Cross-regional power grid data privacy protection collaborative analysis method and system

Abstract

The invention discloses a cross-regional power grid data privacy protection collaborative analysis method and system, and relates to the related field of power system data security, and the method comprises the steps: executing the multi-dimensional attribute vector extraction of power grid data, and building a multi-dimensional attribute vector set; performing mapping matching of an encryption graph by using the multi-dimensional attribute vector set, and establishing an encryption packaging scheme; analyzing the collaborative analysis task, and establishing an analysis result; after the data is fed back to the data providing layer, power grid data dynamic adaptation analysis is executed, and a disturbance level control channel is configured; after the power grid data is encrypted and packaged by using the encryption packaging scheme, calling disturbance control of the packaged data is carried out by using a disturbance level control channel; and generating collaborative risk prediction of the power grid according to a control result, and performing user feedback management. The technical problem that it is difficult to balance data privacy protection and data availability in existing cross-regional power grid data privacy protection collaborative analysis is solved, and the technical effect of improving data availability on the premise that data privacy security is guaranteed is achieved.

Description

Technical Field

[0001] The present application relates to fields related to power system data security, and in particular to a collaborative analysis method and system for cross-regional power grid data privacy protection. Background Art

[0002] In the electric energy sector, the coordinated operation and management of cross-regional power grids is crucial for ensuring the stability of power supply, optimizing resource allocation, and improving overall grid efficiency. As power grids become increasingly information-based and intelligent, the need for cross-regional power grid data sharing and collaborative analysis is becoming increasingly urgent. However, the privacy protection of the massive amounts of data involved has become a key bottleneck restricting the coordinated development of power grids.

[0003] Currently, the primary approach to addressing data privacy protection across regional power grids is through the use of traditional data encryption technologies and permission-based access management strategies. Traditional encryption technologies ensure data confidentiality during transmission and storage by encrypting the data. Permission-based access management strategies strictly control data access based on user roles and permissions. However, while traditional encryption technologies can ensure data security at rest, in cross-regional power grid collaborative analysis scenarios, data needs to dynamically interact and be processed across different regions and entities. Simple static encryption is unable to adapt to the complex logic requirements of collaborative analysis, potentially significantly reducing data availability. Furthermore, permission-based access management strategies, faced with the complex multi-dimensional relationships between cross-regional power grid data, struggle to achieve precise and flexible data access control, leading to the risk of data leakage and failing to meet the dual needs of cross-regional power grid data privacy protection and collaborative analysis.

[0004] Among the current related technologies, cross-regional power grid data privacy protection collaborative analysis has the technical problem of being unable to balance data privacy protection and data availability. Summary of the Invention

[0005] This application provides a cross-regional power grid data privacy protection collaborative analysis method and system, adopts multi-dimensional attribute vector extraction and encryption spectrum mapping matching technology, extracts attribute vectors from power grid data from multiple dimensions such as data sensitivity, historical time series correlation, causal correlation, predicted target variables, control intervention, and establishes a vector set, and then maps and matches to generate an encryption encapsulation scheme. At the same time, based on the analysis of collaborative analysis tasks, combined with dynamic adaptation analysis to configure a disturbance level control channel, and use this channel to call disturbance control and other technical means on the encrypted and encapsulated data, it solves the technical problem of the existing cross-regional power grid data privacy protection collaborative analysis that is difficult to balance data privacy protection and data availability, and achieves the technical effect of improving the availability of data in the collaborative analysis process while ensuring data privacy security.

[0006] The present application provides a cross-regional power grid data privacy protection collaborative analysis method, including: performing multidimensional attribute vector extraction of power grid data, establishing a multidimensional attribute vector set, the dimensions of the multidimensional attribute vector extraction including data sensitivity dimension, historical time series correlation dimension, causal correlation dimension, prediction target variable dimension, and control intervention dimension; performing mapping matching of an encrypted graph using the multidimensional attribute vector set, and establishing an encryption encapsulation scheme; after receiving a collaborative analysis task, parsing the collaborative analysis task, and establishing a parsing result, the parsing result including an analysis logic summary and a data type usage statement; after feeding back the parsing result to a data provision layer, performing dynamic adaptation analysis of the power grid data, and configuring a disturbance level control channel based on the dynamic adaptation analysis result; after encrypting and encapsulating the power grid data using the encryption encapsulation scheme, using the disturbance level control channel to perform call disturbance control of the encapsulated data; generating a collaborative risk prediction of the power grid based on the call disturbance control result, and performing user feedback management on the collaborative risk prediction.

