A data security sharing decision method, system and server of a power internet of things

Abstract

The application provides a data security sharing decision method, system and server of a power internet of things, relates to the technical field of the power internet of things, and collects the privacy data of the power internet of things equipment such as smart meters in the local region, cleans and enhances the data, and generates standard data that can be used for training. The security sharing and model training of the data are carried out, the parameters transferred from the mature model and issued by the central server are partially frozen, the local model training is carried out, the intermediate data generated during the training of the local client is securely shared, and the central server uniformly processes and issues the data, makes decision analysis on the existing demand, gives a decision suggestion, outputs the model for result prediction or classification, searches the expert database decision suggestion, and gives a decision suggestion by comprehensively considering the model result, so that the security sharing of cross-region data is ensured, and decision convenience is provided for professionals.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power internet of things, and in particular to a data security sharing decision method and system for power internet of things and a server. BACKGROUND

[0002] Automated decision-making in the field of power internet of things has been valued, and more and more enterprises have begun to build their own automated decision-making solutions. In the past, decision-making in the field of power internet of things mainly relied on the professional experience and knowledge of experts, and artificial decision-making assisted computer means, which was time-consuming and costly.

[0003] Data has become another object of people's focus in today's society. After the emergence of emerging concepts such as Industry 4.0, the industry and power internet of things field has put forward new requirements for the mining and utilization of various power data. The industry is eager to obtain sufficient data from smart meters or other smart power equipment for decision-making, such as using power consumption data to reasonably allocate power resources in peak and valley, using real-time image data of power equipment to evaluate and warn equipment wear, and using cross-regional power data to develop highly generalized top-level strategies.

[0004] Data such as user power consumption information, regional power distribution lines and internal power equipment in the field are usually identified as private data that needs to be protected. However, there is little power data possessed by each entity, whether it is a country or a region or even a power enterprise entity. Considering the interests and privacy data protection, each entity also has difficulty in aggregating cross-regional data.

[0005] Traditional power internet of things-oriented decision-making systems are often based on small data volume models for training, which may not be effective or not suitable for all regional scenarios. The systems involved do not have migratability and expandability. In addition, it is worth noting that traditional power decision-making systems cannot guarantee the secure sharing of cross-regional data, and various private data in the field of power internet of things cannot be guaranteed in terms of security. Once the private data is leaked, it will have a significant impact on the internet of things. SUMMARY

[0006] The present application provides a data security sharing and decision-making method for power internet of things, which ensures the secure sharing of cross-regional data and provides security for various private data in the field of power internet of things, providing convenience for professionals to make decisions.

[0007] The data security sharing and decision-making method for power internet of things comprises:

[0008] Step A, local client c i ∈C preprocesses the collected power internet of things data, and arranges the local power internet of things device data Datai DataUpdatei=Data+Data i DataUpdatei=Data+Data i );

[0009] wherein C,Data is the total client set and the total power internet of things data set, and RF(·) is the data enhancement method;

[0010] Step B, based on the enhanced data, the local client and the trusted central server cooperatively train, iteratively train the model using federated transfer learning and homomorphic encryption technology to securely share data, and generate an output model Model after completing the training target;

[0011] Step C, based on the output model Model generated in step B and the power internet of things demand Env, use the trained cross-region model to make predictions or analysis, and generate decision suggestions.

[0012] Further need to be explained is that in step C, the model trained by federated transfer learning is used to solve:

[0013] Solution=getSol(Env,Model,Sug)

[0014] wherein Solution is a solution set, getSol(·) is an abstract function for obtaining a solution, and Sug is an auxiliary suggestion retrieved from an expert experience library.

[0015] Further need to be explained is that the step B specifically includes the following steps:

[0016] Step B1, for a specific demand Env, select a mature model TModel for transfer;

[0017] Step B2, each local client receives the model TModel selected by the central server to solve the demand Env and a set of initialized parameters V p ; freeze the first k layers of the model parameters, load the parameters on Tmodel, and then randomly initialize the remaining layer parameters;

[0018] Step B3, the local client encrypts the intermediate data after a round of training;

[0019] Step B4, the central server receives the encrypted intermediate data from the local client and performs parameter updating;

[0020] After parameter updating, the central server issues a new round of parameters V p ;

[0021] Step B5, each local client obtains the latest round of issued parameters V from the central server p , the real data plaintext is obtained after decryption using the private key;

[0022] Step B6, continuously iteratively update the local model parameters, and when the condition is met, the model training is completed:

[0023] Model = Train (TModel)

[0024] wherein Model is the actual model after training, Train(·) is the above iterative training process, and TModel is a model used to solve the demand Env migration.

