Power grid segmentation learning method and system based on V2G network

Abstract

The invention relates to a V2G network optimization technology, and discloses a V2G network-based power grid segmentation learning method, which comprises the following steps that: a client obtains real-time transaction information, and performs sensitivity grading on the real-time transaction information according to a sensitivity function to obtain a plurality of pieces of transaction information with graded sensitivity; obtaining sensitive attributes in the plurality of pieces of transaction information with hierarchical sensitivity, performing noise adding operation on the sensitive attributes, and storing the transaction information with noise added to a cloud storage server; the cloud storage server receives the transaction information added with the noise, and performs parameter optimization training on the LSTM-FCNN multi-task model to obtain a target LSTM-FCNN multi-task model; and the cloud storage server and the aggregator perform cooperative training on the target LSTM-FCNN multi-task model by using a segmentation learning method to obtain a power grid multi-task model based on the V2G network. The invention further provides a power grid segmentation learning system based on the V2G network. According to the invention, the optimization efficiency of the real-time transaction information in the V2G network can be improved, and the stability of the V2G network can be improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of V2G network optimization, and in particular relates to a power grid segmentation learning method and system based on a V2G network. Background Art

[0002] The existing V2G (Vehicle-to-Grid) network is a vehicle-grid interaction technology that allows electric vehicles (EVs) to send electricity back to the grid through a bidirectional charger. Electric vehicle payment transactions focus more on ensuring the security of user sensitive information while achieving data sharing, so most of the current payment mechanisms are based on blockchain.

[0003] Electric vehicles access the V2G network through dedicated parking facilities and provide auxiliary services to the smart grid under the control of local aggregators. Electric vehicles have three states: charging, discharging, and distributed discharging. Charging is initiated by electric vehicles, and electricity is transmitted from the smart grid to electric vehicles. Electric vehicles need to pay electricity bills to the smart grid. Discharging is initiated by the smart grid when the grid is overloaded during peak user demand hours. Electric vehicles can alleviate power shortages by discharging to the grid and receive economic rewards from the smart grid. Distributed discharging is initiated by one or more electric vehicles when the battery capacity is below a certain threshold. Unlike the charging state, electric vehicles with sufficient power will temporarily act as energy suppliers for electric vehicles with insufficient power, and the latter need to pay electricity bills to the supplier.

[0004] In the actual payment mechanism system, the account registration process mainly involves submitting the user-generated account to RA (Registered Account Authentication) for authentication and registration. The user creates a pair of keys and then sends them together with the identity information to RA. After checking the identity, RA replies to the user with a signature generated by RA's private key. Then RA will generate a legal account for the user, and each user can use RA's public key to verify the account. The payment execution process is actually a blockchain transaction, that is, verifying and writing the transaction into the global ledger. The payee can use a different account for each transaction, and then send the registered account and unit price to the payer. The payer checks the global ledger, verifies and calculates the total price, and sends the transaction id to the payee after payment. If the transaction is confirmed to be correct, the payee starts to supply power.

[0005] In the current V2G transaction process, due to the complexity of multi-party collaboration and information exchange, sensitive information such as user identity, transaction amount, charging location, etc. are easily attacked or accidentally leaked during transmission or storage, and there is a lack of effective hierarchical protection measures. In addition, in the process of information interaction and data processing, the existing V2G network usually adopts inefficient optimization strategies and privacy protection technologies, and the multi-party devices in the V2G network do not use segmentation learning methods for multi-task model collaborative training. It is easy to reduce the operating efficiency of real-time transaction information in the V2G network due to the low efficiency of model training in multi-party devices and the exposure of sensitive information, which in turn affects the stability of the entire network and the integrity of the life cycle. Summary of the invention

[0006] The present invention provides a V2G network-based power grid segmentation learning method and system, which can improve the optimization efficiency of real-time transaction information in the V2G network and improve the stability of the V2G network.

[0007] To achieve the above object, the present invention provides a V2G network-based power grid segmentation learning method, comprising:

[0008] The client obtains real-time transaction information from multiple information sources, and classifies the real-time transaction information according to the sensitivity function to obtain transaction information with multiple graded sensitivities, where the real-time transaction information includes the charging time t c , power consumption E, transaction amount A, electricity price fluctuation P, grid load L, user ID (UID), charging station number (SID), transaction ID (TID);

[0009] Acquire sensitive attributes in transaction information of multiple graded sensitivities, perform noise addition operations on the sensitive attributes, and store the transaction information after noise addition in a cloud storage server;

[0010] The cloud storage server receives the transaction information after adding noise, and uses the transaction information after adding noise to perform parameter optimization training on the LSTM-FCNN multi-task model to obtain the target LSTM-FCNN multi-task model;

[0011] The cloud storage server and aggregator use the segmentation learning method to collaboratively train the target LSTM-FCNN multi-task model to obtain a power grid multi-task model based on the V2G network.

