Virtual power plant user data identification method
By using a combination method of data preprocessing, graph neural network and reinforcement learning model in virtual power plants, the virtual power plant user data is solved, and the problems of low data processing efficiency, low accuracy and data security risks in the existing technology are solved, and efficient and accurate data identification and protection are achieved.
Patent Information
- Application Number
- CN202411685601.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-23
- Publication Date
- 2025-03-18
AI Technical Summary
The existing virtual power plant user data processing and identification methods are inefficient and have low accuracy, and cannot deeply explore data associations, which poses a risk of data leakage and abuse.
The combination method of data preprocessing, graph neural network model and reinforcement learning model is adopted to build a user data graph network, extract nodes and edge features through graph neural network, combine reinforcement learning model to identify the sensitivity level of user data, and perform classification processing to ensure data privacy.
It realizes accurate identification and protection of user data of virtual power plants, improves data processing efficiency and accuracy, reduces the risk of data leakage and abuse, and ensures the privacy and security of user data.
Smart Images
Figure CN119646436A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of power plant user data identification, and more specifically, to a method for identifying user data of a virtual power plant. Background Art
[0002] At present, with the vigorous development of smart grid and virtual power plant technologies, the processing and protection of virtual power plant user data has become a key issue that needs to be solved urgently. Under the current technological background, virtual power plants need to process and analyze massive amounts of user data to optimize the allocation and scheduling of power resources.
[0003] However, existing data processing and recognition methods show obvious shortcomings when faced with these large-scale and complex data sets.
[0004] Traditional data processing methods have problems of low efficiency and low accuracy when processing virtual power plant user data. At the same time, these methods are often unable to deeply explore the correlation between data, which limits the virtual power plant's understanding and analysis of user behavior patterns. In addition, existing user data identification methods also mainly rely on simple rule matching or statistical models. These methods have large errors in identifying the sensitivity level of user data, which can easily lead to risks such as data leakage or abuse.
[0005] Therefore, in view of the above situation, the present invention provides a method for identifying user data of a virtual power plant. Summary of the invention
[0006] In order to overcome the above-mentioned defects of the prior art, the present invention provides a virtual power plant user data identification method to solve the problems raised in the above-mentioned background technology.
[0007] To achieve the above object, the present invention provides the following technical solution: a method for identifying user data of a virtual power plant, the method comprising the following steps:
[0008] S1. Data preprocessing, for collecting virtual power plant user data, cleaning and standardizing the data, and building a user data graph network;
[0009] S2. Construct a graph neural network model, wherein the graph neural network model includes an input layer, a hidden layer, and an output layer. The input layer is used to receive the user data graph network as input and extract the features of nodes and edges. The hidden layer adopts a graph convolutional neural network (GCN) or a graph attention network (GAT) graph neural network model. The output layer is used to output the sensitivity level prediction result of the user data.
[0010] S3. Build a reinforcement learning model and design a reinforcement learning framework, including the agent, environment, state space, action space, and reward function. The output of the graph neural network model is used as the input of the reinforcement learning agent. The agent selects actions based on the current state, that is, adjusts the parameters of the graph neural network model. The environment gives feedback based on the agent's actions, that is, rewards or penalties, which are used to update the agent's strategy. Through continuous iterative training, the agent learns the optimal strategy and can accurately identify the sensitivity level of user data.
[0011] S4. Sensitive data identification and protection. Use trained graph neural networks and reinforcement learning models to identify the sensitivity level of user data, and classify and process user data based on the identification results, such as encrypted storage and access control, to ensure the privacy and security of user data. At the same time, monitor the changes in the sensitivity level of user data in real time and take corresponding measures in a timely manner to prevent data leakage and abuse.
[0012] Preferably, in step S2, the mathematical expression of the graph neural network model using a graph convolutional neural network (GCN) or a graph attention network (GAT) is: Among them, H (l) represents the node feature matrix of the lth layer, represents the normalized adjacency matrix, W (l) represents the weight matrix of the lth layer, and σ represents the activation function.
[0013] Preferably, in step S3, the policy update adopts a deep deterministic policy gradient (DDPG) algorithm, and the specific formula is: Among them, θ k represents the policy parameter of the kth iteration, α represents the learning rate, represents the objective function j with respect to parameter θ k The gradient of j(θ k ) represents the objective function.