[0025] Further, in step B1, the central server initializes the configuration parameters P and encrypts them using the public key:

[0026] V p = doEnc (P, PK)

[0027] wherein doEnc(·) is an encryption function, PK is the public key of the encryption process, and V p is the encrypted parameter data.

[0028] Further, the encryption method of the intermediate data in step B3 includes:

[0029]

[0030] wherein doEnc(·) is an encryption function, is the data encrypted by the i-th local client, PK is the public key of the encryption process, and C i is the intermediate data of the i-th local client.

[0031] Further, the execution parameter update method of step B4 includes:

[0032] V p = UD (V p ', V c )

[0033] wherein V p ' is the encrypted data generated by the central server in the last round, UD(·) is a data processing function of the central server, and V c is the total set of intermediate data received from all local clients,

[0034] Further, the method of obtaining the real data plaintext after decryption using the private key in step B5 includes:

[0035] V = doDec(V p , SK)

[0036] Wherein, V is the latest round of parameters obtained from the central server after decryption, doDec(·) is a decryption function, and SK is a private key; the local client updates the parameters in the local model according to the obtained data of this round:

[0037] C' i = UDLoc(V, C i )

[0038] Wherein, C' i It is the actual parameter of the model updated by the i-th local client in this round, UDLoc(·) is a local client parameter update function, V is the latest round of parameters obtained from the central server after decryption, and C i It is the intermediate data of the i-th local client.

[0039] The application also provides a data security sharing and decision system for power internet of things, which comprises: a power data preparation module, a cross-domain collaborative training module and a comprehensive decision module.

[0040] The power data preparation module is used for preparing the basic power internet of things data required for training the model, comprising a data acquisition sub-module and a data preprocessing sub-module; the data acquisition sub-module collects the privacy data of the power internet of things equipment in the local region, and the data preprocessing sub-module cleans and enhances the data to generate standard data that can be used for training;

[0041] The cross-domain collaborative training module is used for data security sharing and model training, comprising a local training sub-module and a cross-domain sharing sub-module; the local training module partially freezes the parameters migrated from the mature model and issued by the central server to perform localized model training, and the cross-domain sharing module securely shares the intermediate data generated during the training of the local client, and then issues the data after being uniformly processed by the central server, while the cross-domain sharing module ensures the security of the intermediate data of the local client by means of homomorphic encryption technology;

[0042] The comprehensive decision module is used for decision analysis on existing demands and gives decision suggestions, comprising a decision prediction sub-module and a comprehensive evaluation sub-module; the decision prediction sub-module uses the model output by the cross-domain collaborative training module to predict or classify the results, and the comprehensive evaluation sub-module gives decision suggestions by combining the model results after searching the expert database decision opinions.

[0043] Further, the power data preparation module is also used for preprocessing the collected power internet of things data based on the local client c i ∈C, and arranging the local power internet of things equipment data Data iData, data augmentation is carried out on the basis of the data, and the enhanced data DataUpdate at the i-th client is obtained i =RF(Data i );

[0044] The cross-domain collaborative training module also performs collaborative training based on the enhanced data, the local client and the trusted central server, iteratively trains the model using federated transfer learning and homomorphic encryption technology to securely share data, and generates an output model Model after completing the training target.

[0045] The comprehensive decision-making module uses the trained cross-region model to generate decision suggestions based on the output model Model generated by the cross-domain collaborative training module and the power Internet of Things demand Env.

[0046] The application also provides a server for realizing the data security sharing and decision-making method of the power Internet of Things, comprising:

[0047] The memory is used to store the computer program and the data security sharing and decision-making method of the power Internet of Things.

[0048] The processor is used to execute the computer program and the data security sharing and decision-making method of the power Internet of Things to realize the steps of the data security sharing and decision-making method of the power Internet of Things.

