code implementation. The underlying digest generation logic is a well-known technology in this field and will not be elaborated here.

[0053] By introducing isomorphic modulus constants to perform modulo mapping, the system forcibly restricts the irregularly large numbers generated by high-order cryptographic algorithms to a range of conventional data bit widths that can be stably processed by the underlying arithmetic logic units of the basic microcontroller. This dimensionality reduction transformation mechanism for algebraic space solves the problems of register memory overflow and data truncation caused by conventional encrypted values ​​directly participating in subsequent accumulation operations.

[0054] After generating the homogeneous module group confusion factor, the trusted execution module performs algebraic decomposition and anti-counterfeiting signature solidification on the plaintext data in its physically isolated environment. The specific implementation steps are as follows: S204, the trusted execution module needs to generate a unique identifier for the current batch of data for subsequent cloud-based blind alignment and matching. The trusted execution module extracts its own hardware device identifier and the high-precision timestamp generated in the aforementioned operation steps, and concatenates the two bytes according to the predetermined data bit width rules.

[0055] The trusted execution module uses an irreversible hash function to perform a hash operation on the concatenated data, generating a non-derivative sequence identifier. This sequence identifier does not contain any information about real-time physical operation data; it serves only as a verification tag for out-of-order communication fault tolerance and packet alignment in subsequent network transmissions. The mathematical expression for this calculation process is as follows: ; In the formula, Indicates the first The sequence identifier generated by an edge device within the current processing cycle; This represents an irreversible hash function; This indicates the corresponding hardware device identifier; This represents a high-precision timestamp indicating the time of the operation. This indicates a byte concatenation operation.

[0056] S205, the trusted execution module, using a globally defined homogeneous modulus constant as a boundary, physically splits the real-time physical operation data into two mutually distributed data packet payloads within the modulus group algebra space. The trusted execution module performs an addition operation on the real-time physical operation data and the homogeneous modulus group confusion factor, and performs a modulo operation on the sum using the homogeneous modulus constant, thereby generating the main data packet payload.

[0057] The trusted execution module subtracts the homogeneous modulus confusion factor from the homogeneous modulus constant and performs a modulo operation on the difference using the homogeneous modulus constant to generate the accompanying data packet payload. The mathematical expression for the above splitting process is as follows: ; ; In the formula, This represents the payload of the generated main data packet; This represents the generated accompanying data packet payload; This represents the real-time physical operation data collected by the trusted execution module from the underlying sensors; Indicates the first Homogeneous module group obfuscation factor derived from the trusted execution module of an edge device; Represents a globally defined isomorphic modulus constant; Represents the constant modulus of isomorphism Perform the remainder mathematical operation.

[0058] Through the aforementioned modular algebraic decomposition operation, the real-time physical operation data is transformed into two data streams exhibiting random surface distribution characteristics. No external network node that intercepts a single data payload can reverse engineer and reconstruct the original physical operation characteristics.

[0059] S206, after data splitting is completed, the trusted execution module constructs independent anti-counterfeiting structures for the two data packet payloads to output the final tamper-proof data message. The trusted execution module has a device private key under an asymmetric encryption system pre-embedded in its hardware secure storage area. It concatenates the main data packet payload with the sequence identifier to obtain the first concatenated data, and simultaneously concatenates the accompanying data packet payload with the sequence identifier to obtain the second concatenated data.

[0060] The trusted execution module invokes the built-in digital signature algorithm, using the device's private key to perform signature calculations on the first and second concatenated data respectively, generating a main message signature digest and a companion message signature digest. For the specific underlying implementation of the digital signature algorithm, those skilled in the art can employ conventional schemes such as standard elliptic curve cryptography algorithms; its underlying digest calculation mechanism is well-known in the field and will not be elaborated upon here.

[0061] The trusted execution module merges and encapsulates the main data packet payload, sequence identifier, and main message signature digest, outputting a first tamper-proof data packet. It then merges and encapsulates the accompanying data packet payload, sequence identifier, and accompanying message signature digest, outputting a second tamper-proof data packet. The mathematical expression for this encapsulation process is as follows: ; ; In the formula, This indicates the first tamper-proof data packet corresponding to the main data packet payload sent to the cloud; This indicates the second tamper-proof data packet corresponding to the accompanying data packet payload sent to the cloud; Indicates the first Each edge device has a built-in, embedded device private key; This indicates a digital signature function performed using the device's private key.

[0062] After the above multi-layered encryption and encapsulation, the first and second anti-tamper data packets are directly pushed by the trusted execution module to the underlying network communication interface of the edge device, ready to be transmitted to the external network.

[0063] See Figure 5 After completing the underlying data splitting and anti-counterfeiting message encapsulation, the system needs to ensure spatial isolation and directional addressing of the two data packets during network transmission. The specific implementation steps are as follows: S301, the edge device's underlying network interface initiates a network connection request to the upper-layer routing gateway. To ensure physical or logical isolation between the two encrypted data streams during transmission, the edge device utilizes the network protocol stack to establish a first encrypted transmission tunnel and a second encrypted transmission tunnel in parallel. For the key exchange mechanism at the underlying layer of the encrypted transmission tunnel, those skilled in the art can employ the bidirectional authentication handshake process of a standard secure transport layer protocol. The underlying communication handshake process is well-known in the field and will not be elaborated upon here.

