Carbon chain grid-oriented data privacy protection method and system

Abstract

The invention discloses a data privacy protection method and system for a carbon chain grid, and relates to the field of data privacy transmission and privacy computation.The data privacy protection method comprises the steps that a user needing to conduct a computing task initiates the computing task to a central server, and the computing task is sent to the central server; the central server initiates a calculation request to each data owner user involved in the calculation task, corresponding federated learning calculation iteration is carried out, and federated learning backdoor defense based on interpretability indexes is carried out in the iteration process of each step; after calculation is finished, the central server sends a final federal learning model to the task initiator; and according to the obtained global and local federated learning models, calculation of interpretability indexes is carried out, different learning rates are given to each local federated learning model according to the interpretability indexes, and model aggregation of the global model is carried out. According to the method, a federal backdoor defense algorithm based on interpretability is utilized, potential backdoor models can be filtered, and normal circulation and utilization of carbon chain grid data are guaranteed.

Description

Technical Field

[0001] The present invention relates to the fields of data privacy transmission and privacy computing, and in particular to a data privacy protection method and system for a carbon chain grid. Background Art

[0002] Some electricity data has low sensitivity and rich shared value, requiring no encrypted storage. To ensure the proper circulation of this data while maintaining privacy and security, research on off-chain data transmission and sharing methods is imperative. Federated learning is a new paradigm in machine learning that aims to enable model training and knowledge sharing across decentralized devices or data sources while protecting user privacy. Unlike centralized machine learning, federated learning shifts model training from data centers to local devices. The decentralized and privacy-preserving nature of federated learning (FL) makes it vulnerable to backdoor attacks, which aim to manipulate the final model's behavior based on attacker-selected inputs. The security and privacy concerns of FL have hindered its community trust and rapid adoption. The ubiquity of backdoor attacks stems from their ability to occur at any stage of the FL model: client-side training, pre-aggregation, during aggregation, and post-aggregation. Current research on backdoor attacks against FL models (horizontal FL) primarily focuses on image recognition and text recognition. Backdoor attack methods can be categorized based on different attack hypotheses.

[0003] As the power industry actively strives for clean, low-carbon development, blockchain, with its distributed ledger and immutable nature, is bringing profound and disruptive changes to the traditional power system, bringing new possibilities, new opportunities, and new developments to the power industry. From a long-term perspective, the introduction of blockchain technology is not only an inevitable choice for addressing climate change, but some power data still has low sensitivity and rich shared value, which does not require encrypted storage. To ensure the normal circulation of this power data while maintaining privacy and security, research on off-chain data transmission and sharing methods is imperative.

[0004] Aiming to "inject new vitality into the electricity and carbon market," and fully integrating the green requirements of economic development, we will focus on researching off-chain data transmission and sharing technologies based on federated learning. We will establish a data privacy trusted computing engine to enable trusted computing and cross-chain sharing of table, image, and graph data. Furthermore, to ensure the security and privacy of data sharing links, we will establish a systematic federated learning backdoor security defense system and framework. Summary of the Invention

[0005] In view of the above-mentioned problems, the present invention is proposed.

[0006] Therefore, the problem to be solved by the present invention is that the decentralized and privacy-preserving characteristics of federated learning (FL) make it vulnerable to backdoor attacks. In order to allow these power data to circulate normally while maintaining privacy and security, research on off-chain data transmission and sharing methods is imperative.

[0007] To solve the above technical problems, the present invention provides the following technical solutions: a data privacy protection method for a carbon chain grid, comprising: a carbon chain grid user who needs to apply for data initiates a carbon chain grid data application request to a central server, the data owner user receives the data application request forwarded from the central server, and decides whether to authorize based on the identity of the requester; a user who needs to perform a computing task initiates a computing task to the central server, selects the data for the computing task based on the data required for the computing task, and also sends the selection status and task ID to the central server; the central server queries the blockchain for the data selected by the computing task initiator to verify the data ownership; the central server initiates a computing request to each data owner user involved in the computing task, performs corresponding federated learning computing iterations, and performs federated learning backdoor defense based on interpretability indicators during each iterative process; after the calculation is completed, the central server sends the final federated learning model to the task initiator, and uploads the hash value of the model to the blockchain for evidence storage; based on the obtained global and local federated learning models, the interpretability index is calculated, different learning rates are assigned to each local federated learning model based on the interpretability index, and model aggregation of the global model is performed.

