Data processing method and system based on dual-mode carrier module

Abstract

The invention discloses a data processing method and system based on a dual-mode carrier module. The method comprises the following steps: determining a first algorithm and a second algorithm conforming to a homomorphic encryption addition algorithm; a first algorithm is issued to the first party, and the first party obtains a first statistical index related to the first electric power information of the first area and processes the first statistical index by adopting the first algorithm to obtain first encrypted data; after first encrypted data is obtained from the first party and verified, the first encrypted data and a second algorithm are sent to a second party, and the second party processes a second statistical index related to second power information of a second area according to the second algorithm and determines a processing result; receiving a processing result, and determining first related data and second related data; and training a power anomaly recognition model according to the first related data and the second related data so as to perform anomaly recognition on the power information of the first region according to the trained power anomaly recognition model.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a data processing method and system based on a dual-mode carrier module. Background Art

[0002] Existing power information anomaly analysis usually relies on manual analysis or trains an identification model based on normal power data and abnormal power data in a single area, and performs power anomaly identification through the identification model.

[0003] However, adopting the above identification scheme, the model relying on manual or small amounts of data training has a low accuracy in power anomaly identification. Summary of the Invention

[0004] The present invention provides a data processing method and system based on a dual-mode carrier module, which can adopt data from multiple parties to train a power anomaly identification model, thereby increasing the amount of training data to improve the accuracy of power anomaly identification.

[0005] To solve the above technical problems, the present invention is implemented as follows:

[0006] In a first aspect, the present application provides a data processing method based on a dual-mode carrier module, the method including: determining a first algorithm and a second algorithm that conform to the homomorphic encryption addition algorithm; sending the first algorithm to a first party, the first party obtaining a first statistical index related to the first power information in a first area, and processing the first statistical index using the first algorithm to obtain first encrypted data; wherein, the first statistical index includes a data anomaly rate, data distribution information, total data volume, peak-valley information; the first power information is obtained based on a dual-mode carrier module, and the dual-mode carrier module includes a single-phase carrier module and a two-way carrier module; after obtaining and verifying the first encrypted data from the first party, sending the first encrypted data and the second algorithm to a second party, the second party processing a second statistical index related to the second power information in a second area using the second algorithm to obtain second encrypted data, and determining a processing result based on the first encrypted data and the second encrypted data; receiving the processing result uploaded by the second party, decrypting the processing result to obtain a data difference, and determining whether the data difference meets a preset threshold to determine first related data from the first power information and second related data from the second power information; training a power anomaly identification model based on the first related data and the second related data, so as to perform anomaly identification on the power information in the first area based on the trained power anomaly identification model.

[0007] Further, training the power anomaly recognition model based on the first relevant data and the second relevant data includes: deploying the power anomaly recognition model to the first party and the second party, and the first party and the second party training the power anomaly recognition model according to the local power information and uploading the trained model parameters; receiving the first model parameters of the first party and the second model parameters of the second party, and performing comprehensive analysis to obtain the parameter similarity of the first model parameters and the second model parameters and the global model parameters; updating the parameters of the power anomaly recognition model according to the global model parameters, and updating the model parameters of the first party and the second party. The first party and the second party perform the next round of training according to the updated power anomaly recognition model and local data; determining the parameter compression schemes for the first party and the second party according to the parameter similarity and sending them to the first party and the second party. The first party and the second party compress the model parameters according to the parameter compression schemes before uploading the model parameters.

[0008] Further, the method further includes: obtaining the compressed model parameters uploaded by the first party and the second party, and obtaining the parameter similarity; restoring the model parameters according to the compressed model parameters and the parameter similarity, so as to determine the global model parameters and the parameter similarity for the next round according to the restored model parameters for the next round of iterative training; wherein, in the process of determining the global model parameters according to the model parameters, determining the parameter loss according to the model parameters of the first party and the model parameters of the second party, and when the parameter loss is the smallest, determining the global model parameters; if the restoration of the model parameters fails, sending a model parameter upload instruction to the first party and the second party to obtain the model parameters, wherein it is determined whether the restoration of the model parameters is successful according to the parameter loss in this round of training and the parameter loss in the previous round.

