Multi-source heterogeneous power data processing method and system

Abstract

The invention discloses a multi-source heterogeneous power data processing method and system. The method comprises the following steps: encrypting original data of power consumers stored in a scattered manner to obtain initial encrypted data; removing non-power associated data in the initial encrypted data to obtain a distributed data set, dividing the distributed data set into data subsets, and distributing each data subset to a corresponding edge node; inputting the data subsets into the constructed local model to obtain privacy data in each edge node; adjusting an encryption scheme for each piece of privacy data according to the heterogeneity of the target power grid; and in response to the power data acquisition instruction, executing a decryption scheme adaptive to the encryption scheme and outputting corresponding privacy data when a security condition is satisfied. According to the multi-source heterogeneous power data processing method and system provided by the invention, the calling effect of the power data can be improved, the confidentiality requirement of the power data is met through a related encryption algorithm, and the intelligent process of processing the multi-source heterogeneous power data is promoted.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a multi-source heterogeneous power data processing method and system. Background Art

[0002] With the rapid development of smart grids and energy Internet, the scale of data related to the power system has grown exponentially, covering multi-dimensional information such as power generation, transmission, power consumption and equipment monitoring.

[0003] In the existing technology, the data related to the power system is simply stored in a fixed location, such as in the core computing node, without any other operations. However, on the one hand, during peak hours, high-load nodes need to handle real-time monitoring, user requests and data analysis tasks at the same time, resulting in an increase of 2-3 orders of magnitude in storage access latency. Experimental data shows that when the node CPU usage rate exceeds 85%, the data call failure rate can reach 34.7%, which is not conducive to the call of power data; on the other hand, power data involves sensitive information, such as user power consumption behavior, equipment operating status, etc. The existing technology will cause serious privacy leakage problems, which is not conducive to the safe and effective management of power data. Summary of the invention

[0004] The present invention provides a multi-source heterogeneous power data processing method and system. By designing a perfect multi-source heterogeneous power data processing flow, the scattered data is managed in a refined manner and sent to the corresponding edge nodes for storage. This can not only improve the calling effect of the power data, but also meet the confidentiality requirements of the power data through relevant encryption algorithms, thereby promoting the intelligent process of multi-source heterogeneous power data processing.

[0005] In order to solve the above technical problems, an embodiment of the present invention provides a multi-source heterogeneous power data processing method, including: Encrypting the original data of each decentralized power user in the target power grid to obtain initial encrypted data; Removing non-power-related data from the initial encrypted data to obtain a distributed data set; Dividing the distributed data set into a number of data subsets, and allocating each of the data subsets to a corresponding edge node, wherein the edge node is determined by topological node information and communication delay matrix information of the target power grid; Obtaining private data in each edge node, wherein the private data is obtained by inputting the corresponding data subset into a constructed local model, and the local model is designed to dynamically adjust the introduced noise intensity through a privacy budget allocation mechanism; adjusting an encryption scheme for each of the private data according to the heterogeneity of the target power grid; In response to the power data acquisition instruction, a decryption scheme adapted to the encryption scheme is executed under a security condition and the corresponding privacy data is output.

[0006] As one of the preferred solutions, the method of encrypting the original data of each decentralized power user in the target power grid to obtain the initial encrypted data includes: Extracting the original data of each decentralized power user and verifying the data source field in the original data of the power user; The identifier is obtained by reading the corresponding access rule from the preset access policy file, and the encryption key matching the verified original data of the power user is generated by using the RSA algorithm; Encrypting the verified original data of the power user based on the AES encryption algorithm and the encryption key to obtain a data block; The identifier and the data block are associated to obtain the initial encrypted data.

[0007] As one of the preferred solutions, the non-power-related data at least includes user identity information; The removing of non-power-related data from the initial encrypted data to obtain a distributed data set includes: Using a K-means clustering algorithm to remove the user identity information in the initial encrypted data, at least obtaining power consumption data, load rate data, and peak-to-valley difference data; Based on the 3σ criterion, abnormal data in the power consumption data, the load rate data and the peak-to-valley difference data are eliminated in sequence; The distributed data set is constructed based on the elimination results.