[0008] As a preferred solution of the carbon chain grid-oriented data privacy protection method described in the present invention, the decision of whether to authorize based on the identity of the requester includes: the data owner user responds to the central server with the data authorization result. If the data authorization result is negative, it directly returns to the central server that no data authorization is performed; if the central server receives a positive authorization result from the data owner, it combines the data requester's ID, the data initiator's ID and the data hash value into a record and uploads them to the blockchain for recording.

[0009] As a preferred solution of the data privacy protection method for the carbon chain grid described in the present invention, the computing tasks include classification and prediction tasks, and the user who initiates the computing task selects the data by selecting the corresponding data hash from the list of authorized data, and performs different categories of computing tasks according to different authorized data; after selecting the computing task and the computing data, the user who initiates the computing task sends the name of the computing task, the selected computing data and its data owner to the central server.

[0010] As a preferred solution of the data privacy protection method for the carbon chain grid described in the present invention, the central server queries the blockchain for the data selected by the computing task initiator, and performs data ownership verification, including querying the blockchain with the computing task initiator as the data requester and the data record based on the data summary and data owner provided by the computing task initiator. If the result returned by the blockchain indicates that there is a data record related to the requester, the data ownership verification is successful and the computing task is carried out normally, otherwise the data ownership verification fails; after the central server fails to perform data ownership verification on the blockchain, it sends information rejecting the computing request to the computing task initiator.

[0011] As a preferred solution of the data privacy protection method for the carbon chain grid described in the present invention, the corresponding federated learning calculation iteration includes: selecting different federated learning algorithm models based on the selected data type, according to whether the data belongs to tabular data, image data or graph data. If it is tabular data, a federated learning algorithm based on a tree model is selected; if it is image and graph data, a federated learning algorithm based on federated average is selected; according to the data feature distribution of each data owner, a horizontal federated learning or a vertical federated learning paradigm is selected. If horizontal federated learning is selected, alignment is performed on the feature dimension of the data; if vertical federated learning is selected, alignment is performed on the sample dimension.

[0012] As a preferred solution of the data privacy protection method for the carbon chain grid described in the present invention, after the central server sends the final federated learning model to the task initiator, the computing task initiator receives the final federated learning model, calculates the hash value of the model, and queries the current model hash value on the blockchain. If the query result shows that the current hash value record exists, the computing task is successful; otherwise, it indicates that there is a transmission error or potential malicious tampering with the model.

[0013] As a preferred solution of the data privacy protection method for the carbon chain grid described in the present invention, the calculation of the explainability index includes calculating the explainability index from three dimensions: attribute index, neuron index and network layer index for each local model; in terms of attribute index, adversarial samples are generated and input to the local model, and the output difference of each local model under the input of the adversarial sample and the original data set sample is recorded; in terms of neuron index, the neuron output distance and the neuron activation distance are calculated, and the output distance is calculated by calculating the output value distance of the output layer neurons of any two models given the same test set, and the activation distance is calculated by calculating the difference in the activation values ​​of the two models; in terms of network layer index, the L2 norm representing the distance between the output values ​​of a certain layer between models under a given input set and the distance of the output value probability distribution of a certain layer are calculated respectively.

[0014] In order to solve the above technical problems, the present invention provides the following technical solutions: a system for a data privacy protection method for a carbon chain grid, comprising: an application and authorization module, a federated learning algorithm module and a backdoor defense algorithm module; the application and authorization module is used to manage and supervise whether data requests are accepted, specifically including: a carbon chain grid user who needs to apply for data initiates a carbon chain grid data application request to a central server, and the data owner user receives the data application request forwarded from the central server and decides whether to authorize based on the identity of the requester; the federated learning algorithm module is used to calculate an iterative federated learning model, specifically including: a user who needs to perform a computing task initiates a computing task to the central server, selects the data for the computing task based on the data required for the computing task, and The selection status and task ID are also sent to the central server; the central server queries the blockchain for the data selected by the initiator of the computing task to verify the data ownership; the central server initiates a computing request to each data owner user involved in the computing task, performs corresponding federated learning computing iterations, and performs federated learning backdoor defense based on interpretability indicators during each iteration; after the calculation is completed, the central server sends the final federated learning model to the task initiator and uploads the hash value of the model to the blockchain for evidence storage; the backdoor defense algorithm module calculates the interpretability indicator based on the obtained global and local federated learning models, assigns different learning rates to each local federated learning model based on the interpretability indicator, and performs model aggregation of the global model.

[0015] A computer device includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of the above-mentioned data privacy protection method for the carbon chain grid are implemented.

[0016] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the above-mentioned data privacy protection method for a carbon chain grid.