[0012] Optionally, before the sensitivity classification of the real-time transaction information is performed according to the sensitivity function, the process further includes performing data standardization processing on the real-time transaction information and performing feature extraction, wherein the data standardization processing on the real-time transaction information and performing feature extraction includes:

[0013] Remove redundant and erroneous information from real-time transaction information, and use data standardization formulas to unify transaction information of different dimensions into the same dimension range;

[0014] The amount characteristics and pattern characteristics in the real-time transaction information are extracted, wherein the amount characteristics include the transaction amount characteristics and the electricity price fluctuation characteristics, and the pattern characteristics include the charging time characteristics, the power consumption characteristics, the grid load characteristics, the user charging frequency characteristics, the charging habit characteristics, the transaction time characteristics and the data collection frequency characteristics.

[0015] Optionally, the step of performing sensitivity grading on the real-time transaction information according to the sensitivity function to obtain transaction information of multiple graded sensitivities includes:

[0016] The sensitivity function is used to calculate the comprehensive sensitivity of each feature in the real-time transaction information;

[0017] Classify the sensitivity of real-time transaction information according to the comprehensive sensitivity and the preset membership function;

[0018] The real-time transaction information is graded according to the sensitivity classification to obtain transaction information of multiple graded sensitivities.

[0019] Optionally, the calculating the comprehensive sensitivity of each feature in the real-time transaction information by using a sensitivity function includes:

[0020] The comprehensive sensitivity S(R) of each feature is calculated using the following formula:

[0021]

[0022] Among them, w i is the feature weight of the i-th feature, i is the subscript index of the feature, S(f i ) is the sensitivity score, Where n is the total number of features, IG(f i ) is the contribution of the feature to the sensitivity classification, IG(f i )=H(S)-H(S|f i ), H(S) is the entropy of the comprehensive sensitivity, H(S|f i ) is the entropy of the comprehensive sensitivity of a given feature.

[0023] Optionally, the real-time transaction information is graded according to the sensitivity classification to obtain transaction information of multiple graded sensitivities, including:

[0024] Using sensitivity classification to classify real-time transaction information into low sensitivity information, medium sensitivity information, and high sensitivity information;

[0025] Storing low-sensitivity information in the blockchain;

[0026] Using a smart contract to set a first access permission for the medium-sensitivity information, and storing the medium-sensitivity information of the first access permission in the blockchain;

[0027] The highly sensitive information is generated with a highly sensitive zero-knowledge proof using zero-knowledge proof generation technology, and a second access permission is set using a smart contract, and the medium-sensitivity information of the second access permission is stored in the blockchain.

[0028] Optionally, performing a noise adding operation on the sensitive attribute and storing the transaction information after the noise is added to the cloud storage server includes:

[0029] Calculate the global sensitivity of sensitive attributes and set the privacy budget parameter ε;

[0030] The Laplace mechanism is used to add noise to sensitive attributes. The specific steps of adding noise to sensitive attributes using the Laplace mechanism are as follows:

[0031] For the original transaction information, generate noise Among them, Δf is the global sensitivity;

[0032] Calculate the protected transaction information x ′ =x+η;

[0033] The transaction information x after adding noise ′ Instead of the original transaction information x, the transaction information after adding noise is obtained, and the transaction information after adding noise is uploaded and stored in the cloud storage server.

[0034] Optionally, the cloud storage server and the aggregator collaboratively train the target LSTM-FCNN multi-task model using a segmentation learning method, including:

[0035] The cloud storage server updates the network parameters of the input layer and LSTM layer in the target hybrid LSTM-FCNN model using the gradient descent update algorithm, and sends the target hybrid LSTM-FCNN model to the aggregator;

[0036] The aggregator uses the gradient descent update algorithm to update the network parameters of the fully connected layer and the output layer in the target hybrid LSTM-FCNN model. After the model parameters are updated, a multi-task model of the power grid based on the V2G network is obtained.

[0037] In order to solve the above problems, the present invention also provides a system for a V2G network-based power grid segmentation learning method, the system comprising: a cloud storage server, an aggregator and a client.