[0064] S302, before pushing data to the underlying network hardware, the edge device adds external routing feature identifiers to the first and second tamper-proof data packets output by the preceding steps. Specifically, the network protocol stack inserts a main routing feature code into the outer transport layer header of the first tamper-proof data packet, and inserts a companion routing feature code into the outer transport layer header of the second tamper-proof data packet.

[0065] External routing identifiers are represented by specific network transmission port numbers or virtual LAN labels. The process of attaching external routing identifiers only changes the outer network protocol header; it does not involve any modification or re-encapsulation of the internal encrypted payload. The external encapsulation expression for this network packet is as follows: ; ; In the formula, Indicates the first The network transmission master data packet with the master route label generated by the edge device; This indicates that the edge device generates network transmission data packets with accompanying routing labels; This indicates the encapsulation operation used to construct the network transport layer protocol header; This represents the main routing signature, which is a globally agreed-upon identifier for the system. This represents the accompanying routing signature, which is a globally agreed-upon feature of the system. This indicates a byte concatenation operation; This indicates the first tamper-proof data message output by the trusted execution module; This indicates the second tamper-proof data message output by the trusted execution module.

[0066] S303, the edge device sends the main network transmission data packet to the routing gateway in the cloud through the first encrypted transmission tunnel, and simultaneously sends the accompanying network transmission data packet to the routing gateway through the second encrypted transmission tunnel.

[0067] The routing gateway's built-in network monitoring daemon captures incoming data streams at designated network interfaces. As a boundary traffic splitter, the routing gateway lacks the private key configuration for decrypting internal payloads; it only reads and parses the outer transport layer header of the packets. By extracting and matching the primary routing signature and accompanying routing signature carried in the outer transport layer header, the routing gateway quickly determines the data type attributes of the captured packets. Using this mechanism, the routing gateway achieves rapid packet type identification without compromising the original packet encryption properties, providing a routing basis for subsequent isolation and distribution.

[0068] After the routing gateway completes the packet feature identification of the incoming data stream at the underlying layer, the system performs asymmetric targeted forwarding operations on the network traffic according to the preset network access control and isolation policies. The specific implementation steps are as follows: S304. The routing gateway maintains a statically bound routing distribution table. This routing distribution table hard maps the primary route signature to the entry address of the dedicated network segment of the primary operation domain, and hard maps the accompanying route signature to the entry address of the dedicated network segment of the offset domain.

[0069] Upon receiving the main network transmission data packet carrying the primary routing signature, the routing gateway performs network address translation (NAT) based on its forwarding table, strips the outer transport layer header, and forwards the main network transmission data packet directly to the network port of the main computing domain. Similarly, upon receiving the accompanying network transmission data packet carrying the accompanying routing signature, the routing gateway forwards the accompanying data packet directly to the network port of the offset domain. Through this targeted distribution mechanism, the two split data packets flow to their respective independent cloud computing environments.

[0070] To meet the physical and logical isolation requirements of the system architecture, the S305's main computing domain and offset domain are deployed on heterogeneous server hardware clusters or isolated virtual machine hypervisors. In the network topology configuration, no direct physical connection or shared network segment is configured between the main computing domain and the offset domain.

[0071] The system configures one-way inbound rules for the network security groups to which the main computing domain and the offset domain belong. The network boundary of the main computing domain only allows data frames whose source address is the routing gateway and conforms to the master data protocol specification to enter, while the network boundary of the offset domain only allows data frames whose source address is the routing gateway and conforms to the accompanying data protocol specification to enter. Any non-compliant data packets that do not conform to the binding are directly dropped and blocked by the underlying firewall at the network segment edge of both domains.

[0072] S306: The system implements a hard-blocking strategy for lateral data content interaction between the main computing domain and the offsetting domain. At the underlying operating system network communication protocol stack of both domains, cross-domain Transmission Control Protocol (TCP) and User Datagram Protocol (UDP) connection establishment permissions are completely disabled. Furthermore, in the underlying virtual switch configuration between the two domains, bidirectional route reachability is set to black hole routing mode.

[0073] When a process within the main computing domain attempts to initiate any communication request containing actual data payload to a network segment in the offset domain, the underlying virtual switch will forcibly redirect the request to a null interface to silently drop the data packet. For the specific implementation of the underlying firewall rule configuration and blackhole routing instructions, those skilled in the art can use the rule distribution mechanism of a standard software-defined network controller. Its network access control list and null interface configuration mechanism are well-known technologies in the field and will not be elaborated upon here.

[0074] Implementing the above-mentioned lateral blocking strategy cuts off the illegal splicing path between the main data packet payload and the accompanying data packet payload from the bottom of the network topology before reaching the final calculation and aggregation stage. Even if any single computing domain is compromised, the attacker cannot obtain the complementary payload parameters required to reconstruct the real-time physical operation data.

[0075] See Figure 6 After the main operation domain and the offset domain receive the directed distribution messages respectively, the system needs to independently verify the legality of the messages in a physically isolated environment and extract alignment features. The specific implementation steps are as follows: In step S401, the main processing domain receives the main network transmission data packet delivered by the routing gateway, while the cancellation domain receives the accompanying network transmission data packet delivered by the routing gateway. Based on the aforementioned lateral blocking strategy, the two domains initiate the data parsing process in parallel without any interaction.