[0049] From the above technical solution, the application has the following advantages:

[0050] The data security sharing and decision-making method of the power Internet of Things provided by the application makes full use of the data recorded by intelligent devices in the field of the power Internet of Things, and uses data enhancement and other means to overcome the problem of small amount of data in special fields to a certain extent. At the same time, transfer learning is used to effectively shorten the model training time and reduce the cost.

[0051] Compared with the existing power decision system, the application uses the federated learning technology combined with homomorphic encryption, solves the problem of cross-domain data sharing, guarantees the security of the private data of the power Internet of Things, and uses the data of all regions in model training, so that the model is more comprehensive and can help top-level decision-making to a certain extent.

[0052] The application provides a decision-making framework after model training, generates a specified number of decision suggestions by combining the model output and the retrieval results of the expert experience library, and provides the suggestion credibility and analysis according to the probability value of the model output.

[0053] The system of this invention can be used for privacy data preprocessing in the power Internet of Things (IoT), cross-domain collaborative training and data sharing, comprehensive decision-making and suggestion analysis, ensuring the secure sharing of cross-regional data, and providing security for privacy data in various power IoT fields, thus providing decision-making convenience for professionals. Attached Figure Description

[0054] To more clearly illustrate the technical solution of the present invention, the accompanying drawings used in the description will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0055] Figure 1 A schematic diagram of a data security sharing and decision-making system for the power Internet of Things;

[0056] Figure 2 This is a schematic diagram of an embodiment of a data security sharing and decision-making system for the power Internet of Things.

[0057] Figure 3 A flowchart illustrating data security sharing and decision-making methods for the power Internet of Things;

[0058] Figure 4 A flowchart illustrating an implementation method for data security sharing and decision-making in the power Internet of Things. Detailed Implementation

[0059] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0060] The data security sharing and decision-making method and system for the power Internet of Things provided by this invention are executed based on a server, namely the central server of this invention.

[0061] For example, such as Figure 1 As shown, the system architecture can include multiple local clients, a network, and a server. The network is the medium used to provide a communication link between the local clients and the server. The network can include various connection types, such as wired or wireless communication links, or fiber optic cables, etc.

[0062] It should be understood that Figure 1The number of local clients, networks and servers in the system is merely illustrative. Any number of local clients, networks and servers can be provided as needed for implementation. For example, the server can be a server cluster composed of multiple servers, etc.

[0063] The local client can interact with the server through the network to receive or send messages, etc.

[0064] Figure 2 A structural schematic diagram of a data security sharing and decision system of a power Internet of Things suitable for use to implement the embodiments of the present disclosure is shown.

[0065] The server involved in the data security sharing and decision system of the power Internet of Things can include a central processing unit which can perform various appropriate actions and processes according to programs stored in a read-only memory or programs loaded from a storage section into a random access memory (RAM). In the RAM, various programs and data required for system operation are also stored. The CPU, ROM and RAM are connected to each other through a bus. An input / output (I / O) interface is also connected to the bus.

[0066] The server can also involve a communication section including a keyboard, a mouse, a liquid crystal display, a speaker and a network interface card including a LAN (Local Area Network) card, a modem, etc.

[0067] A mobile terminal implementing various embodiments of the present disclosure will now be described with reference to the accompanying drawings. In the following description, the suffix "module" or "part" used for an element is merely intended for facilitating description of the present disclosure, and does not have specific meaning or function. Therefore, a "module" and a "part" can be interchangeably used.

[0068] The system includes: a power data preparation module, a cross-domain collaborative training module, and a comprehensive decision module;

[0069] The power data preparation module: for preparing the basic power Internet of Things data required for training the model, including a data acquisition sub-module and a data preprocessing sub-module; the data acquisition sub-module collects the privacy data of the power Internet of Things equipment uniformly in the local domain, and the data preprocessing sub-module cleans and enhances the data to generate standard data that can be used for training;

[0070] The cross-domain collaborative training module is used for safe sharing of data and model training, and comprises a local training submodule and a cross-domain sharing submodule; the local training module partially freezes parameters migrated from a mature model and issued by a central server, and performs localized model training; the cross-domain sharing module safely shares intermediate data generated during local client training, and the central server uniformly processes and issues the intermediate data after safe sharing; and the cross-domain sharing module guarantees the security of the intermediate data of the local client by means of homomorphic encryption technology.