[0017] The beneficial effects of the present invention are as follows: the method of the present invention is aimed at protecting the privacy of carbon chain grid data, and combined with the blockchain, it ensures the off-chain use and sharing of carbon chain data.

[0018] The present invention utilizes a federal backdoor defense algorithm based on explainability, which can filter out potential backdoor models and ensure the normal circulation and utilization of carbon chain grid data.

[0019] The method of the present invention addresses the backdoor security issue in federated learning. It can judge, detect and remove backdoors without assuming the attack strategy of the backdoor attacker in federated learning, effectively protecting the security and privacy of federated learning user participants and service providers. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0021] Figure 1 This is a flow chart of a data privacy protection method for the carbon chain grid in Example 1.

[0022] Figure 2 This is a request and authorization flow chart of a data privacy protection method for the carbon chain grid in Example 1.

[0023] Figure 3 This is a flowchart of the calculation tasks and calculation data selection of a data privacy protection method for a carbon chain grid in Example 1.

[0024] Figure 4 This is a flowchart of data rights confirmation and verification of a data privacy protection method for the carbon chain grid in Example 1.

[0025] Figure 5 This is a flowchart of the federated learning task of a data privacy protection method for the carbon chain grid in Example 1. DETAILED DESCRIPTION

[0026] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0027] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0028] Example 1, reference Figure 1-Figure 5 , which is the first embodiment of the present invention, provides a data privacy protection method for a carbon chain grid, including: Figure 1 As shown:

[0029] Step S1: A CarbonChain Grid user who needs to apply for data initiates a CarbonChain Grid data application request and sends the request to a central server.

[0030] Step S2: The data owner user receives the data application request forwarded from the central server and authorizes the data according to the data owner user's wishes.

[0031] Step S3: The user who needs to perform a computing task initiates the computing task to the central server as needed and generates a task ID. The user selects the data for the computing task based on the data required for the computing task and also sends the selection and task ID to the central server.

[0032] Step S4: The central server queries the blockchain for the data selected by the computing task initiator to verify data ownership.

[0033] Step S5: The central server initiates a computing request to each data owner user involved in the computing task, performs corresponding federated learning computing iterations, and performs federated learning backdoor defense based on explainability indicators during each iteration.

[0034] Step S6: After the calculation is completed, the central server sends the final federated learning model to the task initiator and uploads the hash value of the model to the blockchain for evidence storage.

[0035] Step S7: Calculate the interpretability index based on the global and local federated learning models obtained in S5.

[0036] Step S8: Assign different learning rates to each local federated learning model based on the interpretability index, and perform model aggregation of the global model.

[0037] Reference Figure 2 As shown, step S2 specifically includes: step S201: the data owner user responds to the central server with a data authorization result. If the data authorization result is no, it directly returns to the central server that no data authorization is performed. If it is yes, it proceeds to step S202.

[0038] Step S202: The central server receives a valid authorization request from the data owner, and combines the data requester's ID, the data initiator's ID, and the data hash value into a record, which is then uploaded to the blockchain for recording.

[0039] Reference Figure 3 As shown, step S3 specifically includes, step S301: optional computing tasks include classification and prediction tasks, and the user who initiates the computing task selects data by selecting the corresponding data hash from the list of authorized data. Different authorized data can perform different categories of computing tasks.

[0040] Step S302: After selecting the computing task and computing data, the user who initiates the computing task will send the name of the computing task and the selected computing data and its data owner to the central server.

[0041] Reference Figure 4As shown, step S4 specifically includes step S401: according to the data summary and data owner provided by the computing task initiator, the computing task initiator is taken as the data requester and the data record is queried in the blockchain. If the result returned by the blockchain indicates that the data record exists, the data ownership verification is successful and the computing task is carried out normally. Otherwise, the data ownership verification fails and the process goes to step S402.

[0042] Step S402: After the central server fails to verify the data ownership in the blockchain, it sends a message rejecting the computing request to the initiator of the computing task.

[0043] Reference Figure 5 As shown, step S5 specifically includes: step S501: based on the selected data type, different federated learning algorithm models are selected according to whether the data is tabular data, image data or graph data. If it is tabular data, a federated learning algorithm based on a tree model is selected; for image and graph data, a federated learning algorithm based on a federated average is selected.

[0044] Step S502: Select horizontal federated learning or vertical federated learning paradigm based on the data feature distribution of each data owner. If horizontal federated learning is selected, alignment is performed on the feature dimension of the data; if vertical federated learning is selected, alignment is performed on the sample dimension.