[0038] Optionally, the aggregator determines sensitive attributes of the transaction information and uses a gradient descent update algorithm to update model parameters of a fully connected layer and an output layer in a target hybrid LSTM-FCNN model.

[0039] Optionally, the client performs two-way communication with the electric vehicle equipment in the V2G network, the client performs two-way communication with the cloud storage server, and the cloud storage server provides cloud services to the electric vehicle equipment in the V2G network.

[0040] The present invention performs sensitivity classification on real-time transaction information according to the sensitivity function, can realize graded protection of information with different sensitivities in the transaction information, can improve the management and processing efficiency of the transaction information, in addition, obtains sensitive attributes in the transaction information, and performs noise addition operation on the sensitive attributes, can realize the data security of the transaction information during the transmission process, and furthermore, trains the hybrid LSTM-FCNN model to obtain the multi-task model of the power grid based on the V2G network, can realize the optimization efficiency of the real-time transaction information in the V2G network, and improves the stability of the V2G network. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 A schematic diagram of a flow chart of a V2G network-based power grid segmentation learning method provided in one embodiment of the present invention;

[0042] Figure 2 A schematic diagram of information sensitivity classification of a V2G network-based power grid segmentation learning method provided in one embodiment of the present invention;

[0043] Figure 3 A schematic diagram of noise-added transmission of a V2G network-based power grid segmentation learning method provided in one embodiment of the present invention;

[0044] Figure 4 A schematic diagram of differential privacy learning of a cloud storage server for a V2G network-based power grid segmentation learning method provided in one embodiment of the present invention;

[0045] Figure 5 A schematic diagram of providing cloud services for a V2G network-based power grid segmentation learning method provided in one embodiment of the present invention;

[0046] Figure 6 A system diagram of a V2G network-based power grid segmentation learning device provided by an embodiment of the present invention.

[0047] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0048] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0049] The embodiment of the present application provides a method for learning power grid segmentation based on a V2G network. The execution subject of the method for learning power grid segmentation based on a V2G network includes but is not limited to at least one of the electronic devices such as a server, a terminal, etc. that can be configured to execute the method provided in the embodiment of the present application. In other words, the method for learning power grid segmentation based on a V2G network can be executed by software or hardware installed on a terminal device or a server device, and the software can be a blockchain platform. The server includes but is not limited to: a single server, a server cluster, a cloud server or a cloud server cluster, etc. The server can be an independent server, or it can be a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (CDN), and big data and artificial intelligence platforms.

[0050] Reference Figure 1 FIG. 1 is a flow chart of a V2G network-based power grid segmentation learning method according to an embodiment of the present invention. In this embodiment, the V2G network-based power grid segmentation learning method includes:

[0051] S1. The client obtains real-time transaction information from multiple information sources, and classifies the real-time transaction information according to the sensitivity function to obtain transaction information with multiple graded sensitivities, where the real-time transaction information includes the charging time t c , electricity consumption E, transaction amount A, electricity price fluctuation P, grid load L, user ID (UID), charging station number (SID), transaction ID (TID).

[0052] In the embodiment of the present invention, the information source includes but is not limited to electric vehicle sensors, power grid load monitoring systems, environmental data and other devices that can collect transaction information.

[0053] In the embodiment of the present invention, the sensitivity function refers to a function that defines a sensitivity score for a feature, and the score range is [0, 1], where 1 represents high sensitivity. i is a feature, x i is the eigenvalue.

[0054] When f i When it is of ID type, if f i =UID, then the sensitivity function is S(f i ,x i )=1; otherwise the sensitivity function S(f i ,x i )=0.2.

[0055] When f i When the amount type is set, set the sensitivity function S(f i ,x i )=0.9

[0056] When f i When the time type is set, set the sensitivity function S(f i ,x i )=0.8

[0057] When f i When it is a power resource type (electricity price, power consumption, load), set the sensitivity function Among them, f max is the maximum eigenvalue, f min is the minimum eigenvalue.

[0058] As an embodiment of the present invention, before the sensitivity of the real-time transaction information is graded according to the sensitivity function, the method further includes performing data standardization processing on the real-time transaction information and performing feature extraction.