[0071] The comprehensive decision module is used for decision analysis on existing demands, and gives decision suggestions, and comprises a decision prediction submodule and a comprehensive evaluation submodule; the decision prediction submodule uses the model output by the cross-domain collaborative training module to perform result prediction or classification; and the comprehensive evaluation submodule gives decision suggestions after comprehensively evaluating the model results and searching the expert database for decision opinions.

[0072] In an embodiment of the present application, a possible embodiment will be described below to illustrate the specific implementation of the present application.

[0073] The power data preparation module is further used for preprocessing the collected power Internet of Things data based on the local client c i ∈C, collating the local power Internet of Things device data Data i ∈Data, performing data enhancement, and obtaining enhanced data DataUpdate i at the i-th client. i );

[0074] The cross-domain collaborative training module further performs collaborative training based on the enhanced data, the local client and the trusted central server, iteratively trains the model by using federated transfer learning and homomorphic encryption technology to safely share the data, and generates an output model Model after completing the training target.

[0075] The comprehensive decision module uses the output model Model generated by the cross-domain collaborative training module and the power Internet of Things demand Env to perform prediction or analysis by using the trained cross-region model, and generates decision suggestions.

[0076] The data security sharing and decision system for the power Internet of Things provided by the present application fully utilizes the data recorded by intelligent devices in the power Internet of Things field, and uses data enhancement and other means to overcome the problem of small amount of data in a special field to a certain extent, and uses transfer learning to effectively shorten the model training time and reduce the cost.

[0077] For the existing power decision system, general power perception is used, and the federated learning technology combined with homomorphic encryption is used in the present application, which solves the problem of cross-domain data sharing while ensuring the security of the power Internet of Things privacy data. At the same time, the data of each region are integrated and used in model training, so that the model is more comprehensive and can help top-level decision-making to a certain extent, ensuring the safe sharing of cross-regional data and providing security for various power Internet of Things privacy data, providing decision-making convenience for professionals.

[0078] Based on the above-mentioned data security sharing and decision system of the power Internet of Things, the present application also provides a data security sharing and decision method of the power Internet of Things, as shown in Figure 3 and Figure 4 , the method comprises the following steps:

[0079] Step A, the local client c i ∈C preprocesses the collected power Internet of Things data, organizes the local power Internet of Things device data Data i ∈Data, and then performs data enhancement to obtain the enhanced data DataUpdate at the i-th client. i =RF(Data i ), wherein C, Data are the total client set and the total power Internet of Things data set, and RF(·) is the data enhancement method.

[0080] Step B, based on the enhanced data, the local client and the trusted central server perform collaborative training, use federated transfer learning and homomorphic encryption technology to iteratively train the model while ensuring the security of data sharing, and generate an output model Model after completing the training target.

[0081] Specifically, step B1, for a specific requirement Env, a mature model TModel for transfer is selected. The central server initializes configuration parameters P and encrypts them using a public key:

[0082] V p =doEnc(P,PK)

[0083] Wherein, doEnc(·) is an encryption function, PK is the public key of the encryption process, and V p is the encrypted parameter data.

[0084] Step B2, each local client receives the model TModel selected by the central server to solve the requirement Env and a set of parameters V p initialized after the model. Freeze the first k layers of parameters of the model, load the parameters on Tmodel, and then randomly initialize the remaining layer parameters.

[0085] Step B3, the local client encrypts the intermediate data after one round of training:

[0086]

[0087] where doEnc is the encryption function, is the data encrypted by the i-th local client, PK is the public key of the encryption process, C i is the intermediate data of the i-th local client. Each local client transmits the encrypted data to the central server through a specially protected channel.

[0088] Step B4, the central server receives the encrypted intermediate data from the local client, and performs parameter update:

[0089] V p = UD(V p ) c

[0090] where V p is the encrypted data generated by the central server in the last round, UD(·) is the data processing function of the central server, V c is the total set of intermediate data received from all local clients,

[0091]

[0092] After the calculation is completed, the central server issues the next round of parameters V p .

[0093] Step B5, each local client obtains the latest round of issued parameters V p from the central server, and decrypts them using the private key to obtain the real data plaintext:

[0094] V = doDec(V p , SK)

[0095] where V is the decrypted latest round of issued parameters obtained from the central server, doDec(·) is the decryption function, and SK is the private key. The local client updates the parameters in the local model according to the data obtained in this round:

[0096] C i = UDLoc(V, C i )

[0097] where C i is the actual updated model parameter of the i-th local client in this round, UDLoc(·) is the local client parameter update function, V is the decrypted latest round of issued parameters obtained from the central server, and C i ​Intermediate data for the i-th local client.