[0014] Preferably, in step S1, the data preprocessing also includes performing feature selection and feature dimension reduction processing on the virtual power plant user data to improve the efficiency and accuracy of data processing.
[0015] Preferably, in step S2, when the graph neural network model is constructed, the graph neural network architecture also includes introducing an attention mechanism to improve the learning ability of node and edge features in the user data graph network.
[0016] Preferably, in step S4, the sensitive data identification and protection also includes anonymizing the user data to ensure the privacy and security of the user data.
[0017] Preferably, it also includes a virtual power plant user data identification device, including a data preprocessing module, a graph neural network model module, a reinforcement learning model module and a sensitive data identification and protection module, wherein;
[0018] A data preprocessing module, used to preprocess the virtual power plant user data collected in step S1;
[0019] The graph neural network model module is used to build the graph neural network model in step S2 and contains submodules for executing the GCN or GAT algorithm;
[0020] The reinforcement learning model module is used to build the reinforcement learning model in step S3 and contains a submodule for executing the DDPG algorithm;
[0021] The sensitive data identification and protection module is used to perform sensitive data identification and protection in step S4.
[0022] Preferably, it also includes a feature selection and feature dimension reduction module for performing feature selection and feature dimension reduction processing on virtual power plant user data.
[0023] Preferably, the graph neural network model module further includes an attention mechanism module for improving the learning ability of node and edge features in the user data graph network;
[0024] The reinforcement learning model module also includes a deep deterministic policy gradient algorithm module, which is used to improve the learning efficiency and stability of the agent strategy;
[0025] The sensitive data identification and protection module also includes an anonymization processing module for anonymizing user data to further protect the privacy and security of user data.
[0026] Technical effects and advantages of the present invention:
[0027] The present invention achieves accurate identification of virtual power plant user data by combining multiple steps of data preprocessing, graph neural network model, reinforcement learning model, and sensitive data identification and protection. It can deeply explore the correlation between user data, build a user data graph network, and extract the features of nodes and edges through the graph neural network model, so as to accurately predict the sensitivity level of user data. It not only improves the efficiency and accuracy of data processing, but also provides a solid foundation for subsequent data protection. BRIEF DESCRIPTION OF THE DRAWINGS
[0028] Figure 1 It is the overall flow chart of the present invention. DETAILED DESCRIPTION
[0029] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0030] As attached Figure 1 As shown, the present invention provides a method for identifying user data of a virtual power plant, which specifically includes the following steps:
[0031] S1. Data preprocessing is used to collect virtual power plant user data, clean and standardize the data, and build a user data graph network. The data preprocessing also includes feature selection and feature dimensionality reduction of virtual power plant user data to improve the efficiency and accuracy of data processing.
[0032] Specifically, in this step, user data is collected from the virtual power plant system, including but not limited to electricity consumption, electricity consumption time, user location, and device information. The collected data is cleaned to remove duplicate, erroneous, and invalid data, and the data is converted into a unified format and dimension for subsequent processing. Based on the association between user data, a user data graph network is constructed, in which nodes represent users or devices, and edges represent the relationship between users or devices. In order to improve the efficiency and accuracy of data processing, feature selection and feature dimensionality reduction are performed on the data to retain the most representative features.
[0033] S2. Construct a graph neural network model, which includes an input layer, a hidden layer and an output layer. The input layer is used to receive the user data graph network as input and extract the features of nodes and edges. The hidden layer adopts a graph convolutional neural network (GCN) or a graph attention network (GAT) graph neural network model. The output layer is used to output the sensitivity level prediction result of the user data. When constructing the graph neural network model, the graph neural network architecture also includes introducing an attention mechanism to improve the learning ability of the node and edge features in the user data graph network.
[0034] Specifically, in this step:
[0035] The input layer is used to receive the user data graph network as input and extract the features of nodes and edges;
[0036] The hidden layer uses a graph convolutional neural network (GCN) or a graph attention network (GAT) graph neural network model, and the mathematical expression is: Among them, H (l) represents the node feature matrix of the lth layer, represents the normalized adjacency matrix, W (l)represents the weight matrix of the lth layer, σ represents the activation function;
[0037] The output layer is used to output the sensitivity level prediction results of user data;
[0038] The attention mechanism is introduced to improve the learning ability of node and edge features in the user data graph network.