[0076] The built-in data parsing service in the main computation domain performs transport layer unpacking on the arriving network transmission master data packet, extracting the internally encapsulated first tamper-proof data packet. The main computation domain further deconstructs the first tamper-proof data packet, separating the master data packet payload, sequence identifier, and master packet signature digest. The built-in data parsing service in the offset domain unpacks the arriving network transmission accompanying data packet, extracting the second tamper-proof data packet, and separating the accompanying data packet payload, corresponding sequence identifier, and accompanying packet signature digest.

[0077] In step S402, the main processing domain and the offsetting domain each retrieve the public key of their respective edge devices using the system's built-in certificate management component. This public key exists in pairs with the device private key previously embedded in the edge device, and is used to verify the legitimacy of the encrypted message's origin.

[0078] The main operation field invokes the signature verification algorithm, using the acquired public key to verify the legitimacy of the concatenated data of the main data packet payload and sequence identifier, comparing the calculated digest with the extracted main message signature digest. The cancellation field similarly performs signature verification on the concatenated data of the accompanying data packet payload and sequence identifier. For the digital signature verification process under a public key system, those skilled in the art can use conventional schemes such as the standard elliptic curve digital signature algorithm. The underlying hash comparison and elliptic curve multiplication mechanisms are well-known technologies in the field and will not be elaborated upon here. If the signature verification process in any field fails, it indicates that the data has been tampered with during transmission. The corresponding main operation field directly discards the message data and records a security alarm.

[0079] S403, at the end of the current preset scheduling period, the main operation domain and the offset domain respectively summarize the valid data messages that have passed the verification. The value of the scheduling period is determined by the business response requirements of the virtual power plant. To ensure the real-time performance of power dispatching, the value range is set from 1 second to 10 seconds.

[0080] The main processing domain extracts sequence identifiers from all first-time tamper-proof data packets that have passed signature verification and aggregates them to construct a main sequence identifier set. The main processing domain temporarily stores the main data packet payload and its corresponding sequence identifier as key-value pairs in its own isolated in-memory database. The offsetting domain extracts sequence identifiers from all second-time tamper-proof data packets that have passed signature verification and aggregates them to construct an accompanying sequence identifier set. The offsetting domain also temporarily stores the accompanying data packet payload and its corresponding sequence identifier in its own isolated in-memory database. The mathematical expression for constructing the identifier set is as follows: ; ; In the formula, This represents the set of main sequence identifiers constructed from the main operation domain; This represents the set of accompanying sequence identifiers constructed by the cancellation domain; The first one extracted from the main operation domain The sequence identifier of a valid message; The first term extracted from the offset region The sequence identifier of a valid message; Indicates the total number of valid messages received by the main operation field; This indicates the total number of valid messages received by the cancellation field. Due to the inherent possibility of out-of-order delivery or packet loss in network transmission... and The values ​​are usually not equal before alignment is performed.

[0081] By constructing independent sets of sequence identifiers, the system provides fundamental characteristic credentials for subsequent blind alignment matching in the decision domain without the risk of data leakage. This process isolates the payload content related to real-time physical operation data throughout, ensuring the security boundaries of the underlying logical operations.

[0082] After the sequence identifier sets are constructed in the main operation domain and the offset domain respectively, the system performs message state synchronization according to the payload-free alignment mechanism and clears invalid data caused by network anomalies. The specific implementation steps are as follows: S404, the main computing domain uploads the master sequence identifier set separately to the decision domain in the system cloud architecture through the underlying encrypted management link. The offset domain synchronization uploads the accompanying sequence identifier set to the decision domain.

[0083] The upload process only transmits sequence identifiers generated by an irreversible hash algorithm, without carrying any main data packet payload or accompanying data packet payload. The information synchronization method based on digest exchange ensures that the decision domain, while coordinating the overall message alignment status, cannot access the underlying physical operational data, thus maintaining the confidentiality boundary of data processing at the system architecture level.

[0084] S405, the cloud computing node housing the decision domain integrates a dedicated logic operation unit responsible for handling algebraic and set operations. The decision domain receives the main sequence identifier set and the accompanying sequence identifier set, and performs set intersection operations within the logic operation unit inside the decision domain.

[0085] The purpose of set intersection is to extract common sequence identifier elements that exist simultaneously in two sets, thereby constructing a globally unified alignment identifier set. Elements in the alignment identifier set represent paired data packets that, within the current scheduling cycle, not only completely penetrate the network transport layer but also successfully pass ciphertext validity checks in both independent operation domains. The mathematical expression for set intersection is as follows: ; In the formula, This represents the set of alignment identifiers generated by the decision domain computation; This represents the set of main sequence identifiers constructed from the main operation domain; This represents the set of accompanying sequence identifiers constructed by the cancellation domain; This represents the intersection operation in mathematical sets.

[0086] S406, the decision domain will send the calculated set of alignment identifiers back to the main operation domain and the offset domain respectively through the encrypted management link.

[0087] The main computation domain receives the alignment identifier set and iterates through the previously temporarily stored key-value pair records in the internal isolated in-memory database. For sequence identifiers in the temporarily stored records that are outside the alignment identifier set, the main computation domain determines that the corresponding message has transmission anomalies such as one-sided packet loss or network out-of-order delivery, and belongs to an isolated message that cannot participate in subsequent joint computation. The main computation domain performs physical-level data erasure of the master data packet payload and sequence identifier corresponding to the isolated message from the in-memory database.