[0098] Step B6, continuously iteratively update the local model parameters, and when the condition is met, the model training is completed:

[0099] Model = Train (TModel)

[0100] Where Model is the actual model after training, Train(·) is the above-mentioned iterative training process, and TModel is the model used to solve the demand Env migration.

[0101] Step C, based on the output model Model generated in step B and the power Internet of Things demand Env, the decision system uses the trained cross-regional model for prediction or analysis to generate specific decision suggestions. For the actual demand Env, the model trained by federated transfer learning is used to solve:

[0102] Solution = getSol (Env, Model, Sug)

[0103] Where Solution is a set of solutions, getSol(·) is an abstract function for obtaining solutions, and Sug is the auxiliary suggestion retrieved from the expert experience library.

[0104] Based on the above method, the present application uses federated learning and homomorphic encryption technology to realize the safe sharing of data, so that multiple parties can train the model in collaboration while ensuring that their regional data is not leaked, and then used in the field of power Internet of Things decision, and combined with the expert experience library to give comprehensive decision suggestions,

[0105] The data security sharing and decision method for the power Internet of Things provided by the present application makes full use of the data recorded by intelligent devices in the field of power Internet of Things, and uses data enhancement and other means to overcome the problem of small amount of data in special fields to a certain extent, and uses transfer learning to effectively shorten the model training time and reduce the cost.

[0106] Compared with the existing power decision system, the present application uses federated learning technology combined with homomorphic encryption, which solves the problem of cross-domain data sharing while ensuring the safety of the privacy data of the power Internet of Things, and at the same time, the data of each region is used in model training, making the model more comprehensive, which can help top-level decision-making to a certain extent.

[0107] The present application provides a decision framework after model training, generates a specified number of decision suggestions by combining the model output and the retrieval results of the expert experience library, and provides the suggestion credibility and analysis according to the probability value of the model output.

[0108] The system can be used for power Internet of Things privacy data preprocessing, cross-domain collaborative training and data sharing, comprehensive decision and suggestion analysis.

[0109] Based on the above provided power Internet of Things data security sharing decision method and system, those skilled in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be realized by electronic hardware, computer software or a combination of both, and in order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been described in the above description in general function. Whether the functions are executed in hardware or software mode depends on the specific application and design constraints of the technical solution. The skilled person can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0110] The block diagram of the power Internet of Things data security sharing decision method and system provided by the present application shown in the accompanying drawings is only a functional entity, which does not necessarily correspond to a physically independent entity. That is, these functional entities can be realized in the form of software, or in one or more hardware modules or integrated circuits, or in different networks and / or processor devices and / or microcontroller devices.

[0111] The server involved in the power Internet of Things data security sharing decision method and system provided by the present application is the units and algorithm steps of each example described in combination with the embodiments disclosed herein, which can be realized by electronic hardware, computer software or a combination of both, and in order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been described in the above description in general function. Whether the functions are executed in hardware or software mode depends on the specific application and design constraints of the technical solution. The skilled person can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0112] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application shall not be limited to these embodiments shown herein, but shall conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for data security sharing and decision-making of power Internet of Things, characterized in that, The method comprises: Step A, local client Preprocessing the collected power internet of things data, collating local power internet of things device data , performing data enhancement, obtaining enhanced data at the i-th client ; wherein C, Data is the total client set and the total power internet of things data set, is a method for data augmentation; Step B, based on the enhanced data, the local client and the trusted central server cooperatively train, iteratively train the model using federated transfer learning and homomorphic encryption technology to securely share the data, and generate an output model Model after completing the training target; Step B specifically comprises the following steps: Step B1, for specific needs Env , selecting a mature model to migrate TModel ; Step B2: Each local client receives the data selected by the central server to resolve the requirement. Env model TModel and a set of parameters after initialization ; freeze model before k Layer parameters, loading Tmodel After setting the parameters above, the parameters of the remaining layers are randomly initialized; Step B3, the local client encrypts the intermediate data after one round of training; Step B4, the central server receives the encrypted intermediate data from the local client and performs parameter updating; After the parameter update is completed, the central server issues a new round of parameters ; Step B5, each local client obtains the latest round of parameters from the central server The real data plaintext is obtained after decryption using the private key. Step B6, continuously update the local model parameters, and complete the model training after reaching the condition; wherein Model is the actual model trained, is the iterative training process, TModel is used to solve the demand Env the migrated model; Step C, based on the output model Model generated in step B and the power Internet of Things demand Env, use the trained cross-region model to make predictions or analysis and generate decision suggestions.