[0059] Furthermore, data standardization and feature extraction are performed on real-time transaction information, including:

[0060] Remove redundant and erroneous information from real-time transaction information, and use data standardization formulas to unify transaction information of different dimensions into the same dimension range;

[0061] The amount characteristics and pattern characteristics in the real-time transaction information are extracted, wherein the amount characteristics include the transaction amount characteristics and the electricity price fluctuation characteristics, and the pattern characteristics include the charging time characteristics, the power consumption characteristics, the grid load characteristics, the user charging frequency characteristics, the charging habit characteristics, the transaction time characteristics and the data collection frequency characteristics.

[0062] As an embodiment of the present invention, the real-time transaction information is graded according to the sensitivity function to obtain transaction information with multiple graded sensitivities, including:

[0063] The sensitivity function is used to calculate the comprehensive sensitivity of each feature in the real-time transaction information;

[0064] Classify the sensitivity of real-time transaction information according to the comprehensive sensitivity and the preset membership function;

[0065] The real-time transaction information is graded according to the sensitivity classification to obtain transaction information of multiple graded sensitivities.

[0066] Furthermore, the membership function preset in the embodiment of the present invention adopts the following membership function:

[0067]

[0068] Among them, threshold a is the upper limit of low sensitivity, threshold b is the core value of medium sensitivity, and threshold c is the lower limit of high sensitivity.

[0069] Furthermore, the sensitivity function is used to calculate the comprehensive sensitivity of each feature in the real-time transaction information, including:

[0070] The comprehensive sensitivity S(R) of each feature is calculated using the following formula:

[0071]

[0072] Among them, w i is the feature weight of the i-th feature, i is the subscript index of the feature, S(f i ) is the sensitivity score, Where n is the total number of features, IG(f i ) is the contribution of the feature to the sensitivity classification, IG(f i )=H(S)-H(S|f i ), H(S) is the entropy of the comprehensive sensitivity, H(S|f i ) is the entropy of the comprehensive sensitivity of a given feature.

[0073] The embodiment of the present invention further includes calculating the information entropy of each feature, using the following formula:

[0074]

[0075] Among them, H(f i ) is the characteristic information entropy, p k is the feature f i The probability of taking the value k.

[0076] Furthermore, the real-time transaction information is graded according to the sensitivity classification to obtain transaction information of multiple graded sensitivities, including:

[0077] Using sensitivity classification to classify real-time transaction information into low sensitivity information, medium sensitivity information, and high sensitivity information;

[0078] Storing low-sensitivity information in the blockchain;

[0079] Using a smart contract to set a first access permission for the medium-sensitivity information, and storing the medium-sensitivity information of the first access permission in the blockchain;

[0080] The highly sensitive information is generated with a highly sensitive zero-knowledge proof using zero-knowledge proof generation technology, and a second access permission is set using a smart contract, and the medium-sensitivity information of the second access permission is stored in the blockchain.

[0081] S2. Acquire sensitive attributes in transaction information of multiple graded sensitivities, perform noise adding operations on the sensitive attributes, and store the transaction information after noise addition in a cloud storage server.

[0082] In the embodiment of the present invention, the sensitive attributes refer to attributes such as charging behavior pattern, charging peak time, grid load, and price change correlation.

[0083] As an embodiment of the present invention, performing a noise adding operation on a sensitive attribute and storing the transaction information after the noise is added to a cloud storage server includes:

[0084] Calculate the global sensitivity of sensitive attributes and set the privacy budget parameter ε;

[0085] The Laplace mechanism is used to add noise to sensitive attributes. The specific steps of adding noise to sensitive attributes using the Laplace mechanism are as follows:

[0086] For the original transaction information, generate noise Where Δf is the global sensitivity;

[0087] Calculate the protected transaction information x ′ =x+η;

[0088] The transaction information x after adding noise ′ Instead of the original transaction information x, the transaction information after adding noise is obtained, and the transaction information after adding noise is uploaded and stored in the cloud storage server.

[0089] Exemplarily, the global sensitivity of the calculated sensitive attribute can be obtained by using the following implementation method:

[0090] For example, D and D′ are adjacent data sets;

[0091] Method 1: Proportional function: where count A (D) The number of records that meet condition A, count B (D) is the number of records that meet condition B. Sensitivity of the whole play where c A =count A (D′), c B =count B (D′), the global sensitivity can be obtained by algebraic derivation

[0092] Method 2: Maximum value function: Global sensitivity Δf = max|max(D)-max(D′)| = Δx;

[0093] Method 3: Median or percentile function: f(D) = mid(D), global sensitivity Where Δx is the maximum variation range of a single record, and n is the size of the data set;

[0094] Method 4: Distance function: Global sensitivity Δf = Δd, where Δd is the maximum distance change;

[0095] Method 5: Complex aggregation function: f(D) = ∑ x∈D g(x), g(x) is a nonlinear function of a single record x, and the global sensitivity

[0096] In the embodiment of the present invention, the privacy budget parameter ε can control the strength of privacy protection, and a smaller ε value provides stronger privacy protection.