[0009] Further, the data difference includes data anomaly rate difference, data distribution difference, data volume difference, and peak-valley information difference; judging whether the data difference meets a preset threshold to determine the first relevant data from the first power information and the second relevant data from the second power information includes: screening out the first relevant data and the second relevant data according to the comparison between the processing result and the preset threshold, wherein the differences in at least two of the data anomaly rate, data distribution information, data volume, and peak-valley information of the first relevant data and the second relevant data are less than the preset threshold and at least one difference is greater than the preset threshold.

[0010] Further, the step of verifying the first encrypted data is processed within a secure computing environment. The step of verifying the first encrypted data includes: generating a first standard value and a second standard value, and processing the first standard value and the second standard value using a second algorithm to obtain a first verification data and a second verification data; processing the first encrypted data using the first verification data and the second verification data to obtain a first verification result and a second verification result; and determining whether the first encrypted data passes the verification based on the first verification result and the second verification result.

[0011] Further, the determining whether the first encrypted data passes the verification based on the first verification result and the second verification result includes: determining whether the difference between the original data of the first encrypted data and the first standard value is within a preset difference based on the first verification result to obtain a first judgment result; determining whether the difference between the original data of the first encrypted data and the second standard value exceeds the preset difference based on the second verification result to obtain a second judgment result; and determining whether the first encrypted data passes the verification based on the first judgment result and the second judgment result.

[0012] Further, the method further includes: obtaining power information acquired based on a dual-mode carrier module, and obtaining historical power information; analyzing the Pearson correlation coefficients of the peak-valley information, line loss rate, and total power in the historical power information and the Pearson correlation coefficients of the peak-valley information, line loss rate, and total power in the power information to screen the power information and determine the target power information; and inputting the target power information into a trained power anomaly recognition model for recognition to determine and output the abnormal power information.

[0013] Further, the second party is used to: divide the local power information into multiple data groups, and generate a second statistical index for each data group for processing based on the second statistical index and the second algorithm, where the data groups divided by the second party include a first data set and a second data set, the data included in the two first data sets are different, and the first data set is a subset of the second data set.

[0014] Second aspect, the present application provides a data processing system based on a dual-mode carrier module. The system includes: a homomorphic encryption processing module for determining a first algorithm and a second algorithm that conform to the homomorphic encryption addition algorithm; a first algorithm distribution module for distributing the first algorithm to a first party. The first party obtains a first statistical index related to the first power information in a first area and processes the first statistical index using the first algorithm to obtain first encrypted data. Wherein, the first statistical index includes a data abnormality rate, data distribution information, total data volume, and peak-valley information. The first power information is obtained based on a dual-mode carrier module, and the dual-mode carrier module includes a single-phase carrier module and a two-way carrier module; a second algorithm distribution module for obtaining the first encrypted data from the first party, performing verification, and then sending the first encrypted data and the second algorithm to a second party. The second party processes a second statistical index related to the second power information in a second area using the second algorithm to obtain second encrypted data, and determines a processing result based on the first encrypted data and the second encrypted data; a processing result receiving module for receiving the processing result uploaded by the second party, decrypting the processing result to obtain a data difference, and determining whether the data difference meets a preset threshold, so as to determine first related data from the first power information and second related data from the second power information; an identification model training module for training a power abnormality identification model based on the first related data and the second related data, so as to identify abnormalities in the power information in the first area according to the trained power abnormality identification model.

[0015] Third aspect, the present application provides an electronic device, including: a memory and at least one processor; the memory is used for storing computer execution instructions; the at least one processor is used for executing the computer execution instructions stored in the memory, so that the at least one processor executes the method described in the first aspect.