[0088] S407, after receiving the alignment identifier set, the offsetting domain synchronously performs a message trimming operation. The offsetting domain compares the temporary records in its own memory database, locates the sequence identifiers not included in the alignment identifier set, and executes a destruction instruction on the associated accompanying data packet payload. For the data erasure and overwrite mechanism of the underlying memory database, those skilled in the art can use standard secure erasure algorithms; the underlying storage block zero-value filling logic is a well-known technology in the field and will not be elaborated upon here.

[0089] After a bidirectional pruning operation based on the alignment identifier set, only valid data payloads with fully synchronized states and one-to-one correspondences are retained in the memory databases of the main computation domain and the offset domain. The deep pruning of the underlying storage space eliminates interference caused by communication link jitter, thus removing potential data misalignment risks for subsequent cross-domain isomorphic accumulation calculations.

[0090] See Figure 7 After completing the intersection calculation and orphan message removal based on digest exchange, the main operation domain performs isomorphic modular algebra accumulation calculation on the filtered legitimate data in its independent memory space. The specific implementation steps are as follows: S501, the main computation domain, based on the valid intersection obtained in the previous step, retrieves all main data packet payloads whose sequence identifiers are contained within the valid intersection from the internal temporary storage database. The main computation domain then migrates the extracted main data packet payloads to a kernel-level allocated independent memory space.

[0091] This independent memory space is configured with non-shared page attributes enforced by the hardware memory management unit, prohibiting other application-level processes of the regular operating system from reading or accessing data in this memory segment. Through permission isolation of the underlying memory page table, the system cuts off data leakage paths caused by bypass data eavesdropping and memory dumping at the physical memory level.

[0092] In S502, the main computation domain initializes a dedicated accumulator register in an independent memory space, configuring the initial value of the accumulator register to absolute zero. The main computation domain synchronously counts the total number of sequence identifiers contained in the current valid intersection, and stores the statistical result as the number of the first valid nodes securely.

[0093] The number of first valid nodes accurately reflects the scale of effective data sources actually participating in the accumulation operation within the main computation domain during the current computation cycle. This value not only serves as a boundary control parameter for the current loop computation but also as a core credential for the decision domain to perform symmetry security checks in subsequent system architectures.

[0094] S503, the main computing domain reads the system-wide uniformly set homogeneous modulus constant from the local security configuration file. The value of this homogeneous modulus constant is completely consistent with the parameters used by the edge devices when performing algebraic partitioning of the original physical data, so as to ensure the encapsulation of operations within the same algebraic space.

[0095] The main arithmetic domain performs iterative accumulation operations on the main data packet payload within an independent memory space. After each addition of a single main data packet payload value to the accumulation register, the main arithmetic domain immediately performs a modulo-dimension reduction operation on the current accumulation result using a homogeneous modulo constant, ensuring that any intermediate state value during the operation is strictly limited to the maximum bit width range of the basic arithmetic logic unit. The specific mathematical expression for this modular algebraic accumulation calculation is as follows: ; In the formula, Indicates after the first After each iteration, the intermediate states in the accumulation register are summed. This process is repeated until the iteration is complete. At that time, the final output Defined as the first part of the sum generated by the main operation domain; The absolute zero value configured for system initialization; This is the index for the order of iterative calculations, with values ​​increasing from 1 to... ; This indicates the number of the first valid nodes obtained from the current statistics; Indicates participation in the The main data packet payload is calculated in the next iteration; This represents a globally defined isomorphic modulus constant. Represents the constant modulus of isomorphism Perform the remainder mathematical operation.

[0096] S504, after total After several iterations of modular algebraic computation, the main computational domain outputs the first part of the sum. This is due to the payload of all main data packets involved in the computation. Both are ciphertext forms that incorporate the confusion factor of the isomorphic module group. The first part generated by accumulation also presents a confused value in terms of algebraic characteristics that cannot reflect the real physical operation data.

[0097] After confirming the success of the first part and the output, the main computation domain executes an immediate erase instruction on the physical storage block containing the original main data packet payload in its independent memory space. For the specific implementation of the underlying memory overwrite and destruction instructions, those skilled in the art can use standard memory zeroing algorithms or multi-round random number filling strategies. The underlying storage block reset logic is well-known in the field and will not be elaborated here. After the data erasure operation is completed, the main computation domain retains only the first part and the number of the first valid nodes to await query and scheduling by the decision domain.

[0098] During the synchronous execution phase of the accumulation calculation in the main computation domain, the cancellation domain also performs isomorphic modular algebraic accumulation calculation on the filtered legitimate accompanying data in its independent memory space. The specific implementation steps are as follows: In step S505, the cancellation domain retrieves all accompanying data packet payloads whose sequence identifiers are included in the valid intersection from the internal temporary storage database, based on the valid intersection obtained synchronously in the preceding steps. The cancellation domain then uniformly migrates the extracted accompanying data packet payloads to an independent memory space allocated at the kernel level.

[0099] The aforementioned independent memory space is configured with non-shared page attributes forcibly set by the underlying hardware memory management unit, prohibiting other application-level processes of the conventional operating system from reading or accessing memory segments. Through the system's underlying memory page table permission isolation mechanism, the offset domain prevents by-channel eavesdropping programs from stealing accompanying data payloads at the physical memory level.

[0100] S506, the cancellation field initializes a dedicated accompanying accumulator register in an independent memory space, configuring the initial value of the accompanying accumulator register to an absolute zero value. Simultaneously, the cancellation field counts the total number of sequence identifiers contained in the valid intersection, and temporarily stores the statistical result as the number of second valid nodes.