[0008] As one preferred solution, before dividing the distributed data set into a plurality of data subsets, the method further includes: Acquire physical topology data from a power system database of the target power grid, wherein the physical topology data at least includes node data and connection relationship data; Performing a delay test on the communication network of the target power grid, and constructing delay matrix information based on the obtained communication overhead quantification results; The physical topology data and the delay matrix information are processed based on a node selection algorithm to obtain each edge node.

[0009] As one preferred solution, obtaining the privacy data in each edge node includes: Based on the noise scale parameter determined by the preset privacy budget value and the local sample data, the perturbation data satisfying differential privacy is generated; Using the disturbance data to train an initial local model; Using a differential privacy verification tool to verify the initial model parameters of the initial local model, and dynamically adjusting the initial model parameters based on the verification result to obtain a local model that meets the preset requirements; The private data in each edge node is obtained by using the local model.

[0010] As one of the preferred solutions, the encryption scheme for each of the private data is adjusted according to the heterogeneity of the target power grid, including: Inputting the model parameters corresponding to the local model obtained from each edge node into the Paillier homomorphic encryption algorithm to obtain the encryption model parameters of the current edge node; Designing an adaptive parameter aggregation strategy according to the hierarchical information of the edge node, and at least performing parameter compression on the encryption model parameters; The compression result is processed using the Paillier homomorphic encryption algorithm to generate an encryption scheme for each of the private data.

[0011] As one of the preferred solutions, the types of the privacy data include at least control instruction data, device status data and monitoring log data; the encryption scheme includes at least a first encryption scheme, a second encryption scheme and a third encryption scheme; The first encryption scheme includes the use of national secret SM4 / AES-256 encryption and digital signature based on elliptic curve; the second encryption scheme includes the use of lightweight stream encryption; the third encryption scheme includes embedding key derivation parameters in the log file header.

[0012] As one of the preferred solutions, blockchain technology is used to generate an audit log around the power data acquisition instruction; Processing the operation records in the audit log using the SHA256 hash algorithm to generate and store a corresponding unique identifier; According to the unique identifier, determining whether the data acquisition process complies with preset rules, wherein the preset rules at least include data access rights and calculation compliance; If the preset rule is met, it is determined that the safety condition is met.

[0013] As one preferred solution, the method further comprises: If the security condition is not met, the KNN algorithm is used to analyze the abnormal operation behavior in the corresponding audit log and generate a suspicious behavior report; The data to be isolated is determined according to the suspicious behavior report, and the data to be isolated is encrypted twice.

[0014] Another embodiment of the present invention provides a multi-source heterogeneous power data processing system, including: An encryption module, used for encrypting the original data of each decentralized power user in the target power grid to obtain initial encrypted data; A removal module, used for removing non-power-related data from the initial encrypted data to obtain a distributed data set; A partitioning module, used to divide the distributed data set into a number of data subsets, and assign each of the data subsets to a corresponding edge node, wherein the edge node is determined by the topological node information and communication delay matrix information of the target power grid; A privacy module, configured to obtain private data in each edge node, wherein the private data is obtained by inputting the corresponding data subset into a constructed local model, and the local model is designed to dynamically adjust the introduced noise intensity through a privacy budget allocation mechanism; An encryption scheme module, configured to adjust the encryption scheme for each of the private data according to the heterogeneity of the target power grid; The output module is used to respond to the power data acquisition instruction, execute a decryption scheme adapted to the encryption scheme under the security condition and output the corresponding privacy data.