[0016] Embodiments of the present application can be applied to scenarios of abnormal power information recognition. Existing solutions usually use power information in a single area to train a model. However, the data volume in a single area is small, and due to privacy reasons, data in multiple areas cannot be interacted in plaintext. This solution can complete the matching of model data and the model training process without interacting data in plaintext. It can use data from multiple parties to complete the training, improving both the training data volume and the recognition accuracy. This solution can generate statistical metrics and use the method of homomorphic encryption to match the statistical metrics to select the training data for the model, and then complete the model training process based on the method of federated learning. Specifically, this solution can determine a first algorithm and a second algorithm that conform to the homomorphic encryption addition algorithm; then, send the first algorithm to the first party. The first party obtains the first statistical metrics related to the first power information in the first area and processes the first statistical metrics using the first algorithm to obtain the first encrypted data; where the first statistical metrics include data anomaly rate, data distribution information, total data volume, peak-valley information; the first power information is obtained based on a dual-mode carrier module, and the dual-mode carrier module includes a single-phase carrier module and a two-way carrier module; after obtaining and verifying the first encrypted data from the first party, send the first encrypted data and the second algorithm to the second party. The second party processes the second statistical metrics related to the second power information in the second area using the second algorithm to obtain the second encrypted data, and determines the processing result based on the first encrypted data and the second encrypted data; receive the processing result uploaded by the second party, decrypt the processing result to obtain the data difference, and determine whether the data difference meets a preset threshold, so as to determine the first relevant data from the first power information and the second relevant data from the second power information; where this solution can screen out data that are similar but have some differences as the first relevant data and the second relevant data, and then train a power anomaly recognition model based on the first relevant data and the second relevant data, so as to perform anomaly recognition on the power information in the first area based on the trained power anomaly recognition model. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The drawings described herein are used to provide a further understanding of the present invention and form a part of the present invention. The illustrative embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation to the present invention. In the drawings:

[0018] Figure 1 is a flowchart of a data processing method based on a dual-mode carrier module according to an embodiment of the present application;

[0019] Figure 2 is a structural diagram of a data processing system based on a dual-mode carrier module according to an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0020] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0021] The embodiments of the present application can be applied to the scenario of power information anomaly recognition. The existing solutions usually use the power information of a single area to train the model. However, the data volume of a single area is small, and the data of multiple areas cannot be interacted in plaintext due to privacy reasons. This solution can complete the matching of model data and the model training process without interacting data in plaintext, and can use the data of multiple parties to complete the training, improving both the training data volume and the recognition accuracy. This solution can generate statistical metrics and use the method of homomorphic encryption to match the statistical metrics to select the training data of the model, and then complete the model training process based on the method of federated learning. Specifically, this solution can determine a first algorithm and a second algorithm that conform to the homomorphic encryption addition algorithm; then, send the first algorithm to the first party, and the first party obtains the first statistical metric related to the first power information of the first area and processes the first statistical metric using the first algorithm to obtain the first encrypted data; wherein, the first statistical metric includes data anomaly rate, data distribution information, total data volume, peak-valley information; the first power information is obtained based on a dual-mode carrier module, and the dual-mode carrier module includes a single-phase carrier module and a two-way carrier module; after obtaining and verifying the first encrypted data from the first party, send the first encrypted data and the second algorithm to the second party, and the second party processes the second statistical metric related to the second power information of the second area using the second algorithm to obtain the second encrypted data, and determines the processing result based on the first encrypted data and the second encrypted data; receive the processing result uploaded by the second party, decrypt the processing result to obtain the data difference, and determine whether the data difference meets the preset threshold, so as to determine the first relevant data from the first power information and the second relevant data from the second power information; wherein, this solution can screen out the data that are similar and have some differences as the first relevant data and the second relevant data, and then train the power anomaly recognition model based on the first relevant data and the second relevant data, so as to perform anomaly recognition on the power information of the first area according to the trained power anomaly recognition model.

[0022] Specifically, the embodiments of the present application provide a data processing method based on a dual-mode carrier module, as Figure 1 shown, the method includes:

[0023] Step 102: Determine the first algorithm and the second algorithm that conform to the homomorphic encryption addition algorithm. Homomorphic encryption is a scheme that can perform calculations in the encrypted state and then decrypt the calculation result to obtain the calculation result. This scheme can complete data calculations without exposing the original data. The homomorphic encryption addition algorithm is one of the encryption calculation schemes of the homomorphic encryption algorithm.

