A power grid dispatching-oriented multi-source heterogeneous data credible shunting fusion method and system

Abstract

This invention discloses a trusted data splitting and fusion method and system for power grid dispatching, belonging to the field of power system data processing and trusted computing technology. The method acquires multi-source heterogeneous power data and preprocesses it, performing lightweight encryption, data fingerprint calculation, and digital signature encapsulation to generate data record units. These data record units are mapped to blockchain-based evidence records. Trusted splitting judgment results are generated based on data fingerprint verification results, digital signature verification results, on-chain evidence record matching results, TEE remote measurement results, latency requirements, security levels, and data types. Real-time data undergoes trusted execution processing using the TEE path, while non-real-time data undergoes dense-state fusion calculation using the FHE path. Finally, a comprehensive fusion result is generated and output to the power grid dispatching system. Compared with existing technologies, this invention can balance the real-time performance of power grid dispatching, data privacy protection, and trusted traceability throughout the entire process.

Description

Technical Field

[0001] This invention belongs to the fields of power system data processing, trusted computing, data security and intelligent power grid dispatching technology, and in particular relates to a trusted offloading and fusion method and system for multi-source heterogeneous data for power grid dispatching. Background Technology

[0002] With the development of smart grids, the energy internet, new energy power plants, and distributed power sources, the power grid dispatching system needs to access various types of data from the generation side, new energy power plants, grid sensors, IoT terminals, and dispatching business systems. These data differ significantly in terms of acquisition frequency, data format, real-time performance, security level, reliable sources, and processing methods.

[0003] Existing power data processing solutions typically employ technologies such as encryption, signature, blockchain notarization, trusted execution environment, or homomorphic encryption in the acquisition, transmission, storage, or computing stages. However, these trusted technologies are often deployed in parallel, lacking a continuous trusted computing chain that runs through data record generation, on-chain binding, traffic splitting and determination, trusted execution, encrypted fusion, and result verification.

[0004] For power grid dispatching operations, real-time control data requires low latency and verifiable execution, while historical analysis, forecasting, or statistical data places greater emphasis on privacy protection and traceability. Applying fully homomorphic encryption to all data can easily lead to high computational overhead; while using plaintext or ordinary encryption for all data makes it difficult to guarantee the reliability and traceability of the processing.

[0005] Although current technologies for intelligent data acquisition, transmission, storage, and processing in power systems are constantly improving, the following shortcomings still exist:

[0006] 1) In the existing solutions, there is a lack of coordination among data fingerprints, digital signatures, on-chain evidence storage, TEE measurement and FHE encrypted computation, making it difficult to use trusted computation results to directly drive data diversion and fusion processing.

[0007] 2) Existing power data fusion processes often only focus on data format uniformity and fusion accuracy, without fully expressing the computational correlation between the reliability of data sources, the reliability of processing environments, the reliability of calculation results, and the traceability of logs.

[0008] 3) Fully homomorphic encryption computation has a large overhead. If it is directly used for all scheduling data, it will affect the real-time scheduling service response. Trusted execution environment is suitable for real-time processing, but remote measurement and result verification are required to ensure execution trustworthiness.

[0009] 4) Traditional transmission encryption and storage encryption can protect transmitted and static data, but it is difficult to form a verifiable and trustworthy closed loop for the entire process of collection, storage, distribution, fusion and output. Summary of the Invention

[0010] This invention aims to address the problems in existing power grid dispatch data processing schemes, such as fragmented expression of trusted technologies, insufficient basis for data diversion, difficulty in balancing real-time processing and dense-state computation, and difficulty in reliably tracing the processing process. Specifically, this invention constructs a trusted computing chain through data fingerprinting, digital signatures, on-chain notarization, TEE remote measurement, and FHE dense-state computation, and uses the verification results of this trusted computing chain to drive the diversion and fusion of multi-source heterogeneous power data.