[0015] Compared with the prior art, the embodiments of the present invention have the following advantages: Different from the existing technology that does not take any action on power data, this solution designs a perfect management method for the power data process. On the one hand, the decentralized data is stored at the low-load edge node, which can not only reduce the load pressure of data storage, but also improve the speed and accuracy of calling data. On the other hand, the core data of the power grid is encrypted for privacy and decrypted when the security conditions are met, thereby improving the security and privacy of power data. The whole process starts from the source of the data, and through reasonable methods and steps, the data encryption is infiltrated into each link of the multi-source heterogeneous power data processing (for example, the privacy data stored in the edge node is obtained by the initial original data encryption processing, and the noise intensity adjustment link is introduced). Multiple encryption links can effectively avoid privacy leakage. In addition, the encryption method steps can remove the restrictions on the data storage location. By storing the encrypted data at the edge node, the storage load of the core node can be reduced by 60-80%. For the distributed power grid system in the new era, the overall load will be greatly reduced, thereby realizing the coordinated improvement of the security and efficiency of multi-source heterogeneous power data processing, and promoting the intelligent process of multi-source heterogeneous power data processing. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1is a flow chart of a multi-source heterogeneous power data processing method in one embodiment of the present invention; Figure 2 It is a logic block diagram of a multi-source heterogeneous power data processing system in one embodiment of the present invention. DETAILED DESCRIPTION

[0017] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. The purpose of providing these embodiments is to make the disclosure of the present invention more thorough and comprehensive. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0018] In the description of this application, the terms "first", "second", "third", etc. are used for descriptive purposes only and are not to be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first", "second", "third", etc. may explicitly or implicitly include one or more of the feature. In the description of this application, unless otherwise specified, "plurality" means two or more.

[0019] In the description of the present application, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be a connection between the two elements. The terms "vertical", "horizontal", "left", "right", "upper", "lower" and similar expressions used herein are only for illustrative purposes, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention. The term "and / or" used herein includes any and all combinations of one or more related listed items. For those of ordinary skill in the art, the specific meanings of the above terms in this application can be understood according to specific circumstances.

[0020] In the description of this application, it should be noted that, unless otherwise defined, all technical and scientific terms used in the present invention have the same meaning as those commonly understood by those skilled in the art. The terms used in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. For those of ordinary skill in the art, the specific meanings of the above terms in this application can be understood by specific circumstances.

[0021] An embodiment of the present invention provides a method for processing multi-source heterogeneous power data. For details, see Figure 1 , Figure 1 The flowchart of the multi-source heterogeneous power data processing method in one embodiment of the present invention is shown, including steps S1 to S6: S1. Encrypting the original data of each decentralized power user in the target power grid to obtain initial encrypted data; S2, removing non-power-related data from the initial encrypted data to obtain a distributed data set; S3, dividing the distributed data set into a number of data subsets, and assigning each data subset to a corresponding edge node, wherein the edge node is determined by the topological node information and communication delay matrix information of the target power grid; S4. Obtaining the private data in each edge node, wherein the private data is obtained by inputting the corresponding data subset into the constructed local model, and the local model is designed to dynamically adjust the introduced noise intensity through a privacy budget allocation mechanism; S5. adjusting the encryption scheme for each private data according to the heterogeneity of the target power grid; S6. In response to the power data acquisition instruction, a decryption scheme adapted to the encryption scheme is executed under security conditions and corresponding privacy data is output.

[0022] It should be noted that with the development of grid-connected technology in the new era, the volume of power data has exploded exponentially, and data is usually forced to be stored in core nodes (such as central servers) to ensure physical or logical isolation. This centralized storage mode directly causes the core nodes to bear 100% of the data reading and writing pressure. If it is not perfectly and effectively processed, it will cause the entire power data load to be redundant and data calls to be extremely difficult, which is not conducive to its effective management. In view of this, the present invention designs a perfect data management method, adopts encryption technology and edge node storage technology, and the two technologies are not just a simple technical superposition. For example, encryption technology can not only ensure the security and privacy of power data, but also the encrypted data can reduce the data load and overhead during transmission. The specific reason is that the distributed power grid system can remove the location restriction of data storage through encryption technology. By storing the encrypted data in the edge node, the storage load of the core node can be reduced by 60-80%. In addition, encryption can realize the secure sharing of data, reduce the need for data replication, and also reduce the storage and transmission overhead, which will be described in detail later.