[0024] Step 104: Send the first algorithm to the first party. The first party obtains the first statistical indicators related to the first power information in the first area and processes the first statistical indicators using the first algorithm to obtain the first encrypted data. Among them, the first statistical indicators include data anomaly rate, data distribution information, total data volume, and peak-valley information. The first power information is obtained based on a dual-mode carrier module, and the dual-mode carrier module includes a single-phase carrier module and a two-way carrier module.

[0025] Step 106: After obtaining and verifying the first encrypted data from the first party, send the first encrypted data and the second algorithm to the second party. The second party processes the second statistical indicators related to the second power information in the second area using the second algorithm to obtain the second encrypted data, and determines the processing result based on the first encrypted data and the second encrypted data.

[0026] Step 108: Receive the processing result uploaded by the second party, decrypt the processing result to obtain the data difference, and determine whether the data difference meets the preset threshold, so as to determine the first relevant data from the first power information and the second relevant data from the second power information.

[0027] Step 110: Train a power anomaly recognition model based on the first relevant data and the second relevant data, so as to perform anomaly recognition on the power information in the first area based on the trained power anomaly recognition model.

[0028] Embodiments of the present application can be applied to the scenario of abnormal power information recognition. Existing solutions usually use the power information of a single area to train a model. However, the data volume of a single area is small, and the data of multiple areas cannot be interacted in plaintext due to privacy reasons. This solution can complete the matching of model data and the model training process without interacting data in plaintext, and can use the data of multiple parties to complete the training, improving the training data volume and the recognition accuracy. This solution can generate statistical metrics and use the method of homomorphic encryption to match the statistical metrics to select the training data of the model, and then complete the model training process based on the method of federated learning. Specifically, this solution can determine a first algorithm and a second algorithm that conform to the homomorphic encryption addition algorithm; then, send the first algorithm to the first party, and the first party obtains the first statistical metric related to the first power information of the first area and processes the first statistical metric using the first algorithm to obtain the first encrypted data; wherein, the first statistical metric includes data anomaly rate, data distribution information, total data volume, peak-valley information; the first power information is obtained based on a dual-mode carrier module, and the dual-mode carrier module includes a single-phase carrier module and a two-way carrier module; after obtaining and verifying the first encrypted data from the first party, send the first encrypted data and the second algorithm to the second party, and the second party processes the second statistical metric related to the second power information of the second area according to the second algorithm to obtain the second encrypted data, and determines the processing result based on the first encrypted data and the second encrypted data; receive the processing result uploaded by the second party, decrypt the processing result to obtain the data difference, and determine whether the data difference meets a preset threshold, so as to determine the first relevant data from the first power information and the second relevant data from the second power information; wherein, this solution can screen out data that are similar and have some differences as the first relevant data and the second relevant data, and then train a power anomaly recognition model based on the first relevant data and the second relevant data, so as to perform anomaly recognition on the power information of the first area according to the trained power anomaly recognition model.

[0029] During the process of training the model, the present solution can adopt the method of federated learning for training. Among them, federated learning uses the local models and local data of multiple parties to complete the training, and then uploads the model parameters to the server. The server receives the parameters of multiple parties for aggregation analysis, obtains the global parameters and updates the model, and completes the training process through multiple rounds of iteration. Specifically, as an optional embodiment, the training of the power anomaly recognition model based on the first relevant data and the second relevant data includes: deploying the power anomaly recognition model to the first party and the second party. The first party and the second party train the power anomaly recognition model based on the local power information and upload the trained model parameters; receiving the first model parameters of the first party and the second model parameters of the second party, and conducting comprehensive analysis to obtain the parameter similarity of the first model parameters and the second model parameters and the global model parameters; updating the parameters of the power anomaly recognition model according to the global model parameters, and updating the model parameters of the first party and the second party. The first party and the second party conduct the next round of training based on the updated power anomaly recognition model and local data; determining the parameter compression schemes for the first party and the second party according to the parameter similarity, and sending them to the first party and the second party. The first party and the second party compress the model parameters according to the parameter compression schemes before uploading the model parameters.