[0011] The technical solution adopted in this invention is a reliable data offloading and fusion method for multi-source heterogeneous data oriented towards power grid dispatching, the method comprising the following steps:

[0012] S1. Acquire multi-source heterogeneous power data from the power generation side, new energy power stations, grid sensors and IoT terminals, and perform cleaning, noise reduction, standardization and time alignment processing on the multi-source heterogeneous power data.

[0013] S2 performs lightweight encryption and data fingerprint calculation on the preprocessed data, encapsulating data identifier, node identifier, timestamp, data type, latency level, security level, data fingerprint and digital signature into a data record unit;

[0014] S3, map the data recording unit to a blockchain evidence record, the blockchain evidence record includes a block header and transaction content, and establish the binding relationship between the data recording unit and the on-chain record through the current record hash, Merkle root and operation log summary;

[0015] S4 generates a trusted traffic splitting judgment result based on data fingerprint verification results, digital signature verification results, on-chain evidence record matching results, TEE remote measurement results, data latency requirements, security level and data type.

[0016] S5. When the trusted diversion judgment result indicates that the data to be processed is real-time control data or the latency requirement is lower than the preset latency threshold, the corresponding data is imported into the TEE real-time data path to perform remote measurement, secure loading, data decryption, real-time fusion calculation and result verification.

[0017] S6. When the trusted diversion judgment result indicates that the data to be processed is non-real-time analysis data, prediction data, or statistical data, homomorphic encryption encapsulation is performed on the corresponding data, and homomorphic operation, dense aggregation calculation, and result verification are performed in the dense state.

[0018] S7, through the fusion coordination unit, performs parameter synchronization, status synchronization, result alignment and reliability verification on the real-time results output by the TEE real-time data path and the analysis results output by the FHE non-real-time data path, and generates a comprehensive fusion result.

[0019] S8 associates and stores the integrated fusion results, processing logs, on-chain verification results, and output interface call records, and outputs the integrated fusion results to power grid dispatching, fault prediction, load balancing, power generation plan adjustment, or digital twin update applications.

[0020] Preferably, the lightweight encryption includes: generating a keystream Z based on the ZUC stream cipher using the key K and the initialization vector IV, and then... The ciphertext C is obtained, where M is the original plaintext data, and ⊕ represents the bitwise XOR operation.

[0021] Preferably, the data fingerprint is obtained through... The calculation yields the following: D represents the original data or preprocessed data, NodeID represents the data acquisition node identifier, T represents the timestamp, Type represents the data type, and || represents field concatenation.

[0022] Preferably, the hash of the current record in the blockchain evidence storage record is obtained through... Calculations show that the Merkle root is obtained through... The calculation shows that H_1 to H_n represent the hash values ​​of multiple data records within the same block.

[0023] Preferably, the reliable traffic splitting determination result R is obtained through... The values ​​are: Latency (representing latency requirement), SecLevel (representing security level), Type (representing data type), ChainState (representing on-chain evidence record matching status), SigState (representing digital signature verification status), and MeasureState (representing TEE remote measurement status).

[0024] Preferably, the TEE remote measurement includes hashing the program code, configuration parameters, and runtime image in the trusted execution environment to obtain... The MeasureValue is then compared with a preset trusted baseline value.

[0025] Preferably, the FHE encrypted state fusion calculation includes encrypting non-real-time data x_i into ciphertext Enc(x_i), and then... Execute the fusion function f in the dense state; when the fusion function is a weighted summation... , where w_i is the fusion weight corresponding to the i-th data source;

[0026] In a federated learning scenario, each terminal obtains the model gradient g_i based on local data and uploads the encrypted gradient Enc(g_i). The platform then... Perform dense gradient aggregation;

[0027] Before outputting the integrated fusion result, a record summary is calculated based on Result, ProcessLog, InterfaceLog, and ChainProof. The output record summary will be used for scheduling feedback, fault tracing, or accountability auditing.