2. The data security sharing and decision method for the power Internet of Things according to claim 1, wherein in step C, the model trained by federated transfer learning is used to solve:

3. The data security sharing and decision method for the power Internet of Things according to claim 1, wherein in step B1, the central server initializes configuration parameters P and encrypts them using a public key; where Solution is a set of solutions, Sug is the abstract function to get solutions from the expert experience base.

4. The data security sharing and decision method for the power Internet of Things according to claim 1, wherein the encryption method of the intermediate data in step B3 comprises: PK wherein, is an encryption function, PK is a public key of an encryption process, is encrypted parameter data. The execution parameter updating method of step B4 comprises: The method of obtaining the real data plaintext after decryption using the private key in step B5 comprises: wherein, is an encryption function, is the i intermediate data encrypted by the SK is a public key of the encryption process, is the i intermediate data of the i-th local client. 5.The data security sharing and decision method of power internet of things according to claim 1, wherein, The system adopts the data security sharing and decision method for the power Internet of Things according to any one of claims 1 to 6; wherein, is the encrypted data generated by the central server for the previous round, is the data processing function of the central server, is the total set of intermediate data received from all local clients, . 6.The data security sharing and decision method of power internet of things according to claim 1, wherein, The system comprises: a power data preparation module, a cross-domain cooperative training module, and a comprehensive decision module; wherein, V is the latest round of parameters issued from the central server after decryption, is a decryption function, The power data preparation module: is used to prepare the basic power Internet of Things data required for training the model, comprising a data acquisition sub-module and a data preprocessing sub-module; the data acquisition sub-module collects the privacy data of the power Internet of Things devices in the local region, and the data preprocessing sub-module cleans and enhances the data to generate standard data that can be used for training; is a private key; the local client updates the parameters in the local model according to the data obtained in this round: in, For the first i The parameters of the model actually updated by each local client in this round. Functions for updating parameters for the local client. V This refers to the latest round of parameters obtained from the central server after decryption. For the first i Intermediate data from a local client.

7. A data security sharing and decision system of power internet of things, characterized in that, The cross-domain cooperative training module: is used for secure data sharing and model training, comprising a local training sub-module and a cross-domain sharing sub-module; the local training module partially freezes the parameters migrated from the mature model by the central server and performs local model training, and the cross-domain sharing module securely shares the intermediate data generated during the training of the local client, and then the central server processes and issues them, and the cross-domain sharing module ensures the security of the intermediate data of the local client by means of homomorphic encryption technology; The comprehensive decision module: is used for decision analysis of existing demands and gives decision suggestions, comprising a decision prediction sub-module and a comprehensive evaluation sub-module; the decision prediction sub-module uses the model output by the cross-domain cooperative training module to make result prediction or classification, and the comprehensive evaluation sub-module retrieves expert library decision opinions and gives decision suggestions by comprehensively considering the model results. ​ ​ ​ 8. The data security sharing and decision system of the power internet of things according to claim 7, characterized in that, The power data preparation module is also configured to prepare, based on the local client preprocess the collected power internet of things data, organize the local power internet of things device data , perform data enhancement, and obtain enhanced data at the i th client ; The cross-domain collaborative training module further collaborates with the local client and the trusted central server based on the enhanced data, iteratively trains the model using federated transfer learning and homomorphic encryption technology to securely share the data, and generates an output model Model after completing the training target; The comprehensive decision module uses the output model Model generated by the cross-domain collaborative training module and the power internet of things demand Env to predict or analyze using the trained cross-region model to generate a decision suggestion. 9.A server for implementing a data security sharing and decision method of a power internet of things, characterized in that, It comprises: a memory for storing a computer program and a power internet of things data security sharing and decision method; a processor for executing the computer program and the power internet of things data security sharing and decision method to realize the steps of the power internet of things data security sharing and decision method according to any one of claims 1 to 6.

Images