[0007] In a possible implementation, after the analysis result is fed back to the data provision layer, dynamic adaptation analysis of the power grid data is performed, and the disturbance level control channel is configured based on the dynamic adaptation analysis result, and the following processing is performed: the target power grid data is located by the data provision layer; the usage sensitivity analysis network of the data provision layer is used to calculate the usage sensitivity according to the target power grid data and the data type usage statement to generate a first adaptation analysis result; the reversible backtracking network of the data provision layer is called, and after the analysis logic summary is sent to the reversible backtracking network, the original data restoration capability of the target power grid data in the collaborative analysis task scenario is analyzed to establish a second adaptation analysis result; the sensitivity mean of the multidimensional attribute vector set is obtained, and a third adaptation analysis result is established based on the sensitivity mean; the dynamic adaptation analysis result is established based on the first adaptation analysis result, the second adaptation analysis result, and the third adaptation analysis result.

[0008] In a possible implementation, the usage sensitivity analysis network using the data provision layer performs usage sensitivity calculation based on the target power grid data and the data type usage statement, and performs the following processing: using the preprocessing layer of the usage sensitivity analysis network to convert the data type usage statement into a function chain structure; mapping the target power grid data to each module participating in the function chain, identifying directly used fields, derived reasoning fields, and indirectly affected fields; using the calculation layer of the usage sensitivity analysis network to perform usage path tension calculation on each identified field, and completing the usage sensitivity calculation based on the usage path tension calculation result.

[0009] In a possible implementation, the calculation layer is as follows: ;in, Characterize the results of sensitivity calculations for usage, is the total number of fields, Represents any field, Characterization The basic sensitivity weight of each field, Characterization Whether the field is directly used by the current purpose task, Characterization Whether the field is indirectly used by the task, Characterization The field usage and control conflict of each field, 、 are the weight factors for the indirect use component and the use control conflict component respectively.

[0010] In a possible implementation, the disturbance level control channel is configured based on the dynamic adaptation analysis results, and the following processing is performed: a basic trust level of the target power grid data is generated according to the multidimensional attribute vector set; a usage trust level is established according to the data type usage statement; a field decision constraint is constructed using the basic trust level and the usage trust level, and the disturbance level control channel is corrected using the field decision constraint.

[0011] In a possible implementation, after the analysis logic summary is sent to the reversible traceability network, the original data restoration capability analysis of the target power grid data in the collaborative analysis task scenario is performed, and the following processing is performed: the data dependency structure in the analysis logic summary is extracted and converted into a data access graph; a field reconstruction chain is constructed using the data access graph, and the field reconstruction chain represents the reverse restoration path; the reversibility index of each original data field is calculated using the field reconstruction chain, and the calculated feature dimensions include the shortest path length from the original data field to the output result, the calculation feature type in the path, the participation frequency of the original data field, the statistical feature significance, and the stability of the field reconstruction chain; and the restoration capability analysis is completed based on the reversibility index.

[0012] In a possible implementation, after receiving the collaborative analysis task, the following processing is performed: activating the task ontology identification mechanism, and using the task ontology identification mechanism to perform deep semantic modeling of task intent based on the collaborative analysis task; using the deep semantic modeling results of task intent to perform conflict detection, and the conflict detection includes use exceeding authority and use drift; using conflict detection to report conflict points and prompt warnings.

[0013] In a possible implementation, the configuration of the disturbance level control channel based on the dynamic adaptation analysis results further performs the following processing: establishing an adaptation game model of task efficiency and risk cost; synchronizing the dynamic adaptation analysis results to the adaptation game model and executing a balance game; and configuring the disturbance level control channel using the balance game results.

[0014] In a possible implementation, the use of the balance game result to configure the disturbance level control channel performs the following processing: creating a disturbance mapping table, which is mapped one-to-one with the task value; after evaluating the task value of the collaborative analysis task, calling the disturbance mapping table to establish the basic disturbance; configuring the random disturbance factor, and using the basic disturbance, the random disturbance factor and the balance game result to configure the disturbance level control channel.