[0028] A trusted offloading and fusion system for multi-source heterogeneous data in power grid dispatch, implementing the aforementioned method, includes: a multi-source data acquisition terminal for acquiring generation-side data, new energy power plant data, power grid sensor data, and IoT terminal data; a trusted transmission and storage layer for performing data encryption, secure transmission, data fingerprint generation, blockchain notarization, and encrypted storage; a trusted offloading and fusion layer for performing trusted offloading determination based on data fingerprint verification results, digital signature verification results, on-chain notarization record matching results, TEE remote measurement results, latency requirements, security levels, and data types, and allocating data to the TEE real-time path or the FHE non-real-time path; a trusted management layer for performing identity authentication, digital signature, log auditing, and traceability verification; and an application output layer for outputting the integrated fusion results to power grid dispatch, fault prediction, load balancing, and digital twin applications.

[0029] An electronic device includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the method.

[0030] A computer-readable storage medium storing a computer program that, when executed by a processor, implements the method.

[0031] Compared with existing technologies, this invention treats data fingerprint calculation, digital signature verification, on-chain evidence storage matching, TEE remote measurement, and FHE encrypted state calculation as a continuous trusted calculation process, so that trusted technologies are no longer just independent security modules, but directly participate in data diversion judgment and fusion processing.

[0032] This invention processes low-latency real-time data through the TEE path and processes non-real-time analysis data through the FHE path, thus ensuring real-time scheduling response capabilities while retaining dense state fusion and privacy protection capabilities.

[0033] This invention binds the data recording unit, blockchain evidence record, operation log summary, and trusted verification path together to make the data source, processing process, fusion result, and application output verifiable, traceable, and tamper-proof.

[0034] This invention is applicable to power business scenarios such as power grid dispatching, fault prediction, load balancing, power generation plan adjustment and digital twin status update, and can form a closed loop of operation monitoring and dispatch feedback. Attached Figure Description

[0035] Figure 1 The system overall architecture diagram provided for embodiments of the present invention;

[0036] Figure 2 A flowchart of trusted traffic splitting and fusion provided for embodiments of the present invention;

[0037] Figure 3 This is a diagram illustrating the data fingerprint and on-chain trusted record structure provided in an embodiment of the present invention.

[0038] Figure 4 This is a flowchart of the collaborative processing of TEE and FHE provided in an embodiment of the present invention;

[0039] Figure 5 This is a closed-loop diagram of application output and scheduling feedback provided in an embodiment of the present invention. Detailed Implementation

[0040] The specific embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that these embodiments are for explaining the present invention and not for limiting the scope of protection of the present invention. Those skilled in the art can substitute encryption algorithms, hash algorithms, blockchain platforms, trusted execution environments, or homomorphic encryption schemes without departing from the core ideas of the present invention.

[0041] Example 1: Overall System Architecture

[0042] like Figure 1 As shown, the system of this invention includes a multi-source data acquisition terminal, a trusted transmission and storage layer, a trusted load balancing and fusion layer, a trusted management layer, and an application output layer. The multi-source data acquisition terminal is used to acquire data from power generation, renewable energy plant data, power grid sensor data, and IoT terminal data; the trusted transmission and storage layer is used to perform data encryption, secure transmission, data fingerprint generation, blockchain notarization, and encrypted storage; the trusted load balancing and fusion layer is used to perform load balancing decisions based on trusted computation results and power grid business attributes; the trusted management layer is used for identity authentication, digital signatures, log auditing, and traceability verification; and the application output layer is used to serve power grid dispatching, fault prediction, load balancing, and digital twins.

[0043] Example 2: Trusted Flow Splitting and Fusion Method

[0044] like Figure 2 As shown, the system first receives multi-source heterogeneous power data and performs data cleaning, noise reduction, standardization, and time alignment. Subsequently, the system performs data fingerprint generation and on-chain verification on the preprocessed data, and uses the corresponding trusted state as input for the power diversion decision.