[0023] First, in the above embodiment, the decentralized stored data will be encrypted at the beginning, fully reflecting the protection of data security and privacy. Optionally, considering that there are many sources of power data, in order to reduce the data load, this embodiment will verify the source of the data, and only encrypt the data that passes the verification, thereby reducing the load of encryption on the node. For the source of data, the data source field in the original data of the power user can be extracted and verified, or it can be verified according to the address bit of the data itself. After passing the verification, in order to ensure the encryption effect, the relevant encryption technology adopts a combination of keys and algorithms. Specifically, the identifier is obtained by reading the corresponding access rules from the preset access policy file, and the RSA algorithm is used to generate an encryption key that matches the original data of the power user that has passed the verification. The access policy file may contain different rules, such as role-based access control (RBAC) or attribute-based access control (ABAC). The identifier may refer to a user ID, role, attribute or other unique identifier. In an encrypted environment, such as ABE, the policy file may define which attribute combinations can decrypt the data, and the system needs to read these policies to generate corresponding public or private key parameters. Then, the original data of the power user that has passed the verification is encrypted based on the AES encryption algorithm and the above encryption key to obtain a data block, and then the identifier and the data block are associated to obtain the initial encrypted data. Of course, when the amount of data is small, the data may not be verified, which is determined by the actual multi-source heterogeneous power data processing requirements. Similarly, the relevant encryption technology can also be determined according to the actual multi-source heterogeneous power data processing requirements, and is not specifically limited in the embodiments of the present invention.

[0024] In order to further reduce the data load and improve the response rate effect of subsequent data calls, this embodiment will also remove non-power-related data in the data. Optionally, the data can be screened by a preset power grid model to remove non-power-related data. This embodiment uses another clustering algorithm to remove it. Specifically, the K-means clustering algorithm is used to remove non-power-related data in the initial encrypted data. Of course, non-power-related data at least includes user identity information. In order to further ensure the quality of the data, this embodiment will adopt a series of data processing methods. Optionally, based on the 3σ criterion, abnormal data in the remaining data is eliminated to ensure data quality. After the above processing, power-related data, such as power consumption data, load rate data, and peak-to-valley difference data, will be obtained. These data need to be distributed, stored, and encrypted to achieve the effectiveness of multi-source heterogeneous power data processing.

[0025] The embodiment of the present invention stores privacy data in edge nodes. The principle of this design is to combine the physical topology and communication performance of the power grid, select the optimal edge node to minimize the data transmission delay and match the computing resources. To achieve this effect, it is necessary to determine each edge node in advance. Optionally, based on the physical topology data obtained from the power system database of the target power grid, it reflects the physical connection relationship of the power grid (such as the hierarchical structure of substation-distribution cabinet-user terminal) through a node diagram. Of course, the attributes of each physical node should also be considered comprehensively, such as the computing power (CPU / GPU resources), storage capacity, and supported protocols (such as MQTT, IEC 61850) of the marked node.

[0026] In order to reduce the communication delay of subsequent data calls, the communication network of the target power grid can be tested for delay, and the delay matrix information can be constructed based on the quantified results of the communication overhead. Specifically, the communication delay values ​​between nodes can be obtained through PING tests or historical transmission records, and the delay matrix Dij can be constructed, where Dij represents the delay from node i to node j. The delay matrix needs to be updated periodically (e.g., every 5 minutes) to adapt to network fluctuations (such as link interruptions caused by faults).

[0027] Finally, the physical topology data and delay matrix information are processed based on the node selection algorithm to obtain each edge node. Optionally, the weighted scoring model is used to calculate the comprehensive score of each candidate edge node Nk using the following formula: Sk=α Tk+β C Among them, Tk is the topological matching degree, which represents whether the level of the node in the power grid is related to the data subset; Ck is the communication efficiency, which represents the average delay from the node to other related nodes based on the delay matrix; α, β are weight coefficients (such as α=0.6, β=0.4).

[0028] By selecting the node with the highest comprehensive score as the belonging node of the data subset, the data subset needs to be allocated to the corresponding edge node.