[0039] S3. Build a reinforcement learning model and design a reinforcement learning framework, including the agent, environment, state space, action space and reward function. The output of the graph neural network model is used as the input of the reinforcement learning agent. The agent selects actions based on the current state, that is, adjusts the parameters of the graph neural network model. The environment gives feedback based on the agent's actions, that is, rewards or penalties, which are used to update the agent's strategy. Through continuous iterative training, the agent learns the optimal strategy and can accurately identify the sensitivity level of user data.
[0040] Specifically, in this step, specifically:
[0041] Design a reinforcement learning framework including agent, environment, state space, action space and reward function, where;
[0042] The agent is used to use the output of the graph neural network model as the input of the reinforcement learning agent;
[0043] The environment gives feedback based on the agent's actions, i.e. rewards or penalties, which are used to update the agent's strategy;
[0044] The state space represents the current state of the agent, that is, the prediction result of the sensitivity level of user data;
[0045] The action space represents the actions that the agent can take, i.e. adjusting the parameters of the graph neural network model;
[0046] The reward function gives feedback, i.e. reward or punishment, based on the agent's actions. The reward function can make the agent tend to take actions that can accurately identify the sensitivity level of user data.
[0047] The policy update uses the deep deterministic policy gradient (DDPG) algorithm for policy update. The specific formula is: Among them, θ k represents the policy parameter of the kth iteration, α represents the learning rate, represents the objective function j with respect to parameter θ k The gradient of j(θ k ) represents the objective function.
[0048] S4. Sensitive data identification and protection. Use the trained graph neural network and reinforcement learning model to identify the sensitivity level of user data, and classify the user data based on the identification results, such as encrypted storage and access control, to ensure the privacy and security of user data. At the same time, monitor the changes in the sensitivity level of user data in real time, and take corresponding measures in time to prevent data leakage and abuse. The sensitive data identification and protection also includes anonymization of user data to ensure the privacy and security of user data.
[0049] Specifically, in this step, the trained graph neural network and reinforcement learning model are used to identify the sensitivity level of user data, and the user data is classified and processed according to the identification results, such as encrypted storage and access control. For data with higher sensitivity levels, stricter encryption and access control strategies are adopted. The sensitivity level changes of user data are monitored in real time, potential data leakage and abuse risks are discovered and handled in a timely manner, and user data is anonymized, which can further protect the privacy and security of user data.
[0050] Embodiment 2,
[0051] In order to implement the above method, the embodiment of the present invention also proposes a virtual power plant user data identification device, including a data preprocessing module, a graph neural network model module, a reinforcement learning model module and a sensitive data identification and protection module;
[0052] The data preprocessing module is used to preprocess the virtual power plant user data collected in step S1;
[0053] The graph neural network model module is used to build the graph neural network model in step S2 and includes a submodule for executing the GCN or GAT algorithm. In addition, it also includes an attention mechanism module for improving the learning ability of node and edge features in the user data graph network;
[0054] The reinforcement learning model module is used to build the reinforcement learning model in step S3 and includes a submodule for executing the DDPG algorithm. The module also includes a deep deterministic policy gradient algorithm module to improve the learning efficiency and stability of the agent strategy;
[0055] A sensitive data identification and protection module, used to perform sensitive data identification and protection in step S4, and the module also includes an anonymization processing module, used to anonymize user data;
[0056] In addition, the device also includes a feature selection and feature dimension reduction module, which is used to perform feature selection and feature dimension reduction processing on virtual power plant user data to improve the efficiency and accuracy of data processing.
[0057] Through the implementation of the above-mentioned method and device, the present invention can realize the accurate identification and protection of user data of virtual power plants. Specifically, the method can accurately identify the sensitivity level of user data, and classify and process user data according to the identification results, such as encrypted storage and access control. At the same time, by real-time monitoring of changes in the sensitivity level of user data, potential data leakage and abuse risks can be discovered and handled in a timely manner, thereby effectively protecting the privacy and security of user data.