[0045] Lightweight encryption and data fingerprint generation at the data acquisition end

[0046] At the data acquisition end, for resource-constrained power generation-side acquisition terminals, grid sensors, and IoT terminals, ZUC stream cipher can be used to quickly encrypt the raw data. Given the original plaintext data M, key K, and initialization vector IV, the ZUC algorithm generates a keystream Z:

[0047]

[0048] Ciphertext C is obtained by bitwise XORing plaintext M with keystream Z:

[0049]

[0050] Here, ⊕ represents a bitwise XOR operation. The above encryption calculation is used to perform initial protection in low-power acquisition terminals.

[0051] For data record D, the SM3 hash algorithm is used to generate data fingerprint H:

[0052]

[0053] In this data fingerprint, NodeID represents the identifier of the data collection node, T represents the timestamp, Type represents the data type, and || represents field concatenation. This data fingerprint is used for on-chain notarization, integrity verification, and trusted traffic distribution determination.

[0054] Secure transmission, on-chain evidence storage and encrypted computation

[0055] During the transmission phase, the acquisition end and the platform end negotiate using a secure communication channel based on SM2 / SM4. Let the private keys of the two communicating parties be a and b, and the base point of the SM2 curve be G. Then the public keys of both parties can be represented as:

[0056]

[0057] The shared key material can be calculated using elliptic curve key exchange:

[0058]

[0059] During the blockchain notarization phase, the current record hash is calculated based on the block header and transaction content:

[0060]

[0061] The hash values ​​of multiple data records within a block further form the Merkle root:

[0062]

[0063] During the storage phase, disk encryption methods such as SM4-XTS can be used to protect the data written to disk, and storage keys can be derived based on passwords, salt values, and iteration parameters.

[0064]

[0065] Trustworthy shunting determination calculation

[0066] Trusted traffic splitting is not determined solely by data type, but rather by a comprehensive consideration of data trustworthiness and service latency requirements. The trusted traffic splitting result R is determined by the following factors:

[0067]

[0068] Among them, Latency represents latency requirements, SecLevel represents security level, Type represents data type, ChainState represents on-chain evidence record matching status, SigState represents digital signature verification status, and MeasureState represents TEE remote measurement status.

[0069] When ChainState or SigState is abnormal, the corresponding data does not directly enter the fusion computation, but instead enters the isolation, verification, or re-collection process; when the corresponding data is real-time control data and TEE measurement passes, it enters the TEE real-time path; when the corresponding data is historical analysis, prediction, or statistical data and on-chain verification passes, it enters the FHE encrypted path.

[0070] The aforementioned reliable data routing determination results are used to determine whether data enters the real-time data path or the non-real-time data path. The real-time data path mainly processes low-latency data such as control commands, real-time measurements, and fault alarms; the non-real-time data path mainly processes data such as historical statistics, trend forecasts, load analysis, and model training.

[0071] Example 3: Data fingerprinting and on-chain trusted record structure

[0072] like Figure 3 As shown, a data record unit includes a data ID, node ID, timestamp, data type, latency level, security level, data fingerprint, and digital signature. After field mapping and encapsulation, the data record unit generates a blockchain evidence record, which includes a block header and transaction content. The block header includes the previous block hash, the current record hash, the Merkle root, the timestamp, and the difficulty / random number. The transaction content includes the transaction index, the data record unit index, the operation type, and the operation log summary.

[0073] In the trusted verification path, the system performs on-chain queries according to data ID, time, or node, compares the local fingerprint with the on-chain fingerprint, and confirms the validity of the digital signature through public key verification; when the hash is consistent and the chain verification passes, the record is deemed trusted and valid.

[0074] Example 4: Co-processing of TEE and FHE

[0075] TEE Trusted Execution and FHE Dense-State Fusion Computation

[0076] In the TEE real-time data path, before performing data decryption and fusion computation, the system remotely measures the program code, configuration parameters, and running image in the trusted execution environment:

[0077]

[0078] When MeasureValue matches the preset trusted baseline value, data decryption and real-time fusion calculation are allowed to be performed in TEE; otherwise, execution is rejected and an audit log is recorded.