[0101] The second effective node count maps to the actual scale of effective data sources participating in the accumulation of the offset domain within the current computation cycle. Setting a second effective node count helps provide a verification benchmark for the decision domain to perform network link symmetry security checks in subsequent system architectures.

[0102] S507, the cancellation domain extracts a globally uniformly set isomorphic modulus constant from the local security configuration file. The isomorphic modulus constant is consistent with the parameters used by the main operation domain to ensure the operational closure of the two-domain data within the same algebraic space and the feasibility of subsequent restoration.

[0103] The cancellation domain performs an iterative accumulation operation on the accompanying data packet payload within an independent memory space. After each accompanying data packet payload value is added to the accompanying accumulation register, the cancellation domain immediately performs a modulo-dimension reduction operation on the current accumulation result using a homogeneous modulus constant, forcibly constraining the intermediate state value within the maximum bit width range of the basic microcontroller's arithmetic logic module. The specific mathematical expression for the above modular algebraic accumulation calculation is as follows: ; In the formula, Indicates after the first After each iteration of the cumulative calculation, the intermediate state sum in the accumulation register is added. When the iteration is complete... At that time, the final output Defined as the second part of the sum generated by the offset domain calculation; The absolute zero value configured for system initialization; This is the index for the order of iterative calculations, with values ​​increasing from 1 to... ; This indicates the number of second-valid nodes obtained from the current statistics; Indicates the extracted first One valid accompanying data packet payload; Represents a globally defined isomorphic modulus constant; Represents the constant modulus of isomorphism Perform the remainder mathematical operation.

[0104] S508, Total Completed After several modular algebraic iterations, the cancellation domain outputs the second part of the sum. This is due to all valid accompanying data packet payloads. All of these are encrypted data generated after algebraic splitting in the early stage. The second part generated by the cancellation domain also presents confused values ​​that cannot reflect the real physical operation data in terms of algebraic characteristics.

[0105] After confirming the second part and successful output, the offset domain executes an immediate erase instruction on the physical storage block containing the original accompanying data packet payload in its independent memory space. For the specific implementation of the underlying memory overwrite and destruction instructions, those skilled in the art can use standard memory zeroing algorithms or multi-round random number filling strategies. The underlying storage block reset logic is well-known in the art and will not be elaborated here. After the erase operation is complete, the offset domain retains only the second part and the number of second valid nodes, awaiting query and scheduling by the decision domain.

[0106] See Figure 8 After the main computation domain and the cancellation domain have each completed the blind computation of the encrypted data, the decision domain performs a security verification of the underlying computation results. The specific implementation steps are as follows: S601, the decision domain initiates a result query command in parallel to the main computation domain and the offset domain through an internal independent encrypted management link. Upon receiving the command, the main computation domain packages and uploads the calculated first part sum and the number of the first valid nodes to the decision domain. Simultaneously, upon receiving the command, the offset domain packages and uploads the calculated second part sum and the number of the second valid nodes to the decision domain. The decision domain temporarily stores these four parameters in its cache register, preparing to perform integrity verification.

[0107] S602, the security decision logic unit within the decision domain extracts the number of first and second valid nodes, and performs numerical consistency comparison and validity lower limit determination. Under normal system operation, since the main operation domain and the offset domain have already performed isolated message removal operations based on the valid intersection before the operation, the data samples actually participating in the modular algebraic accumulation calculation in the two domains have a one-to-one mapping relationship. The number of first valid nodes should be equal to the number of second valid nodes, and must be greater than the system's preset minimum valid participating node threshold (e.g., greater than zero, to prevent false zero-value judgments caused by network-wide communication interruption). The logical expression for the comparison process is as follows: ; In the formula, This represents the symmetric security decision state generated by the decision domain; Indicates the number of first valid nodes reported by the main computation domain; Indicates the number of second valid nodes reported by the offset domain; This represents the logical truth value indicating that the decision has passed. Indicates a logical false value indicating failure; Mathematical logic statements indicating that a condition is true.

[0108] S603: When the security decision status output is a logical false value, it indicates that the main operation domain or the offset domain encountered an illegal data injection, memory overwrite failure, or underlying timing collapse during the execution of the accumulation operation. This anomaly disrupts the data processing alignment mechanism between the two domains, and the decision domain immediately triggers the system's security circuit breaker mechanism.

[0109] Under the circuit breaker mechanism, the decision domain directly clears the first and second parts of its cache register, refusing to perform subsequent merge operations. For the system-level circuit breaker's post-circuit breaker exception logging and security alarm issuance mechanism, those skilled in the art can use the interface of a standard security information and event management system. The underlying exception event throwing mechanism is a well-known technology in this field and will not be elaborated upon here.

[0110] S604: When the security determination output is a logical true value, the decision domain confirms that the underlying isolated computing environment is not contaminated and that the data processing chain remains highly symmetrical. The decision domain removes the security interception restrictions for the current computing cycle, allowing the temporarily stored first and second parts to enter the final data restoration process. Implementing the above determination strategy enables macroscopic verification of the integrity of the distributed computing architecture without decrypting the physical runtime data.