[0030] Among them, the data of multiple parties has a certain similarity, so the corresponding model parameters will have similarity. Therefore, the present solution can determine the parameter compression scheme according to the similarity of the model parameters in the previous round. Each party compresses the model parameters uploaded in the next round according to the parameter compression scheme, which can reduce the number of model parameters for interaction. For the solution that requires multiple rounds of iterative training, the present solution can reduce the transmission pressure. Specifically, as an optional embodiment, the method further includes: obtaining the compressed model parameters uploaded by the first party and the second party, and obtaining the parameter similarity; restoring the model parameters according to the compressed model parameters and the parameter similarity, so as to determine the global model parameters and the parameter similarity in the next round according to the restored model parameters for the next round of iterative training; among them, during the process of determining the global model parameters according to the model parameters, determining the parameter loss according to the model parameters of the first party and the model parameters of the second party. When the parameter loss is the smallest, determining the global model parameters; if the restoration of the model parameters fails, sending a model parameter upload instruction to the first party and the second party to obtain the model parameters, where it is determined whether the model parameters are successfully restored according to the parameter loss in this round of training and the parameter loss in the previous round. If the restoration of the model parameters fails, the present solution can require each party to upload all the parameters and conduct an analysis of the global parameters. Moreover, the present solution can judge whether the parameters are successfully restored according to the difference between the parameter loss in this round and the parameter loss in the previous round.

[0031] In the process of screening data, this solution can screen out data that are similar but have some differences for training. Because if the data have a high similarity, it may lead to overfitting of the model. Therefore, this solution preserves both the similarity and the differences of the data to better train the model. Specifically, as an optional embodiment, the data differences include data anomaly rate differences, data distribution differences, total data volume differences, and peak-valley information differences; the judgment of whether the data differences meet the preset threshold to determine the first relevant data from the first power information and the second relevant data from the second power information includes: comparing the processing result with the preset threshold to screen out the first relevant data and the second relevant data, where the differences in at least two of the data anomaly rate, data distribution information, total data volume, and peak-valley information of the first relevant data and the second relevant data are less than the preset threshold and at least one difference is greater than the preset threshold.

[0032] This solution can verify the first encrypted data to prevent problems caused by incorrect encryption, etc. During the verification process, this solution can generate a normal value and an extreme value (or an extremely small value). Generally speaking, statistical indicators will have a small difference from the normal value and a large difference from the extreme value. Therefore, this solution analyzes the data in a secure computing environment to check whether the first encrypted data is within the range of the normal value. Among them, the secure computing environment can not expose the standard value and the data, and only outputs the result. Specifically, as an optional embodiment, the step of verifying the first encrypted data is processed in a secure computing environment. The step of verifying the first encrypted data includes: generating a first standard value and a second standard value, and processing the first standard value and the second standard value using a second algorithm to obtain a first verification data and a second verification data; using the first verification data and the second verification data to process the first encrypted data to obtain a first verification result and a second verification result; determining whether the first encrypted data passes the verification based on the first verification result and the second verification result. Specifically, as an optional embodiment, the determination of whether the first encrypted data passes the verification based on the first verification result and the second verification result includes: determining whether the difference between the original data of the first encrypted data and the first standard value is within the preset difference based on the first verification result to obtain a first judgment result; determining whether the difference between the original data of the first encrypted data and the second standard value exceeds the preset difference based on the second verification result to obtain a second judgment result; determining whether the first encrypted data passes the verification based on the first judgment result and the second judgment result.

[0033] In this solution, a screening condition can be set first to filter out some power information that is basically within the normal range. For example, if the power information is less than the target value, it can be not analyzed, or the consistency between the historical power information and the current-stage power information can be analyzed to determine whether the data may be abnormal. When the data may be abnormal, the data is input into the model for analysis. This solution can analyze the Pearson correlation coefficients of peak-valley information, line loss rate, and total power to determine whether the power information may be abnormal for further analysis. Specifically, as an optional embodiment, the method further includes: obtaining the power information obtained based on the dual-mode carrier module and obtaining the historical power information; analyzing based on the Pearson correlation coefficients of peak-valley information, line loss rate, and total power in the historical power information and the Pearson correlation coefficients of peak-valley information, line loss rate, and total power in the power information, screening the power information to determine the target power information; inputting the target power information into the trained power anomaly recognition model for recognition, determining the abnormal power information and outputting it.