[0079] In the FHE non-real-time data path, the non-real-time data x_i is encrypted into ciphertext Enc(x_i), and the system executes the fusion function f in encrypted state:

[0080]

[0081] When the fusion function is a weighted sum, it can be expressed as:

[0082]

[0083] Here, w_i represents the fusion weight for each data source. This calculation process allows non-real-time analysis data to participate in aggregation calculations without exposing its plaintext.

[0084] In a federated learning scenario, each terminal obtains the model gradient g_i based on local data and uploads the encrypted gradient Enc(g_i). The platform then performs dense-state aggregation.

[0085]

[0086] The aggregation results, after being authorized and decrypted, are used to update the global model, thereby preventing the original data from each terminal from leaving the domain directly.

[0087] like Figure 4 As shown, the TEE real-time data processing link and the FHE non-real-time data processing link are synchronized for parameters, status, results, and reliability through a fusion coordination unit. Parameter synchronization includes synchronizing data time windows, data source identifiers, fusion weights, and business scenario parameters; status synchronization includes synchronizing data processing status, on-chain verification status, and execution environment status; result alignment includes correlating real-time results and analysis results according to timestamps, data types, and scheduling objects.

[0088] Example 5: Digital Signatures, Trusted Output, and Scheduling Feedback Closed Loop

[0089] Digital signatures, signature verification, and trusted output

[0090] The data source node uses digital signature algorithms such as SM2 to sign the data record unit. Let the private key be d, the message digest to be signed be e, and the signature value be (r, s). Then, the verification end verifies the signature based on the corresponding public key to confirm that the data source has not been forged.

[0091]

[0092] Before outputting the integrated results, the system will again correlate the integrated results, processing logs, interface call records, and on-chain verification results to form an output record summary:

[0093]

[0094] The output record summary can be used for subsequent scheduling feedback, fault tracing, and accountability auditing.

[0095] like Figure 5 As shown, the integrated results are input into the scheduling decision engine and used for grid scheduling, fault prediction, load balancing, power generation plan adjustment, and digital twin updates. After scheduling instructions are issued to grid operation objects such as transmission networks, substations, distribution networks, and power plants, the system continues to collect status data, operation monitoring, fault alarms, load changes, and new energy fluctuation information, and uses this feedback information as new multi-source heterogeneous data input, thereby forming a reliable distribution fusion and grid scheduling feedback closed loop.

[0096] Alternative implementation methods

[0097] Without affecting the technical effect of the present invention, the ZUC, SM2, SM3, SM4 and other algorithms can be replaced with cryptographic algorithms with equivalent security capabilities; the blockchain can be a consortium blockchain, a private blockchain or other trusted ledgers with tamper-proof recording capabilities; the TEE can be a trusted hardware or trusted software environment that supports remote measurement and isolated execution; and the FHE can be a homomorphic encryption scheme that supports addition, multiplication or approximate numerical calculations.

Claims

1. A multi-source heterogeneous data credible shunting fusion method for power grid dispatching, characterized in that, Includes the following steps: S1. Acquire multi-source heterogeneous power data from the power generation side, new energy power stations, grid sensors and IoT terminals, and perform cleaning, noise reduction, standardization and time alignment processing on the multi-source heterogeneous power data. S2 performs lightweight encryption and data fingerprint calculation on the preprocessed data, encapsulating data identifier, node identifier, timestamp, data type, latency level, security level, data fingerprint and digital signature into a data record unit; S3, map the data recording unit to a blockchain evidence record, the blockchain evidence record includes a block header and transaction content, and establish the binding relationship between the data recording unit and the on-chain record through the current record hash, Merkle root and operation log summary; S4 generates a trusted traffic splitting judgment result based on data fingerprint verification results, digital signature verification results, on-chain evidence record matching results, TEE remote measurement results, data latency requirements, security level and data type. S5. When the trusted diversion judgment result indicates that the data to be processed is real-time control data or the latency requirement is lower than the preset latency threshold, the corresponding data is imported into the TEE real-time data path to perform remote measurement, secure loading, data decryption, real-time fusion calculation and result verification. S6. When the trusted diversion judgment result indicates that the data to be processed is non-real-time analysis data, prediction data, or statistical data, homomorphic encryption encapsulation is performed on the corresponding data, and homomorphic operation, dense aggregation calculation, and result verification are performed in the dense state. S7, through the fusion coordination unit, performs parameter synchronization, status synchronization, result alignment and reliability verification on the real-time results output by the TEE real-time data path and the analysis results output by the FHE non-real-time data path, and generates a comprehensive fusion result. S8 associates and stores the integrated fusion results, processing logs, on-chain verification results, and output interface call records, and outputs the integrated fusion results to power grid dispatching, fault prediction, load balancing, power generation plan adjustment, or digital twin update applications.