[0111] After the symmetric blocking decision logic outputs a pass instruction, the decision domain utilizes the mathematical properties of modular arithmetic to merge the two unordered ciphertext accumulation results to remove interference data injected from the underlying layer. The specific implementation steps are as follows: S605: The decision domain retrieves the first and second partial sums temporarily stored in the cache register. The decision domain reads the homogeneous modulo constant from the system's global security configuration center. Within the logic unit, the decision domain performs an algebraic addition operation on the first and second partial sums, and then performs a modulo-reduction operation on the homogeneous modulo constant again using the result. The mathematical expression for the algebraic merging operation is as follows: ; In the formula, This represents the total effective power data of the virtual power plant calculated and output by the decision domain; This represents the first part of the sum reported by the main operation domain; This represents the second part of the sum reported by the offset field; This represents the number of valid nodes that are equal after consistency verification. Represents a globally uniformly defined isomorphic modulus constant; Represents the constant modulus of isomorphism Perform the remainder mathematical operation.

[0112] S606, an algebraic merging operation, can accurately restore the underlying logic of data, relying on the closed nature of addition operations within the homogeneous module group space. During the data splitting stage of edge device access, the real-time physical operation data acquired by the underlying hardware is evolved into the main data packet payload and the accompanying data packet payload through algebraic splitting logic.

[0113] By substituting its decomposition logic into the accumulation process of the main operational domain and the cancellation domain, a mathematical proof of the reconstruction deduction process can be obtained. The mathematical expression of the deduction process is as follows: ; ; ; ; ; ; In the formula, Indicates participation in the The main data packet payload is calculated in the next iteration; Indicates participation in the The accompanying data packet payload is calculated in the next iteration; Indicates participation in the The real-time physical operation data of the next iteration (i.e., the underlying physical plaintext corresponding to the effective data payload that successfully enters the operation queue after pre-alignment filtering). Indicates the relationship with the first The confusion factor of the corresponding isomorphic mode group is calculated in each iteration; Indicates from the first The number of valid nodes up to the first The continuous summation operation of each valid node.

[0114] After the above algebraic merging and modular recombination, the confusion factor of the isomorphic modular group is... It is eliminated in mathematical deduction in the form of positive and negative offsetting.

[0115] S607: To ensure that the output total effective power data of the virtual power plant reflects the actual parameters of the physical world, the system sets an overflow prevention boundary for the value range of the homogeneous modulus constant. The homogeneous modulus constant must be greater than the maximum threshold of the sum of real-time physical operation data output by all edge devices under the control of the virtual power plant.

[0116] This maximum threshold is determined by multiplying the total rated installed capacity of the registered virtual power plant assets by a set safety margin factor, which is set to range from 1.2 to 1.5. When the conditions are met... When the algebraic conditions are met, the modulo operation no longer truncates the actual sum. In this case, the derivation formula directly simplifies to: ; S608, after the complete offsetting of the above-mentioned obfuscation factors, the decision domain obtains the total effective power data of the virtual power plant of the entire computing cluster without ever touching the plaintext operation data of any single edge device throughout the entire process.

[0117] The decision domain writes the calculated total effective power data into a time-series database in the cloud for use by the upper-level power dispatching system for business invocation and peak-shaving decisions. The specific implementation of data persistence in the time-series database and data presentation on the front-end visual monitoring panel can be accomplished by those skilled in the art using standard structured query language commands and chart components. The underlying data persistence mechanism is a well-known technology in the field and will not be elaborated upon here. This deduction and reconstruction mechanism achieves compatibility between micro-level data isolation and shielding and precise calculation of macro-level data.

[0118] To aid in understanding the technical solution of this invention, a specific application embodiment is provided below to illustrate the content of this invention in detail. The embodiment describes a distributed virtual power plant connected to three edge devices (i.e., the total number of valid edge devices configured for concurrent access in the system). Each edge device registered asset has a rated installed capacity of 1000kW.

[0119] The decision domain receives the access handshake requests initiated by these three edge devices and performs certificate authentication. After successful authentication, the decision domain generates a dynamic access control token with timestamp boundaries and transmits it through secure communication data packets to the independent hardware security register of the trusted execution module of the corresponding edge device.

[0120] The system determines the homogeneous modulus constant by comprehensively considering the product of the maximum safety boundary value that the corresponding data type of the underlying physical sensor can accommodate and the maximum number of concurrent access nodes. The maximum security boundary value for a single node is set to the product of the rated installed capacity and the safety margin coefficient (e.g., 1.33), i.e., 1000 × 1.33 = 1330; the maximum number of concurrent access nodes is 3. The calculated maximum threshold is: 1330 × 3 = 3990. The system uses the homogeneous modular constant in the global security configuration center. The value is set to 4000. This setting ensures that the sum of the real-time physical operation data collected by all three concurrently accessing edge devices is less than the homogeneous modulus constant. Overflow prevention boundary conditions.

[0121] Suppose that at the current moment, the real-time physical operation data collected by the three edge devices are as follows: , , Each edge device's trusted execution module derives a homogeneous module group obfuscation factor based on parameters such as the fixed dynamic access control token, and sets them to... , , And calculate and generate their respective sequence identifiers.

[0122] According to the formula and The trusted execution module calculates the output load: Edge device 1: Main data packet payload Accompanying data packet payload ; Edge device 2: Main data packet payload Accompanying data packet payload ; Edge device 3: Main data packet payload Accompanying data packet payload .

[0123] After the split is completed, the trusted execution module uses the device private key to independently sign and encapsulate the separated payload and the corresponding sequence identifier, and outputs the first tamper-proof data packet and the second tamper-proof data packet.