[0058] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims
1. A method for identifying user data of a virtual power plant, characterized in that: The method comprises the following steps: S1. Data preprocessing, for collecting virtual power plant user data, cleaning and standardizing the data, and building a user data graph network; S2. Construct a graph neural network model, wherein the graph neural network model includes an input layer, a hidden layer, and an output layer. The input layer is used to receive the user data graph network as input and extract the features of nodes and edges. The hidden layer adopts a graph convolutional neural network (GCN) or a graph attention network (GAT) graph neural network model. The output layer is used to output the sensitivity level prediction result of the user data. S3. Build a reinforcement learning model and design a reinforcement learning framework, including the agent, environment, state space, action space, and reward function. The output of the graph neural network model is used as the input of the reinforcement learning agent. The agent selects actions based on the current state, that is, adjusts the parameters of the graph neural network model. The environment gives feedback based on the agent's actions, that is, rewards or penalties, which are used to update the agent's strategy. Through continuous iterative training, the agent learns the optimal strategy and can accurately identify the sensitivity level of user data. S4. Sensitive data identification and protection. Use trained graph neural networks and reinforcement learning models to identify the sensitivity level of user data, and classify and process user data based on the identification results, such as encrypted storage and access control, to ensure the privacy and security of user data. At the same time, monitor the changes in the sensitivity level of user data in real time and take corresponding measures in a timely manner to prevent data leakage and abuse.
2. The method for identifying user data of a virtual power plant according to claim 1, characterized in that: In step S2, the mathematical expression of the graph neural network model using the graph convolutional neural network (GCN) or the graph attention network (GAT) is: Among them, H (l) represents the node feature matrix of the lth layer, represents the normalized adjacency matrix, W (l) represents the weight matrix of the lth layer, and σ represents the activation function.
3. The method for identifying user data of a virtual power plant according to claim 1, characterized in that: In step S3, the policy update adopts the deep deterministic policy gradient (DDPG) algorithm, and the specific formula is: Among them, θ k represents the policy parameter of the kth iteration, α represents the learning rate, represents the objective function j with respect to parameter θ k The gradient of j(θ k ) represents the objective function.
4. The method for identifying user data of a virtual power plant according to claim 1, characterized in that: In step S1, the data preprocessing also includes feature selection and feature dimension reduction processing of virtual power plant user data to improve the efficiency and accuracy of data processing.
5. The method for identifying user data of a virtual power plant according to claim 1, characterized in that: In step S2, when the graph neural network model is constructed, the graph neural network architecture also includes the introduction of an attention mechanism to improve the learning ability of node and edge features in the user data graph network.
6. The method for identifying user data of a virtual power plant according to claim 1, characterized in that: In step S4, the sensitive data identification and protection also includes anonymizing the user data to ensure the privacy and security of the user data.
7. The method for identifying user data of a virtual power plant according to claim 1, characterized in that: It also includes a virtual power plant user data identification device, including a data preprocessing module, a graph neural network model module, a reinforcement learning model module and a sensitive data identification and protection module, wherein; A data preprocessing module, used to preprocess the virtual power plant user data collected in step S1; The graph neural network model module is used to build the graph neural network model in step S2 and contains submodules for executing the GCN or GAT algorithm; The reinforcement learning model module is used to build the reinforcement learning model in step S3 and contains a submodule for executing the DDPG algorithm; The sensitive data identification and protection module is used to perform sensitive data identification and protection in step S4.
8. The method for identifying user data of a virtual power plant according to claim 7, characterized in that: It also includes feature selection and feature dimension reduction modules, which are used to perform feature selection and feature dimension reduction processing on virtual power plant user data.
9. The method for identifying user data of a virtual power plant according to claim 7, characterized in that: The graph neural network model module also includes an attention mechanism module for improving the learning ability of node and edge features in the user data graph network; The reinforcement learning model module also includes a deep deterministic policy gradient algorithm module, which is used to improve the learning efficiency and stability of the agent strategy; The sensitive data identification and protection module also includes an anonymization processing module for anonymizing user data to further protect the privacy and security of user data.