[0034] For the second party, if the second party aggregates all the data and then calculates the statistical indicators, it may miss some data that can be used for model training. Therefore, this solution can group the data of the second party and aggregate multiple groups into one group, and then obtain multiple sets of second statistical indicators to analyze the first statistical indicator and the second statistical indicators to determine the data that can be used for model training. Specifically, as an optional embodiment, the second party is used to: divide the local power information into multiple data groups and generate second statistical indicators for each data group to process according to the second statistical indicators and the second algorithm. Among them, the data groups divided by the second party include a first data set and a second data set. The data included in the two first data sets is different, and the first data set is a subset of the second data set.

[0035] Based on the above embodiments, the embodiment of the present application further provides a data processing system based on a dual-mode carrier module, as Figure 2 shown. The system includes:

[0036] A homomorphic encryption processing module 202, which is used to determine a first algorithm and a second algorithm that conform to the homomorphic encryption addition algorithm.

[0037] A first algorithm distribution module 204, which is used to distribute the first algorithm to the first party. The first party obtains the first statistical indicators related to the first power information in the first area and processes the first statistical indicators using the first algorithm to obtain the first encrypted data. Among them, the first statistical indicators include data anomaly rate, data distribution information, data total amount, and peak-valley information; the first power information is obtained based on the dual-mode carrier module, and the dual-mode carrier module includes a single-phase carrier module and a two-way carrier module.

[0038] The second algorithm distribution module 206 is configured to obtain the first encrypted data from the first party, perform verification, and then send the first encrypted data and the second algorithm to the second party. The second party processes the second statistical metrics related to the second power information in the second region according to the second algorithm to obtain the second encrypted data, and determines the processing result based on the first encrypted data and the second encrypted data.

[0039] The processing result receiving module 208 is configured to receive the processing result uploaded by the second party, decrypt the processing result to obtain the data difference, and determine whether the data difference meets a preset threshold, so as to determine the first relevant data from the first power information and the second relevant data from the second power information.

[0040] The recognition model training module 210 is configured to train a power anomaly recognition model based on the first relevant data and the second relevant data, and perform anomaly recognition on the power information in the first region according to the trained power anomaly recognition model.

[0041] The implementation manner of the embodiment of the present application is similar to that of the above method embodiment. The specific implementation manner can refer to the specific implementation manner of the above method embodiment, which will not be elaborated here.

[0042] Embodiments of the present application can be applied to scenarios of abnormal power information recognition. Existing solutions usually use power information in a single area to train a model. However, the data volume in a single area is small, and due to privacy reasons, data in multiple areas cannot be interacted in plaintext. This solution can complete the matching of model data and the model training process without interacting data in plaintext. It can use data from multiple parties to complete the training, improving both the training data volume and the recognition accuracy. This solution can generate statistical metrics and use the method of homomorphic encryption to match the statistical metrics to select the training data of the model, and then complete the model training process based on the method of federated learning. Specifically, this solution can determine a first algorithm and a second algorithm that conform to the homomorphic encryption addition algorithm; then, send the first algorithm to the first party. The first party obtains the first statistical metric related to the first power information in the first area and processes the first statistical metric using the first algorithm to obtain the first encrypted data; wherein, the first statistical metric includes the data anomaly rate, data distribution information, total data volume, and peak-valley information; the first power information is obtained based on a dual-mode carrier module, and the dual-mode carrier module includes a single-phase carrier module and a two-way carrier module; after obtaining and verifying the first encrypted data from the first party, send the first encrypted data and the second algorithm to the second party. The second party processes the second statistical metric related to the second power information in the second area using the second algorithm to obtain the second encrypted data, and determines the processing result based on the first encrypted data and the second encrypted data; receive the processing result uploaded by the second party, decrypt the processing result to obtain the data difference, and determine whether the data difference meets a preset threshold, so as to determine the first relevant data from the first power information and the second relevant data from the second power information; wherein, this solution can screen out data that are similar but have some differences as the first relevant data and the second relevant data, and then train a power anomaly recognition model based on the first relevant data and the second relevant data, so as to perform anomaly recognition on the power information in the first area based on the trained power anomaly recognition model.