[0015] The present application also provides a cross-regional power grid data privacy protection collaborative analysis system, including: a multidimensional attribute vector extraction module, used to perform multidimensional attribute vector extraction of power grid data and establish a multidimensional attribute vector set, wherein the dimensions of the multidimensional attribute vector extraction include data sensitivity dimension, historical time series correlation dimension, causal correlation dimension, prediction target variable dimension, and control intervention dimension; an encryption encapsulation scheme establishment module, used to perform mapping and matching of the encryption graph using the multidimensional attribute vector set to establish an encryption encapsulation scheme; a collaborative analysis task parsing module, used to parse the collaborative analysis task after receiving it and establish a parsing result, wherein the parsing result includes an analysis logic summary and a data type usage statement; a disturbance level control channel configuration module, used to perform dynamic adaptive analysis of power grid data after feeding the parsing result back to the data provision layer, and configure the disturbance level control channel based on the dynamic adaptive analysis result; a call disturbance control module, used to perform call disturbance control of the encapsulated data using the disturbance level control channel after encrypting and encapsulating the power grid data using the encryption encapsulation scheme; and a collaborative risk prediction generation module, used to generate a collaborative risk prediction of the power grid based on the call disturbance control result, and perform user feedback management on the collaborative risk prediction.

[0016] The cross-regional power grid data privacy protection collaborative analysis method and system proposed in this application first performs multi-dimensional attribute vector extraction of power grid data to establish a multi-dimensional attribute vector set. The dimensions of the multi-dimensional attribute vector extraction include data sensitivity dimension, historical time series correlation dimension, causal correlation dimension, prediction target variable dimension, and control intervention dimension. Then, the multi-dimensional attribute vector set is used to perform mapping and matching of the encrypted graph to establish an encryption encapsulation scheme. Then, after receiving the collaborative analysis task, the collaborative analysis task is parsed to establish the parsing result. The parsing result includes an analysis logic summary and a data type usage statement. After the parsing result is fed back to the data provision layer, dynamic adaptive analysis of the power grid data is performed. Based on the dynamic adaptive analysis result, a disturbance level control channel is configured. After the power grid data is encrypted and encapsulated using the encryption encapsulation scheme, the disturbance level control channel is used to call disturbance control of the encapsulated data. Finally, a collaborative risk prediction of the power grid is generated based on the call disturbance control result, and the collaborative risk prediction is managed for user feedback. The technical effect of improving the availability of data in the collaborative analysis process while ensuring data privacy security is achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings of the embodiments of the present invention are briefly introduced below. Flowcharts are used in this application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed in precise order. Instead, various steps may be processed in reverse order or simultaneously as needed. Furthermore, other operations may be added to these processes, or one or more operations may be removed from these processes.

[0018] Figure 1 A flowchart of a cross-regional power grid data privacy protection collaborative analysis method provided in an embodiment of the present application.

[0019] Figure 2 A schematic diagram of the structure of the cross-regional power grid data privacy protection collaborative analysis system provided in an embodiment of the present application.

[0020] Explanation of the accompanying symbols: multi-dimensional attribute vector extraction module 10, encryption encapsulation scheme establishment module 20, collaborative analysis task parsing module 30, disturbance level control channel configuration module 40, calling disturbance control module 50, collaborative risk prediction generation module 60. DETAILED DESCRIPTION

[0021] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below.

[0022] In order to make the purpose, technical solutions and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings. The described embodiments should not be regarded as limiting this application. All other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0023] In the following description, reference is made to “some embodiments”, which describes a subset of all possible embodiments, but it will be understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict, and the terms “first\second” involved are merely used to distinguish similar objects and do not represent a specific ordering of the objects. The terms “including” and “having” and any variations are intended to cover non-exclusive inclusions. For example, a process, method, system, product or server that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or modules that are not clearly listed or that are inherent to these processes, methods, products or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this application belongs. The terms used herein are for the purpose of describing the embodiments of this application only.