[0029] In order to further ensure the security and privacy of data, in the embodiment of the present invention, the above data subset will be converted into private data, optionally by introducing moderate noise, and the relevant noise mechanism can refer to the existing technology, which will not be described in detail in the embodiment of the present invention. The details are as follows: Based on the noise scale parameter determined by the preset privacy budget value and the local sample data, the perturbation data satisfying differential privacy is generated; The initial local model is obtained by training with perturbation data; Use the differential privacy verification tool to verify the initial model parameters of the initial local model, and dynamically adjust the initial model parameters based on the verification results to obtain a local model that meets the preset requirements; The private data in each edge node is obtained using a local model.

[0030] For the privacy data in the edge node, this embodiment will also set up a corresponding encryption scheme, which improves the security and privacy of power data through multiple encryption steps. Optionally, in the above embodiment, the model parameters corresponding to the local model obtained from each edge node are input into the Paillier homomorphic encryption algorithm to obtain the encryption model parameters of the current edge node, and then an adaptive parameter aggregation strategy is designed according to the hierarchical information of the edge node, at least the encryption model parameters are compressed, and finally the compression result is processed by the Paillier homomorphic encryption algorithm to generate an encryption scheme for each privacy data.

[0031] Paillier homomorphic encryption allows calculations to be performed in an encrypted state, which is very useful for protecting the privacy of model parameters during transmission. The above-mentioned adaptation refers to the way of adjusting aggregation according to the hierarchical information of the nodes. For example, nodes at different levels may have different weights or compression ratios. The above-mentioned parameter compression can adopt common compression methods, such as quantization, sparsification, or the use of encoding technology. It should be noted that compression in an encrypted state needs to be combined with the characteristics of homomorphic encryption to ensure that the compression operation does not affect the security of the encrypted data.

[0032] For different types of privacy data, the encryption methods are different. This embodiment needs to be adaptively adjusted for different data types. Optionally, the types of privacy data include control instruction data, device status data, and monitoring log data.

[0033] For control instruction data, such as circuit breaker switching instructions, voltage regulation instructions, and energy storage charging and discharging instructions, since their equipment environment is the high-voltage substation controller (high computing power) and distributed energy controller (medium computing power) environment, the corresponding first encryption scheme adopts the national secret SM4 / AES-256 encryption, superimposed with the elliptic curve-based digital signature (ECDSA), and the key is rotated every hour. In addition, the relevant communication protocols also need to be adaptively adjusted.

[0034] For equipment status data, such as transformer temperature, line current value, circuit breaker status, etc., the corresponding second encryption scheme is to use lightweight stream encryption (such as Chacha20) and data block encryption (one key per 1KB).

[0035] For monitoring log data, such as operation logs, abnormal alarm records, access audits, etc., since the device environment is a local log server (medium computing power) and a blockchain node (distributed storage), the corresponding third encryption scheme can use AES-CBC encryption and embed key derivation parameters (PBKDF2) in the log file header.

[0036] In the above step S6, the security condition can be a rule set by the user, such as setting a data field within a certain range, or setting other security conditions. Optionally, blockchain technology is used to generate an audit log around the power data acquisition instruction; the operation record in the audit log is processed by the SHA256 hash algorithm to generate and store the corresponding unique identifier; based on the unique identifier, it is judged whether the data acquisition process complies with the preset rules, wherein the preset rules at least include data access rights and calculation compliance; if it complies with the preset rules, it is judged that the security condition is met. Blockchain technology is different from the data throughput of the power grid itself. It can realize the secure recording of data, thereby facilitating the subsequent judgment process of security conditions.

[0037] Under non-safe conditions, in order to ensure the security and privacy of data, relevant alarm / warning signals can be generated. Optionally, in the above embodiment, if the security conditions are not met, the KNN algorithm is used to analyze the abnormal operation behavior in the corresponding audit log, and generate a suspicious behavior report. The data to be isolated is determined based on the suspicious behavior report, and the isolated data is re-encrypted, thereby achieving effective security protection for the data and avoiding data loss or anomalies caused by hacker attacks.