[0097] S3. The cloud storage server receives the transaction information after adding noise, and uses the transaction information after adding noise to perform parameter optimization training on the LSTM-FCNN multi-task model to obtain the target LSTM-FCNN multi-task model.

[0098] In an embodiment of the present invention, the LSTM-FCNN multi-task model is a deep learning model that combines a long short-term memory network (LSTM) and a fully convolutional neural network (FCN). The LSTM-FCNN multi-task model includes an input layer, 2 LSTM layers, 2 fully connected layers, and an output layer, which can realize functions such as charging demand prediction, power grid load prediction, charging behavior classification, and power grid status classification.

[0099] The embodiment of the present invention can use an adaptive learning rate Adam optimizer to perform parameter optimization training on the LSTM-FCNN multi-task model, wherein the adaptive learning rate Adam optimizer calculates its own learning rate for each parameter, so that Adam can have different learning speeds on different parameters, thereby accelerating the convergence speed of the LSTM-FCNN multi-task model.

[0100] S4, the cloud storage server and the aggregator use the segmentation learning method to collaboratively train the target LSTM-FCNN multi-task model to obtain a power grid multi-task model based on the V2G network.

[0101] As an embodiment of the present invention, the cloud storage server and the aggregator use a segmentation learning method to collaboratively train the target LSTM-FCNN multi-task model, including:

[0102] The cloud storage server updates the network parameters of the input layer and LSTM layer in the target hybrid LSTM-FCNN model using the gradient descent update algorithm, and sends the target hybrid LSTM-FCNN model to the aggregator;

[0103] The aggregator uses the gradient descent update algorithm to update the network parameters of the fully connected layer and the output layer in the target hybrid LSTM-FCNN model. After the model parameters are updated, a multi-task model of the power grid based on the V2G network is obtained.

[0104] The embodiment of the present invention uses a gradient descent update algorithm to update the input layer and LSTM layer parameters in the target hybrid LSTM-FCNN model, and the following implementation steps can be used:

[0105] Gradient clipping: Clip the gradient of each sample to a preset norm range to prevent a single sample from having too much influence on the gradient. The preset norm range is a pre-set fixed value, denoted as O. The appropriate O value can be selected through experiments and cross-validation.

[0106] Add noise: Add noise ζ~N(0,σ 2 I), where σ controls the noise intensity and I is the identity matrix.

[0107] Exemplarily, the input layer and LSTM layer parameter updates in the target hybrid LSTM-FCNN model can be updated using the following gradient descent update algorithm:

[0108] Let W c is the weight of the model input layer and LSTM layer parameters in the client, b c The bias of the model input layer and LSTM layer parameters in the client. c is the noisy gradient received by the client. 1 is the first learning rate. The update formula is: in is the weight gradient, is the bias gradient.

[0109] Exemplarily, the aggregator uses the gradient descent update algorithm to update the model parameters of the fully connected layer and the output layer in the target hybrid LSTM-FCNN model, which can be updated using the following gradient descent update algorithm:

[0110] Let W s is the weight of the fully connected layer and output layer of the model in the cloud storage server, b s is the bias of the fully connected layer and output layer of the model in the cloud storage server. s is the gradient calculated by the cloud storage server. η 2 is the second learning rate. U is the gradient clipping threshold, ζ is the added noise;

[0111] The update formula is as follows:

[0112] Among them, clip(g s ,U) represents the gradient g s The norm of is limited to within U, is the weight gradient after clipping and noise addition, is the bias gradient after clipping and noise addition.

[0113] In the embodiment of the present invention, the input layer and the LSTM layer are responsible for preliminary feature extraction and time series data processing, and their functions include: feature extraction, dimension compression, and data transmission.

[0114] In the embodiment of the present invention, the fully connected layer and the output layer are responsible for further feature fusion and final prediction, and their functions include: feature fusion, final prediction, loss calculation and gradient transfer.