2. The method according to claim 1, characterized in that, The lightweight encryption comprises: generating a key stream Z based on a ZUC sequence cipher according to a key K and an initial vector IV, and performing a bitwise exclusive OR operation on the key stream Z and the original plaintext data M to obtain ciphertext C, wherein the original plaintext data M is obtained by performing a bitwise exclusive OR operation on the plaintext data M and the initial vector IV. The ciphertext C is obtained, wherein M is the original plaintext data, and represents a bitwise exclusive OR operation.

3. The method according to claim 1, characterized in that, The data fingerprint is calculated by D || NodeID || T || Type, wherein D represents original data or preprocessed data, NodeID represents a collection node identifier, T represents a timestamp, Type represents a data type, and || represents field splicing.

4. The method according to claim 1, characterized in that, The current record hash in the blockchain record is obtained by calculation, and the Merkle root is obtained by calculation, wherein H_1 to H_n represent hash values of a plurality of data records in the same block.

5. The method according to claim 1, characterized in that, The trusted shunt determination result R is determined by determining, wherein Latency represents a latency requirement, SecLevel represents a security level, Type represents a data type, ChainState represents a chain on record matching state, SigState represents a digital signature verification state, and MeasureState represents a TEE distance measurement state.

6. The method according to claim 1, characterized in that, The TEE remote degree measurement includes hash calculation on program code, configuration parameters and running image in the trusted execution environment, to obtain and compares the MeasureValue with a preset trusted reference value.

7. The method according to claim 1, characterized in that, The FHE encrypted fusion calculation includes encrypting non-real-time data x_i into ciphertext Enc(x_i), and then... Execute the fusion function f in the dense state; when the fusion function is a weighted summation... , where w_i is the fusion weight corresponding to the i-th data source; In a federated learning scenario, each terminal obtains the model gradient g_i based on local data and uploads the encrypted gradient Enc(g_i). The platform then... Perform dense gradient aggregation; Before outputting the integrated fusion result, a record summary is calculated based on Result, ProcessLog, InterfaceLog, and ChainProof. The output record summary will be used for scheduling feedback, fault tracing, or accountability auditing.

8. A multi-source heterogeneous data trusted offloading and fusion system for power grid dispatching that implements the method described in any one of claims 1-7, characterized in that, include: Multi-source data acquisition terminal, used to acquire data from power generation, new energy power plant, power grid sensor, and IoT terminal; The Trusted Transmission and Storage Layer is used to perform data encryption, secure transmission, data fingerprint generation, blockchain notarization, and encrypted storage; the Trusted Distribution and Fusion Layer is used to perform trusted distribution judgment based on data fingerprint verification results, digital signature verification results, on-chain notarization record matching results, TEE remote measurement results, latency requirements, security level, and data type, and allocate data to the TEE real-time path or the FHE non-real-time path. Trusted management layer, used to perform identity authentication, digital signatures, log auditing, and source tracing verification; The application output layer is used to output the integrated results to power grid dispatching, fault prediction, load balancing and digital twin applications.

9. An electronic device, characterized in that, The method includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method according to any one of claims 1 to 7.

Images