[0024] The embodiment of the present application provides a collaborative analysis method for cross-regional power grid data privacy protection, such as Figure 1 As shown, the method includes: Step S100, performing multidimensional attribute vector extraction of power grid data, establishing a multidimensional attribute vector set, wherein the dimensions of the multidimensional attribute vector extraction include data sensitivity dimension, historical time series correlation dimension, causal correlation dimension, predicted target variable dimension, and control intervention dimension.

[0025] Specifically, for the data sensitivity dimension, expert systems are used to assess the risk level of data leakage. For example, decision trees or rule engines can be used to automatically assess and classify data sensitivity. For the historical time series correlation dimension, time series analysis tools (such as Python's statsmodels library) are used to extract time-related features. For example, ARIMA models or seasonal decomposition can be used to analyze time series data. For the causal correlation dimension, causal inference frameworks (such as the DoWhy library) are applied to identify causal relationships. For example, Granger causality tests or causal graph models can be used to determine causal relationships between variables. For the prediction target variable dimension, relevant features are selected based on the prediction target (such as using scikit-learn's feature_selection module). For example, recursive feature elimination (RFE) or model-based feature selection methods can be used. For the control intervention dimension, the impact of control strategies on the data is analyzed and control-related features are extracted. For example, simulations or historical data analysis can be used to determine the specific impact of control strategies on the data.

[0026] For example, consider a power grid dataset containing the following fields: user ID, timestamp, electricity consumption (kWh), weather conditions (sunny, cloudy, rainy), holidays (yes / no), and device status (normal, faulty). For the data sensitivity dimension, the sensitivity of each field is assessed, marking the user ID as "high sensitivity," electricity consumption as "medium sensitivity," and other fields, such as weather conditions and holidays, as "low sensitivity." For the historical time series correlation dimension, the data trends over time are analyzed, extracting hourly and daily averages of electricity consumption, as well as hourly and daily rates of change. For the causal correlation dimension, Granger causality tests are used to determine the causal relationship between weather conditions and electricity consumption. For the prediction target variable dimension, relevant features are selected based on the prediction objective. For example, if the goal is to predict future electricity consumption, electricity consumption over the past 24 hours, weather conditions over the past week, and holiday information are selected as predictive features. For the control intervention dimension, the impact of control strategies on the data is analyzed, extracting the time periods when demand response plans were implemented and the corresponding electricity consumption changes. The above extracted features are combined into a multidimensional attribute vector set, each vector representing a feature set of a data record.

[0027] Step S200: performing mapping matching of the encrypted graph using the multi-dimensional attribute vector set to establish an encryption encapsulation solution.

[0028] Specifically, a graph is created that associates different data attribute combinations with different encryption methods. These association rules can be stored in a database management system. For example, high-sensitivity data is associated with the CP-ABE encryption method, and low-sensitivity data is associated with the FHE-VTS encryption method. The multidimensional attribute vector extracted in step S100 is matched with the encryption graph to identify the attributes of each data item and find the corresponding encryption method in the graph. Specifically, a machine learning classifier can be used to automatically match encryption methods. Based on the mapping results, the most appropriate encryption method is selected for each data item, and an encryption encapsulation scheme is established. This scheme details how the data will be encrypted and how access to the encrypted data will be managed and controlled. This scheme can be implemented using security protocols and encryption standards (such as AES and RSA). For example, if the data item is marked as highly sensitive and requires strict access control, the CP-ABE encryption method is selected.

[0029] For example, for high-sensitivity + controllable attributes, a policy-based CP-ABE + dynamic signature chain hash binding encryption method is used, allowing access only to specific policy signature chains matching the policy. For causal + predictive attributes, the polynomial approximate homomorphic encryption (HE-Poly) encryption method is used to preserve the gradient and residual direction of the model structure. This is suitable for scenarios that require predictive analysis in an encrypted state, such as the inference phase of machine learning models. For medium-sensitivity + derivable attributes, the perturbation injection secret state computation (Noisy Enc) encryption method is used to introduce semantically reversible perturbations into the secret state to protect the true change trajectory. This is suitable for scenarios with medium-sensitivity data that require some form of data analysis. For low-sensitivity + time series attributes, the fuzzy verifiable homomorphic sequence encryption (FHE-VTS) encryption method is used to ensure time series pattern analysis capabilities and hide node meaning. This is suitable for scenarios with low data sensitivity but the need to protect data time series patterns.