[0115] In the embodiment of the present invention, after distributing the target hybrid LSTM-FCNN model to the aggregator, and the aggregator using the gradient descent algorithm to update the model parameters of the fully connected layer and the output layer in the target hybrid LSTM-FCNN model, the aggregator also uses the cloud storage server to receive the model parameters of the fully connected layer and the output layer, and uses the power grid multi-task model based on the V2G network to complete the prediction output of the model. The following implementation steps can be adopted:

[0116] Fully connected layer 1 → Fully connected layer 2 → Output layer → Loss calculation (the second half processes the intermediate representation to generate the predicted output. Depending on the task type, the difference between the predicted output and the true label is calculated using the selected loss function)

[0117] Calculate the loss function and calculate the gradient through back propagation (calculate the gradient based on the loss function, perform gradient clipping, add noise, and finally update the model parameters)

[0118] Calculate the loss function to determine the loss function based on the task type:

[0119] 1. Loss function of regression tasks (such as charging demand prediction, grid load prediction): Where N is the number of samples, y z is the true value of the zth sample, is the predicted value of the zth sample.

[0120] 2. Loss function of classification tasks (such as charging behavior classification, grid status classification):

[0121] 1. Second Classification:

[0122] 2. Multi-classification:

[0123] Where C is the number of categories, y z,cis the true label of the zth sample in the cth class (usually 0 or 1), is the probability of the cth class predicted by the model.

[0124] In the embodiment of the present invention, the multi-task model of the power grid based on the V2G network can realize functions such as charging demand prediction, power grid load prediction, charging behavior classification and power grid status classification.

[0125] After obtaining the multi-task model of the power grid based on the V2G network, the embodiment of the present invention further includes cloud service application of the multi-task model of the power grid based on the V2G network, which is implemented by the following steps:

[0126] Step 1: Cloud service platform architecture design.

[0127] Multi-layer architecture design:

[0128] Data layer: stores differentially private data from trams and smart grids.

[0129] Service layer: provides data processing, analysis, and machine learning model training functions.

[0130] Interface layer: Provides a unified API interface for trams and power grids, supporting data uploading, model calling and result acquisition.

[0131] Security layer: Integrates differential privacy, authentication, and permission management mechanisms to ensure the security of data and services.

[0132] Step 2: Service interface and API design.

[0133] Unified interface standards: Design RESTful APIs to provide standardized service calling methods; define clear request and response formats, and support multiple data types and operations.

[0134] In the embodiment of the present invention, a RESTful API is designed: HTTP methods (GET, POST, PUT, DELETE); requests use the JSON format, including necessary parameters and data; responses use a unified JSON structure, including a status code, message, and data.

[0135] Authentication and authorization: Integrate the OAuth 2.0 authentication mechanism to ensure that only authorized trams and grid devices can access cloud services.

[0136] Implementation steps: 1. Set up the authorization server; 2. Client authentication (obtain access token); 3. Resource access; 4. Token verification.

[0137] Service catalog and documentation: Provide detailed API documentation and usage guides to facilitate developers and devices to integrate cloud services; establish a service catalog to list available model services, data analysis services, etc.

[0138] Step 3: Cloud service application for electric vehicles and power grid

[0139] Tram side application:

[0140] (1) Smart charging recommendation:

[0141] (i) The electric car obtains recommendations for the best charging time and location by calling the cloud storage server model service.

[0142] (ii) Provide personalized charging solutions based on real-time electricity prices, charging station load and user habits.

[0143] (2) Energy consumption optimization:

[0144] The cloud storage server is used to analyze the results, optimize the energy consumption management strategy of the electric vehicle, and improve the battery usage efficiency.

[0145] Grid-side applications:

[0146] (1) Load forecasting and scheduling:

[0147] (i) The power grid calls the cloud service to obtain the forecast results of electric vehicle charging demand and optimize power dispatch.

[0148] (ii) Sense charging peak in advance to balance grid load and avoid overload.

[0149] (2) Demand response management:

[0150] (i) Based on the analysis of cloud storage servers, a demand response strategy is implemented to guide electric vehicles to charge during off-peak hours.

[0151] (ii) Provide incentive mechanisms to encourage users to participate in grid regulation and improve overall energy efficiency.

[0152] Step 4: User feedback and service improvement

[0153] User feedback collection:

[0154] (1) Provide channels (such as API feedback interface, customer service support) to collect opinions and suggestions from tram and power grid users.

[0155] (2) Classify and analyze feedback to identify problems and areas for improvement in services.

[0156] Service optimization:

[0157] (1) Continuously optimize the functions and performance of cloud services based on user feedback and data analysis results.