[0043] Based on the above embodiments, the present application further provides an electronic device, including: a memory and at least one processor; the memory is used to store computer execution instructions; the at least one processor is used to execute the computer execution instructions stored in the memory, so that the at least one processor executes the method as described in the above embodiments.

[0044] An embodiment of the present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements each process of the above-described method embodiment for processing data and can achieve the same technical effects. To avoid repetition, it will not be elaborated here. Among them, the computer-readable storage medium is, for example, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disc, etc.

[0045] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0046] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to embodiments of the present invention. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of flows and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the specified functions in one Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0047] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured article including instruction means, and the instruction means implements the specified functions in one Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0048] These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process. Thus, the instructions executed on the computer or other programmable device provide means for implementing the specified functions in one Figure 1 one flow or multiple flows and / or blocks Figure 1Steps for the functions specified in one or more boxes.

[0049] In a typical configuration, a computing device includes one or more processors (CPUs), an input / output interface, a network interface, and memory.

[0050] The memory may include non-permanent memory in the form of computer-readable media, random access memory (RAM), and / or non-volatile memory such as read-only memory (ROM) or flash memory (flash RAM). Memory is an example of computer-readable media.

[0051] Computer-readable media includes permanent and non-permanent, removable and non-removable media that can store information by any method or technology. The information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassette tapes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transitory media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include transitory computer-readable media such as modulated data signals and carrier waves.

[0052] It should also be noted that the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but also other elements not expressly listed, or elements that are inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the statement "comprising an..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus that comprises the element.

[0053] Those skilled in the art will appreciate that the embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0054] The above are only embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, various modifications and variations can be made to the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the scope of the claims of the present invention.

Claims

1. A data processing method based on a dual-mode carrier module, characterized in that, The method includes: Determine a first algorithm and a second algorithm that conform to the homomorphic encryption addition algorithm; Send the first algorithm to the first party. The first party obtains the first statistical indicators related to the first power information in the first area, and processes the first statistical indicators using the first algorithm to obtain the first encrypted data. Among them, the first statistical indicators include data anomaly rate, data distribution information, total data volume, peak-valley information; the first power information is obtained based on a dual-mode carrier module, and the dual-mode carrier module includes a single-phase carrier module and a two-way carrier module; After obtaining and verifying the first encrypted data from the first party, send the first encrypted data and the second algorithm to the second party. The second party processes the second statistical indicators related to the second power information in the second area using the second algorithm to obtain the second encrypted data, and determines the processing result based on the first encrypted data and the second encrypted data; Receive the processing result uploaded by the second party, decrypt the processing result to obtain the data difference, and determine whether the data difference meets the preset threshold, so as to determine the first relevant data from the first power information and the second relevant data from the second power information; Train a power anomaly recognition model based on the first relevant data and the second relevant data, so as to recognize anomalies in the power information of the first area based on the trained power anomaly recognition model.

2. The method according to claim 1, wherein The training of the power anomaly recognition model based on the first relevant data and the second relevant data includes: Deploy the power anomaly recognition model to the first party and the second party. The first party and the second party train the power anomaly recognition model based on the local power information and upload the trained model parameters; Receive the first model parameters of the first party and the second model parameters of the second party, and perform comprehensive analysis to obtain the parameter similarity of the first model parameters and the second model parameters and the global model parameters; Update the parameters of the power anomaly recognition model according to the global model parameters, and update the model parameters of the first party and the second party. The first party and the second party perform the next round of training based on the updated power anomaly recognition model and local data; Determine the parameter compression schemes for the first party and the second party according to the parameter similarity, and send them to the first party and the second party. The first party and the second party compress the model parameters according to the parameter compression schemes before uploading the model parameters.

3. The method according to claim 2, characterized in that The method further includes: Obtain the compressed model parameters uploaded by the first party and the second party, and obtain the parameter similarity; Restore the model parameters according to the compressed model parameters and the parameter similarity, so as to determine the global model parameters and the parameter similarity for the next round based on the restored model parameters for the next round of iterative training; among them, in the process of determining the global model parameters according to the model parameters, determine the parameter loss according to the model parameters of the first party and the model parameters of the second party. When the parameter loss is the smallest, determine the global model parameters; If the restoration of the model parameters fails, send a model parameter upload instruction to the first party and the second party to obtain the model parameters, where it is determined whether the model parameters are successfully restored according to the parameter loss in this round of training and the parameter loss in the previous round.