[0045] Step S6 specifically includes: Step S601: the central server sends the final federated learning model to the computing task initiator, calculates the hash value of the model, and uploads the computing task initiator, task ID and the hash value of the model as a record to the blockchain.

[0046] Step S602: After receiving the final federated learning model, the computing task initiator calculates the hash value of the model and queries the hash value on the blockchain. If the query result shows that the record exists, the computing task is successful. Otherwise, it indicates that the model may have a transmission error or potential malicious tampering.

[0047] Step S7 specifically includes: Step S701: Calculate interpretability metrics for each local model based on three dimensions: attribute metrics, neuron metrics, and network layer metrics. For attribute metrics, generate adversarial examples as input to the local model and record the difference in output between each local model and the original dataset sample.

[0048] Step S702: Based on neuron indicators, the neuron output distance and neuron activation distance are calculated. The former is to calculate the numerical output distance of the output layer neurons of the two models given the same test set, and the latter is to calculate the difference in their activation values ​​(which can only be 0 or 1).

[0049] Step S703: In terms of network layer indicators, the L2 norm representing the distance between the output values ​​of a certain layer between models under a given input set and the distance between the probability distributions of the output values ​​of a certain layer (measured using JS divergence) are calculated respectively.

[0050] Step S8 specifically includes calculating the variance of each indicator based on the interpretability index calculated by each model and normalizing the variance. If the global maximum learning rate is η max , the normalized variance is s, the scaling factor is β, then the learning rate η of the i-th local model i Expressed as,

[0051]

[0052] Among them, s avg Expressed as mean square error.

[0053] Example 2 is the second embodiment of the present invention, which is different from the first embodiment in that: a data privacy protection system for the carbon chain grid, including an application and authorization module, a federated learning algorithm module and a backdoor defense algorithm module; the application and authorization module is used to manage and supervise whether the data request is accepted, specifically including: a carbon chain grid user who needs to apply for data initiates a carbon chain grid data application request to the central server, the data owner user receives the data application request forwarded from the central server, and decides whether to authorize based on the identity of the requester; the federated learning algorithm module is used to calculate the iterative federated learning model, specifically including: a user who needs to perform a computing task initiates a computing task to the central server, and the data for the computing task is selected based on the data required by the computing task. The selection status and task ID are also sent to the central server; the central server queries the blockchain for the data selected by the initiator of the computing task to verify the data ownership; the central server initiates computing requests to the data owners involved in the computing task, performs corresponding federated learning computing iterations, and performs federated learning backdoor defense based on interpretability indicators during each iteration; after the calculation is completed, the central server sends the final federated learning model to the task initiator and uploads the hash value of the model to the blockchain for evidence storage; the backdoor defense algorithm module calculates the interpretability indicator based on the obtained global and local federated learning models, assigns different learning rates to each local federated learning model based on the interpretability indicator, and performs model aggregation of the global model.

[0054] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0055] The logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing the logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (e.g., a computer-based system, a system including a processor, or other system that can fetch and execute instructions from an instruction execution system, apparatus, or device). For purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0056] More specific examples (a non-exhaustive list) of computer-readable media include the following: an electrical connection with one or more wires (electronic devices), a portable computer disk cartridge (magnetic devices), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disc read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering, or processing in another suitable manner as necessary, and then stored in a computer memory.

[0057] It should be understood that various parts of the present invention can be implemented using hardware, software, firmware, or a combination thereof. In the above-described embodiments, multiple steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, it can be implemented using a combination of any of the following technologies known in the art: a discrete logic circuit having logic gate circuits for implementing logic functions on data signals, an application-specific integrated circuit having suitable combinational logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0058] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A data privacy protection method for the carbon chain grid, characterized by: include, The CarbonChain Grid user who needs to apply for data initiates a CarbonChain Grid data application request to the central server. The data owner receives the data application request forwarded by the central server and decides whether to authorize it based on the identity of the requester. The user who needs to perform a computing task initiates the computing task to the central server, selects the data for the computing task based on the data required by the computing task, and sends the selection and task ID to the central server; The central server queries the blockchain for the data selected by the computing task initiator to verify data ownership; The central server initiates computing requests to each data owner involved in the computing task, performs corresponding federated learning computing iterations, and performs federated learning backdoor defense based on explainability indicators during each iteration. After the calculation is completed, the central server sends the final federated learning model to the task initiator and uploads the hash value of the model to the blockchain for storage; Based on the obtained global and local federated learning models, the interpretability index is calculated, a different learning rate is assigned to each local federated learning model according to the interpretability index, and model aggregation of the global model is performed.