[0038] Another embodiment of the present invention provides a multi-source heterogeneous power data processing system. For details, see Figure 2 , Figure 2 The following is a logic block diagram of a multi-source heterogeneous power data processing system in one embodiment of the present invention, which includes: The encryption module 11 is used to encrypt the original data of each decentralized power user in the target power grid to obtain initial encrypted data; A removal module 12, used to remove non-power-related data from the initial encrypted data to obtain a distributed data set; A partitioning module 13 is used to partition the distributed data set into a number of data subsets, and allocate each of the data subsets to a corresponding edge node, wherein the edge node is determined by the topological node information and the communication delay matrix information of the target power grid; A privacy module 14, configured to obtain the privacy data in each edge node, wherein the privacy data is obtained by inputting the corresponding data subset into a constructed local model, and the local model is designed to dynamically adjust the introduced noise intensity through a privacy budget allocation mechanism; An encryption scheme module 15, configured to adjust the encryption scheme for each of the private data according to the heterogeneity of the target power grid; The output module 16 is used to respond to the power data acquisition instruction, execute a decryption scheme adapted to the encryption scheme under the security condition and output the corresponding privacy data.

[0039] The multi-source heterogeneous power data processing method and system provided by the embodiments of the present invention have the following beneficial effects: Different from the existing technology that does not take any action on power data, this solution designs a perfect management method for the power data process. On the one hand, the decentralized data is stored at the low-load edge node, which can not only reduce the load pressure of data storage, but also improve the speed and accuracy of calling data. On the other hand, the core data of the power grid is encrypted for privacy and decrypted when the security conditions are met, thereby improving the security and privacy of power data. The whole process starts from the source of the data, and through reasonable methods and steps, the data encryption is infiltrated into each link of the multi-source heterogeneous power data processing (for example, the privacy data stored in the edge node is obtained by the initial original data encryption processing, and the noise intensity adjustment link is introduced). Multiple encryption links can effectively avoid privacy leakage. In addition, the encryption method steps can remove the restrictions on the data storage location. By storing the encrypted data at the edge node, the storage load of the core node can be reduced by 60-80%. For the distributed power grid system in the new era, the overall load will be greatly reduced, thereby realizing the coordinated improvement of the security and efficiency of multi-source heterogeneous power data processing, and promoting the intelligent process of multi-source heterogeneous power data processing.

[0040] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. A multi-source heterogeneous power data processing method, characterized in that: include: Encrypting the original data of each decentralized power user in the target power grid to obtain initial encrypted data; Removing non-power-related data from the initial encrypted data to obtain a distributed data set; Dividing the distributed data set into a number of data subsets, and allocating each of the data subsets to a corresponding edge node, wherein the edge node is determined by topological node information and communication delay matrix information of the target power grid; Obtaining private data in each edge node, wherein the private data is obtained by inputting the corresponding data subset into a constructed local model, and the local model is designed to dynamically adjust the introduced noise intensity through a privacy budget allocation mechanism; adjusting an encryption scheme for each of the private data according to the heterogeneity of the target power grid; In response to the power data acquisition instruction, a decryption scheme adapted to the encryption scheme is executed under a security condition and the corresponding privacy data is output.

2. The multi-source heterogeneous power data processing method according to claim 1, characterized in that: The method of encrypting the original data of each decentralized power user in the target power grid to obtain initial encrypted data includes: Extracting the original data of each decentralized power user and verifying the data source field in the original data of the power user; The identifier is obtained by reading the corresponding access rule from the preset access policy file, and the encryption key matching the verified original data of the power user is generated by using the RSA algorithm; Encrypting the verified original data of the power user based on the AES encryption algorithm and the encryption key to obtain a data block; The identifier and the data block are associated to obtain the initial encrypted data.