4. The method according to claim 3, wherein The data difference includes data anomaly rate difference, data distribution difference, total data quantity difference, and peak-valley information difference; Judging whether the data difference meets a preset threshold to determine first relevant data from the first power information and second relevant data from the second power information includes: Based on comparing the processing result with the preset threshold, filtering out the first relevant data and the second relevant data, where the difference in at least two of the data anomaly rate, data distribution information, total data quantity, and peak-valley information between the first relevant data and the second relevant data is less than the preset threshold and the difference in at least one item is greater than the preset threshold.

5. The method according to claim 4, characterized in that The step of verifying the first encrypted data is processed in a secure computing environment. The step of verifying the first encrypted data includes: Generating a first standard value and a second standard value, and processing the first standard value and the second standard value using a second algorithm to obtain a first verification data and a second verification data; Processing the first encrypted data using the first verification data and the second verification data to obtain a first verification result and a second verification result; Determining whether the first encrypted data passes the verification based on the first verification result and the second verification result.

6. The method according to claim 5, characterized in that, The determining whether the first encrypted data passes the verification based on the first verification result and the second verification result includes: Determining whether the difference between the original data of the first encrypted data and the first standard value is within a preset difference based on the first verification result to obtain a first judgment result; Determining whether the difference between the original data of the first encrypted data and the second standard value exceeds the preset difference based on the second verification result to obtain a second judgment result; Determining whether the first encrypted data passes the verification based on the first judgment result and the second judgment result.

7. The method according to claim 6, characterized in that, The method further includes: Obtaining power information acquired based on a dual-mode carrier module, and obtaining historical power information; Analyzing based on the Pearson correlation coefficients of peak-valley information, line loss rate, and total power quantity in the historical power information and the Pearson correlation coefficients of peak-valley information, line loss rate, and total power quantity in the power information, screening the power information, and determining target power information; Inputting the target power information into a trained power anomaly recognition model for recognition, determining abnormal power information, and outputting it.

8. The method according to claim 7, wherein The second party is used for: Dividing the local power information into multiple data groups, and generating a second statistical index for each data group to be processed based on the second statistical index and the second algorithm. The data groups divided by the second party include a first data set and a second data set. The data included in the two first data sets is different, and the first data set is a subset of the second data set.

9. A data processing system based on a dual-mode carrier module, characterized in that, The system includes: A homomorphic encryption processing module for determining a first algorithm and a second algorithm that conform to the homomorphic encryption addition algorithm; A first algorithm distribution module for distributing the first algorithm to the first party. The first party obtains a first statistical index related to the first power information in the first region, and processes the first statistical index using the first algorithm to obtain first encrypted data; where the first statistical index includes data anomaly rate, data distribution information, total data quantity, and peak-valley information; the first power information is acquired based on a dual-mode carrier module, and the dual-mode carrier module includes a single-phase carrier module and a two-way carrier module; The second algorithm distribution module is used to obtain the first encrypted data from the first party, perform verification, and then send the first encrypted data and the second algorithm to the second party. The second party processes the second statistical indicators related to the second power information in the second area according to the second algorithm to obtain the second encrypted data, and determines the processing result based on the first encrypted data and the second encrypted data; The processing result receiving module is used to receive the processing result uploaded by the second party, decrypt the processing result to obtain the data difference, and determine whether the data difference meets the preset threshold, so as to determine the first relevant data from the first power information and the second relevant data from the second power information; The recognition model training module is used to train the power anomaly recognition model based on the first relevant data and the second relevant data, so as to perform anomaly recognition on the power information in the first area according to the trained power anomaly recognition model.

10. An electronic device, characterized in that, Comprising: A memory and at least one processor; The memory is used to store computer execution instructions; The at least one processor is used to execute the computer execution instructions stored in the memory, so that the at least one processor executes the method according to any one of claims 1-8.

Images