[0124] Edge devices attach external routing signatures to their output tamper-proof data packets and send them to the routing gateway. The routing gateway performs distribution control, routing the main network transmission data packets with the attached primary routing signature to the main processing domain, and routing the accompanying network transmission data packets with the attached secondary routing signature to the offset domain.

[0125] Both the main operation field and the offset field perform public key validation on the received message. After successful validation, the main operation field extracts the sequence identifiers from the message to form a main sequence identifier set, and the offset field extracts the sequence identifiers to form a companion sequence identifier set.

[0126] The decision domain obtains the two sets mentioned above through the network and calculates their intersection to obtain the alignment identifier set. After receiving the alignment identifier set, the main operation domain and the cancellation domain use it to check and discard the data packet payloads corresponding to isolated messages whose sequence identifiers are not included in the intersection.

[0127] After the data alignment check is completed, the two domains perform modular algebraic accumulation calculations on the valid data in their respective memory spaces: The first part of the output from the main operand domain: Meanwhile, the number of the first valid nodes is recorded as 3.

[0128] The second part of the output of the cancellation domain: Meanwhile, the number of second valid nodes is recorded as 3.

[0129] After obtaining and comparing the number of the first and second valid nodes in the decision domain, and confirming that the values ​​are consistent (the judgment state is logically true), algebraic merging and modulo-based dimensionality reduction operations are performed on the sum of the first and second parts: ; Verification shows that the total effective power data of the virtual power plant calculated by the decision domain, 1900, is equivalent to the sum of the underlying real physical operation data (500+800+600), and the deduction and restoration mechanism logic is valid.

[0130] To verify the actual effect of the present invention, relevant experimental verification and effect comparison were carried out in conjunction with the above application embodiments.

[0131] See Figure 9The figure shows a comparison of the numerical distribution dimensions of the three edge devices connected to the system during the current scheduling period. The horizontal axis represents different edge devices, and the vertical axis represents the magnitude of the numerical distribution dimension. The figure shows that the real-time physical operation data collected by edge device 1, edge device 2, and edge device 3 are 500, 800, and 600, respectively.

[0132] After algebraic splitting calculation by the trusted execution module, the main data packet payload and accompanying data packet payload of edge device 1 are 1700 and 2800, respectively; the corresponding payloads of edge device 2 are 300 and 500, respectively; and the corresponding payloads of edge device 3 are 2700 and 1900, respectively. In terms of numerical distribution, the original real-time physical operation data characteristics are hidden, and the payload data after splitting into two paths exhibits an irregular discrete distribution trend.

[0133] The reason for the aforementioned irregular numerical trends is that the system derives a homogeneous modulus confusion factor using parameters such as dynamic access control tokens, and performs modulo-dimensionality reduction operations based on the homogeneous modulus constant. This processing mechanism changes the original linear mapping relationship of the data. Even if the data stream of a single network interface is intercepted, it is impossible to reconstruct the original physical operating characteristics through reverse calculation, thereby achieving data isolation during network transmission.

[0134] See Figure 10 The figure shows a comparison of the time consumption of algebraic merging operations under different concurrent access node scales. The horizontal axis of the figure represents the number of first or second effective nodes participating in the calculation, with values ​​ranging from 100 to 1000; the vertical axis represents the time consumption of the system's algebraic merging operations, in milliseconds.

[0135] As can be seen from the curve trend in the figure, the computation time of the traditional fully homomorphic ciphertext aggregation scheme increases non-linearly with the number of nodes, exceeding 1600 milliseconds when the number of nodes reaches 1000. In contrast, the computation time curve of the isomorphic modular algebraic reconstruction calculation of this invention has a lower slope, showing a generally linear and gentle trend, with the computation time remaining within the range of tens of milliseconds.

[0136] The reason for the aforementioned differences in processing time lies in the dimensionality reduction transformation mechanism employed in this invention for the algebraic space. When performing algebraic merging operations, the decision domain utilizes the closure property of addition operations within the homogeneous module group space to eliminate confusion factors of the homogeneous module groups through positive and negative offsetting. This restoration process does not employ large-number polynomial multiplication and private key decryption calculations, reducing the computational load on the underlying hardware memory management unit and cache registers, thus meeting the response time requirements of high-frequency power dispatching in power dispatching systems.

[0137] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can occur depending on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A virtual power plant data security protection system, characterized in that, include: The decision domain is used to send dynamic access control tokens to edge devices; An edge device is used to acquire real-time physical operation data, derive a homogeneous module group confusion factor using the dynamic access control token, and perform algebraic splitting calculation on the real-time physical operation data using the homogeneous module group confusion factor to generate the main data packet payload and the accompanying data packet payload. A routing gateway is used to route the main data packet payload to the main computing domain and the accompanying data packet payload to the offset domain; The main operation domain is used to perform modular algebraic accumulation calculation on the main data packet payload and output the first part of the sum; The cancellation domain is used to perform a modular algebraic accumulation calculation on the accompanying data packet payload and output the second part of the sum; The decision domain is also used to obtain the sum of the first part and the sum of the second part, perform algebraic merging and modulo dimensionality reduction operations on the sum of the first part and the sum of the second part, and output the total effective power data of the virtual power plant.

2. The virtual power plant data security protection system according to claim 1, characterized in that, The decision domain receives the access handshake request initiated by the edge device and performs certificate authentication. After successful authentication, it allocates access control parameters that include timestamp boundaries and permission matrices. The timestamp boundaries consist of an effective start time and an expiration end time. The decision domain concatenates the hardware device identifier, the effective start time, the expiration deadline, and the permission matrix to form a basic token payload. It then uses the decision domain's private key to perform a digital signature calculation on the basic token payload to generate a token digest. Finally, it merges the basic token payload and the token digest to form the dynamic access control token.