3. The multi-source heterogeneous power data processing method according to claim 1, characterized in that: The non-power-related data includes at least user identity information; The removing of non-power-related data from the initial encrypted data to obtain a distributed data set includes: Using a K-means clustering algorithm to remove the user identity information in the initial encrypted data, at least obtaining power consumption data, load rate data, and peak-to-valley difference data; Based on the 3σ criterion, abnormal data in the power consumption data, the load rate data and the peak-to-valley difference data are eliminated in sequence; The distributed data set is constructed based on the elimination results.

4. The multi-source heterogeneous power data processing method according to claim 1, characterized in that: Before dividing the distributed data set into a plurality of data subsets, the method further includes: Acquire physical topology data from a power system database of the target power grid, wherein the physical topology data at least includes node data and connection relationship data; Performing a delay test on the communication network of the target power grid, and constructing delay matrix information based on the obtained communication overhead quantification results; The physical topology data and the delay matrix information are processed based on a node selection algorithm to obtain each edge node.

5. The multi-source heterogeneous power data processing method according to claim 1, characterized in that: The obtaining of the privacy data in each edge node includes: Based on the noise scale parameter determined by the preset privacy budget value and the local sample data, the perturbation data satisfying differential privacy is generated; Using the disturbance data to train an initial local model; Using a differential privacy verification tool to verify the initial model parameters of the initial local model, and dynamically adjusting the initial model parameters based on the verification result to obtain a local model that meets the preset requirements; The private data in each edge node is obtained by using the local model.

6. The multi-source heterogeneous power data processing method according to claim 1, characterized in that: The step of adjusting the encryption scheme for each of the private data according to the heterogeneity of the target power grid includes: Inputting the model parameters corresponding to the local model obtained from each edge node into the Paillier homomorphic encryption algorithm to obtain the encryption model parameters of the current edge node; Designing an adaptive parameter aggregation strategy according to the hierarchical information of the edge node, and at least performing parameter compression on the encryption model parameters; The compression result is processed using the Paillier homomorphic encryption algorithm to generate an encryption scheme for each of the private data.

7. The multi-source heterogeneous power data processing method according to claim 1, characterized in that: The types of privacy data include at least control instruction data, device status data and monitoring log data; the encryption scheme includes at least a first encryption scheme, a second encryption scheme and a third encryption scheme; The first encryption scheme includes the use of national secret SM4 / AES-256 encryption and digital signature based on elliptic curve; the second encryption scheme includes the use of lightweight stream encryption; the third encryption scheme includes embedding key derivation parameters in the log file header.

8. The multi-source heterogeneous power data processing method according to claim 1, characterized in that: Using blockchain technology to generate an audit log around the power data acquisition instructions; Processing the operation records in the audit log using the SHA256 hash algorithm to generate and store a corresponding unique identifier; According to the unique identifier, determining whether the data acquisition process complies with preset rules, wherein the preset rules at least include data access rights and calculation compliance; If the preset rule is met, it is determined that the safety condition is met.

9. The multi-source heterogeneous power data processing method according to claim 8, characterized in that: The method further comprises: If the security condition is not met, the KNN algorithm is used to analyze the abnormal operation behavior in the corresponding audit log and generate a suspicious behavior report; The data to be isolated is determined according to the suspicious behavior report, and the data to be isolated is encrypted twice.

10. A multi-source heterogeneous power data processing system, characterized in that: include: An encryption module is used to encrypt the original data of each decentralized power user in the target power grid to obtain initial encrypted data; A removal module, used to remove non-power-related data from the initial encrypted data to obtain a distributed data set; A partitioning module, used to divide the distributed data set into a number of data subsets, and assign each of the data subsets to a corresponding edge node, wherein the edge node is determined by the topological node information and communication delay matrix information of the target power grid; A privacy module, configured to obtain private data in each edge node, wherein the private data is obtained by inputting the corresponding data subset into a constructed local model, and the local model is designed to dynamically adjust the introduced noise intensity through a privacy budget allocation mechanism; An encryption scheme module, configured to adjust the encryption scheme for each of the private data according to the heterogeneity of the target power grid; The output module is used to respond to the power data acquisition instruction, execute a decryption scheme adapted to the encryption scheme under the security condition and output the corresponding privacy data.

Images