2. The data privacy protection method for the carbon chain grid according to claim 1, characterized in that: The decision of whether to authorize based on the identity of the requester includes: the data owner user responding to the central server with a data authorization result; if the data authorization result is negative, the central server is directly notified of the data authorization; If the central server receives a yes authorization result from the data owner, it will compile a record of the data requester's ID, the data initiator's ID and the data hash value and upload them to the blockchain for recording.

3. The data privacy protection method for the carbon chain grid according to claim 2, characterized in that: The computing tasks include classification and prediction tasks. The user who initiates the computing task selects the data by selecting the corresponding data hash from the list of authorized data, and performs different types of computing tasks based on the authorized data. After selecting the computing task and computing data, the user who initiates the computing task sends the name of the computing task and the selected computing data and its owner to the central server.

4. The data privacy protection method for the carbon chain grid according to claim 3, characterized in that: The central server queries the blockchain for the data selected by the computing task initiator to verify the data ownership, including: Based on the data summary and data owner provided by the computing task initiator, the computing task initiator is queried on the blockchain as the data requester and the data record. If the result returned by the blockchain indicates that the data record related to the requester exists, the data ownership verification is successful and the computing task is carried out normally. Otherwise, the data ownership verification fails. After the central server fails to verify the data ownership in the blockchain, it sends a message to the initiator of the computing task to reject the computing request.

5. The data privacy protection method for the carbon chain grid according to claim 4, characterized in that: The performing of the corresponding federated learning computation iteration includes selecting different federated learning algorithm models based on the selected data type, depending on whether the data is tabular data, image data, or graph data. If the data is tabular data, a federated learning algorithm based on a tree model is selected; if the data is image and graph data, a federated learning algorithm based on federated averaging is selected. Based on the data feature distribution of each data owner, choose the horizontal federated learning or vertical federated learning paradigm. If horizontal federated learning is selected, alignment is performed on the feature dimension of the data; if vertical federated learning is selected, alignment is performed on the sample dimension.

6. The data privacy protection method for the carbon chain grid according to claim 5, characterized in that: After the central server sends the final federated learning model to the task initiator, the computing task initiator receives the final federated learning model, calculates the hash value of the model, and queries the current model hash value on the blockchain. If the query result shows that the current hash value record exists, the computing task is successful. Otherwise, it indicates that there is a transmission error or potential malicious tampering with the model.

7. The data privacy protection method for the carbon chain grid according to claim 6, characterized in that: The calculation of the interpretability index includes calculating the interpretability index from three dimensions: attribute index, neuron index and network layer index, taking each local model as a unit; In terms of attribute indicators, we generate adversarial samples and input them to local models, and record the output difference between each local model when the adversarial sample is input and the original dataset sample; In terms of neuron indicators, the neuron output distance and neuron activation distance are calculated. The output distance is calculated by calculating the distance between the output layer neuron values ​​of any two models given the same test set, and the activation distance is calculated by calculating the difference between the activation values ​​of the two models. In terms of network layer indicators, the L2 norm representing the distance between the output values ​​of a certain layer between models under a given input set and the distance between the probability distributions of the output values ​​of a certain layer are calculated respectively.

8. A data privacy protection system for a carbon chain grid, applying the data privacy protection method for a carbon chain grid as claimed in any one of claims 1 to 7, characterized in that: Including application and authorization module, federated learning algorithm module and backdoor defense algorithm module; The application and authorization module is used to manage and supervise whether data requests are accepted. Specifically, the Carbon Chain Grid user who needs to apply for data initiates a Carbon Chain Grid data application request to the central server. The data owner user receives the data application request forwarded by the central server and decides whether to authorize it based on the identity of the requester. The federated learning algorithm module is used to calculate the iterative federated learning model. Specifically, the user who needs to perform a computing task initiates the computing task to the central server. Based on the data required for the computing task, the user selects the data for the computing task and sends the selection and task ID to the central server. The central server queries the blockchain for the data selected by the computing task initiator to verify the data ownership. The central server initiates computing requests to each data owner involved in the computing task and performs corresponding federated learning computing iterations. In each iterative process, the user performs federated learning backdoor defense based on interpretability indicators. After the calculation is completed, the central server sends the final federated learning model to the task initiator and uploads the hash value of the model to the blockchain for storage; The backdoor defense algorithm module calculates the interpretability index based on the obtained global and local federated learning models, assigns different learning rates to each local federated learning model according to the interpretability index, and performs model aggregation of the global model.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of a data privacy protection method for a carbon chain grid according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of a data privacy protection method for a carbon chain grid according to any one of claims 1 to 7 are implemented.

Images