[0158] (2) Introduce new technologies and algorithms to improve the intelligence and convenience of services.

[0159] Community and Ecosystem Construction:

[0160] (1) Establish a developer community and encourage third-party developers to develop new applications based on cloud services.

[0161] (2) Open some data and model interfaces to promote the prosperity and development of the V2G network ecosystem.

[0162] The present invention performs sensitivity classification on real-time transaction information according to the sensitivity function, can realize graded protection of information with different sensitivities in the transaction information, can improve the management and processing efficiency of the transaction information, in addition, obtains sensitive attributes in the transaction information, and performs noise addition operation on the sensitive attributes, can realize the data security of the transaction information during the transmission process, and furthermore, trains the hybrid LSTM-FCNN model to obtain the multi-task model of the power grid based on the V2G network, can realize the optimization efficiency of the real-time transaction information in the V2G network, and improves the stability of the V2G network.

[0163] like Figure 2 , which is a schematic diagram of information sensitivity classification of a V2G network-based power grid segmentation learning method provided in one embodiment of the present invention.

[0164] like Figure 3 , which is a schematic diagram of noise-added transmission of a V2G network-based power grid segmentation learning method provided in one embodiment of the present invention.

[0165] like Figure 4 As shown, it is a schematic diagram of differential privacy learning of a cloud storage server of a V2G network-based power grid segmentation learning method provided by an embodiment of the present invention.

[0166] like Figure 5 As shown, it is a schematic diagram of providing cloud services for a V2G network-based power grid segmentation learning method provided in one embodiment of the present invention.

[0167] like Figure 6 , which is a system diagram of a V2G network-based power grid segmentation learning device provided by an embodiment of the present invention.

[0168] As an embodiment of the present invention, the system includes: a cloud storage server, an aggregator and a client.

[0169] Furthermore, the aggregator determines sensitive attributes of the transaction information and uses a gradient descent update algorithm to update model parameters of the fully connected layer and the output layer in the target hybrid LSTM-FCNN model.

[0170] Furthermore, the client performs two-way communication with the electric vehicle equipment in the V2G network, the client performs two-way communication with the cloud storage server, and the cloud storage server provides cloud services to the electric vehicle equipment in the V2G network.

[0171] It should be understood that the embodiment is for illustration only and the scope of the patent application is not limited to this structure.

[0172] It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0173] Therefore, no matter from which point of view, the embodiments should be regarded as illustrative and non-restrictive, and the scope of the present invention is limited by the appended claims rather than the above description, so it is intended that all changes falling within the meaning and scope of the equivalent elements of the claims are included in the present invention. Any attached figure mark in the claims should not be regarded as limiting the claims involved.

[0174] The blockchain referred to in this invention is a new application mode of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanism, encryption algorithm, etc. Blockchain is essentially a decentralized database, a string of data blocks generated by cryptographic methods. Each data block contains a batch of network transaction information, which is used to verify the validity of its information (anti-counterfeiting) and generate the next block. Blockchain can include the underlying blockchain platform, platform product service layer, and application service layer.

[0175] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Among them, artificial intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0176] In addition, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices stated in the system claim can also be implemented by one unit or device through software or hardware. The words first, second, etc. are used to indicate names, and do not indicate any particular order.

[0177] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. A power grid segmentation learning method based on V2G network, characterized in that: The method comprises: The client obtains real-time transaction information from multiple information sources, and classifies the real-time transaction information according to the sensitivity function to obtain transaction information with multiple graded sensitivities, where the real-time transaction information includes the charging time t c , power consumption E, transaction amount A, electricity price fluctuation P, grid load L, user ID (UID), charging station number (SID), transaction ID (TID); Acquire sensitive attributes in transaction information of multiple graded sensitivities, perform noise addition operations on the sensitive attributes, and store the transaction information after noise addition in a cloud storage server; The cloud storage server receives the transaction information after adding noise, and uses the transaction information after adding noise to perform parameter optimization training on the LSTM-FCNN multi-task model to obtain the target LSTM-FCNN multi-task model; The cloud storage server and aggregator use the segmentation learning method to collaboratively train the target LSTM-FCNN multi-task model to obtain a power grid multi-task model based on the V2G network.