3. The virtual power plant data security protection system according to claim 2, characterized in that, The edge device has a built-in trusted execution module; The decision domain encapsulates the dynamic access control token into a secure communication data packet and inserts a hardware penetration identifier into the header of the secure communication data packet. After the edge device recognizes the hardware penetration identifier, it triggers a security monitor call instruction in the kernel layer of the edge device's built-in operating system to pass through the secure communication data packet to the physical isolation memory area dedicated to the trusted execution module. The trusted execution module extracts the dynamic access control token and stores it in an independent hardware security register.

4. The virtual power plant data security protection system according to claim 3, characterized in that, The edge device acquires the real-time physical operation data fed back by the underlying physical sensors through the underlying bus driver within the timestamp boundary; The trusted execution module extracts the dynamic access control token, the hardware device identifier, and the high-precision timestamp corresponding to the current business action, and concatenates them to construct the underlying obfuscation seed. It then uses an irreversible hash function to perform a hash operation on the underlying obfuscation seed and performs a modulo-dimension reduction operation on the resulting discrete hash number to derive the homogeneous modulus group obfuscation factor.

5. The virtual power plant data security protection system according to claim 4, characterized in that, The trusted execution module is used for: Extract the hardware device identifier and concatenate it with the high-precision timestamp, and then use the irreversible hash function to generate a sequence identifier; The real-time physical operation data is added to the homogeneous module group confusion factor, and the sum is moduloed using the homogeneous module constant to generate the main data packet payload. The accompanying data packet payload is generated by subtracting the homogeneous modulus confusion factor from the homogeneous modulus constant and performing a modulo operation on the difference result using the homogeneous modulus constant. The main data packet payload is concatenated with the corresponding sequence identifier to obtain the first concatenated data, and the accompanying data packet payload is concatenated with the corresponding sequence identifier to obtain the second concatenated data. The first concatenated data and the second concatenated data are signed using the device private key, and then merged and encapsulated to output the first tamper-proof data packet and the second tamper-proof data packet.

6. The virtual power plant data security protection system according to claim 5, characterized in that, The edge device inserts a main routing signature into the first tamper-proof data packet to form a network transmission main data packet, and inserts an accompanying routing signature into the second tamper-proof data packet to form a network transmission accompanying data packet; After receiving the main network transmission data packet, the routing gateway forwards the main network transmission data packet to the main computing domain. After receiving the accompanying network transmission data packet, it forwards the accompanying network transmission data packet to the offset domain.

7. The virtual power plant data security protection system according to claim 5, characterized in that, The main operation domain and the offset domain each perform public key validity verification; The main operation domain extracts the verification mark by aggregating the sequence marks in the message to construct a main sequence mark set, which is then uploaded separately to the decision domain. The offset domain extracts the verification mark by aggregating the sequence marks in the message to construct an accompanying sequence mark set, which is then uploaded to the decision domain. The decision domain performs a set intersection operation to extract common sequence identifier elements that exist simultaneously in the main sequence identifier set and the accompanying sequence identifier set, constructs an alignment identifier set, and sends the alignment identifier set back to the main operation domain and the cancellation domain.

8. A virtual power plant data security protection system according to claim 7, characterized in that, The main operation domain performs the modular algebraic summation calculation on the main data packet payload contained in the alignment identifier set, outputs the first partial sum, and counts the total number of sequence identifiers contained in the effective intersection as the number of first effective nodes.

9. A virtual power plant data security protection system according to claim 8, characterized in that, The offset field performs the modulo algebraic summation calculation on the accompanying data packet payload contained in the alignment identifier set, outputs the second part sum, and counts the total number of sequence identifiers contained therein as the second effective node number; Under the condition that the number of the first effective nodes is the same as the number of the second effective nodes and is greater than the minimum effective participating node threshold, the decision domain performs an algebraic addition operation on the first part and the second part, and then performs the modulo-dimension reduction operation on the isomorphic modulus constant to output the total effective power data of the virtual power plant.

10. A method for data security protection in a virtual power plant, characterized in that, The virtual power plant data security protection system according to any one of claims 1-9 includes the following steps: Send dynamic access control tokens to edge devices; Acquire real-time physical operation data, derive homogeneous module group obfuscation factor using the dynamic access control token, and perform algebraic splitting calculation on the real-time physical operation data using the homogeneous module group obfuscation factor to generate main data packet payload and accompanying data packet payload. The main data packet payload is routed and distributed to the main computation domain, and the accompanying data packet payload is routed and distributed to the offset domain; For the main data packet payload, perform modular algebraic accumulation calculation and output the first part sum; at the same time, perform modular algebraic accumulation calculation for the accompanying data packet payload and output the second part sum. Extract the first part and the second part, perform algebraic merging and modulo dimensionality reduction operations on the first part and the second part, and output the total effective power data of the virtual power plant.

Citation Information

Patent Citations

  • Virtual power plant data security protection method based on block chain

    CN116756772A

  • Data security protection system and method for virtual power plant

    CN118171293A

  • Virtual power plant data security protection method, platform, device and system based on cloud computing platform

    CN119299147B

  • Virtual power plant data security protection method and system and electronic equipment

    CN121637535A