2. The V2G network-based power grid segmentation learning method according to claim 1, characterized in that: Before the sensitivity classification of the real-time transaction information is performed according to the sensitivity function, the method further includes performing data standardization processing on the real-time transaction information and performing feature extraction, wherein the data standardization processing on the real-time transaction information and performing feature extraction include: Remove redundant and erroneous information from real-time transaction information, and use data standardization formulas to unify transaction information of different dimensions into the same dimension range; The amount characteristics and pattern characteristics in the real-time transaction information are extracted, wherein the amount characteristics include the transaction amount characteristics and the electricity price fluctuation characteristics, and the pattern characteristics include the charging time characteristics, the power consumption characteristics, the grid load characteristics, the user charging frequency characteristics, the charging habit characteristics, the transaction time characteristics and the data collection frequency characteristics.

3. The V2G network-based power grid segmentation learning method according to claim 1, characterized in that: The step of performing sensitivity grading on the real-time transaction information according to the sensitivity function to obtain transaction information with multiple graded sensitivities includes: The sensitivity function is used to calculate the comprehensive sensitivity of each feature in the real-time transaction information; Classify the sensitivity of real-time transaction information according to the comprehensive sensitivity and the preset membership function; The real-time transaction information is graded according to the sensitivity classification to obtain transaction information of multiple graded sensitivities.

4. The V2G network-based power grid segmentation learning method according to claim 3, characterized in that: The method of calculating the comprehensive sensitivity of each feature in the real-time transaction information using the sensitivity function includes: The comprehensive sensitivity S(R) of each feature is calculated using the following formula: Among them, w i is the feature weight of the i-th feature, i is the subscript index of the feature, S(f i ) is the sensitivity score, Where n is the total number of features, IG(f i ) is the contribution of the feature to the sensitivity classification, IG(f i )=H(S)-H(S|f i ), H(S) is the entropy of the comprehensive sensitivity, H(S|f i ) is the entropy of the comprehensive sensitivity of a given feature.

5. The V2G network-based power grid segmentation learning method according to claim 3, characterized in that: The real-time transaction information is graded according to the sensitivity classification to obtain transaction information of multiple graded sensitivities, including: Using sensitivity classification to classify real-time transaction information into low sensitivity information, medium sensitivity information, and high sensitivity information; Storing low-sensitivity information in the blockchain; Using a smart contract to set a first access permission for the medium-sensitivity information, and storing the medium-sensitivity information of the first access permission in the blockchain; The highly sensitive information is generated into a highly sensitive zero-knowledge proof using zero-knowledge proof generation technology, and the second access permission is set using a smart contract, and the medium-sensitivity information of the second access permission is stored in the blockchain.

6. The V2G network-based power grid segmentation learning method according to claim 1, characterized in that: The performing of the noise adding operation on the sensitive attribute and storing the transaction information after the noise is added to the cloud storage server includes: Calculate the global sensitivity of sensitive attributes and set the privacy budget parameter ε; The Laplace mechanism is used to add noise to sensitive attributes. The specific steps of adding noise to sensitive attributes using the Laplace mechanism are as follows: For the original transaction information, generate noise Among them, Δf is the global sensitivity; Calculate the protected transaction information x ′ =x+η; The transaction information x after adding noise ′ Instead of the original transaction information x, the transaction information after adding noise is obtained, and the transaction information after adding noise is uploaded and stored in the cloud storage server.

7. The V2G network-based power grid segmentation learning method according to claim 1, characterized in that: The cloud storage server and the aggregator collaboratively train the target LSTM-FCNN multi-task model using a segmentation learning method, including: The cloud storage server updates the network parameters of the input layer and LSTM layer in the target hybrid LSTM-FCNN model using the gradient descent update algorithm, and sends the target hybrid LSTM-FCNN model to the aggregator; The aggregator uses the gradient descent update algorithm to update the network parameters of the fully connected layer and the output layer in the target hybrid LSTM-FCNN model. After the model parameters are updated, a multi-task model of the power grid based on the V2G network is obtained.

8. A system for implementing the V2G network-based power grid segmentation learning method according to any one of claims 1 to 7, the system comprising: Cloud storage servers, aggregators, and clients.

9. The system of the V2G network-based power grid segmentation learning method according to claim 8, comprising: The aggregator determines the sensitive attributes of the transaction information and uses the gradient descent algorithm to update the model parameters of the fully connected layer and the output layer in the target hybrid LSTM-FCNN model.

10. The system of the V2G network-based power grid segmentation learning method according to claim 8, comprising: The client performs two-way communication with the electric vehicle equipment in the V2G network, the client performs two-way communication with the cloud storage server, and the cloud storage server provides cloud services to the electric vehicle equipment in the V2G network.

Images