[0030] Consider the following power grid data attributes: Data A: High sensitivity, used for predictive analysis. Data B: Medium sensitivity, requiring time series analysis. Data C: Low sensitivity, requiring time series pattern protection. For Data A, due to its high sensitivity and predictive use, CP-ABE + dynamic signature chain hash binding is chosen to ensure data access only by authorized users and protect data integrity. For Data B, due to its medium sensitivity and the need for time series analysis, Noisy Enc is chosen to enable effective time series analysis while protecting data privacy. For Data C, due to its low sensitivity and the need for time series pattern protection, FHE-VTS is chosen to ensure accurate time series pattern analysis while concealing the data's specific meaning. This approach allows for the flexible selection of the most appropriate encryption method based on the specific attributes and usage scenarios of the data, ensuring effective data utilization while protecting data privacy.

[0031] Step S300: After receiving the collaborative analysis task, the collaborative analysis task is parsed to create a parsing result, which includes an analysis logic summary and a data type usage statement.

[0032] Specifically, receive a description of a collaborative analysis task from a user or other system, use natural language processing (NLP) technology to parse the task description, and extract key information. For example, use text classification to identify the task type (such as prediction, classification, clustering, etc.), and use entity recognition to extract data fields and analysis objectives. Based on the parsing results, formulate a summary of the analysis logic, including determining the analysis steps, such as data cleaning, feature engineering, model selection, parameter tuning, etc. For example, if the task is to predict electricity demand, the analysis logic may include steps such as time series analysis, feature selection, model training and validation. Clarify the role of different data types in the analysis task. For example, timestamps can be used for time series analysis, and weather conditions can be used as features input into the prediction model. The statement may include data preprocessing requirements, such as normalization and missing value processing.

[0033] In one possible implementation, after receiving the collaborative analysis task, step S300 further includes step S310, activating a task ontology recognition mechanism and utilizing it to perform deep semantic modeling of the task intent based on the collaborative analysis task. Specifically, natural language processing (NLP) technology and ontology methods are used to analyze the task description, extract key information, and construct a semantic model of the task. Through the task ontology recognition mechanism, deep semantic modeling of the collaborative analysis task is performed to accurately understand the task intent and requirements.

[0034] Step S320 uses the deep semantic modeling results of the task intent to perform conflict detection. This includes detection of unauthorized use and drift. Specifically, based on the semantic modeling results obtained in step S310, the task intent is checked for potential conflicts with the data usage policy, including unauthorized use (the task intent exceeds the permitted scope of data use) and drift (the task intent deviates from the original data usage).

[0035] Step S330: Utilize conflict detection to report conflict points and issue warnings. Specifically, the conflict detection results are used to automatically generate warning information and notify relevant personnel through the user interface or message system.

[0036] For example, consider a power grid data analysis platform that processes power grid data from different regions and performs collaborative analysis tasks. A user submits a task to analyze electricity demand in a specific region and forecast electricity load for the coming week. The system activates the task ontology recognition mechanism and uses natural language processing (NLP) technology to parse the task description. The system identifies key entities (such as "power demand," "specific region," and "coming week") and relationships (such as "analysis" and "forecast"). Based on the identified entities and relationships, the system constructs a semantic model that clearly defines the task's objective as power demand analysis and load forecasting. The system then checks whether this task conflicts with existing data usage policies or rules. For example, it checks whether load forecasting using data from a specific region is permitted. If the policy stipulates that certain data can only be used for internal analysis, but the task requires the results to be published publicly, this constitutes unauthorized use. If the data was originally intended for short-term load forecasting, but the task requires long-term forecasting, this constitutes usage drift. The system automatically detects these potential conflicts and assesses whether they violate data usage policies. If the system discovers that a task's requested data usage exceeds policy-permitted usage, it generates an alert stating, "Use beyond authorization detected: The task requires the results to be published publicly, but policy stipulates that the data can only be used for internal analysis." This alert is displayed to the task submitter through the user interface and sent to the system administrator via email or text message. This implementation approach uses conflict detection to proactively identify and prevent potential data usage conflicts, ensuring data compliance and reducing the risk of data misuse.

[0037] Step S400: After feeding back the analysis result to the data providing layer, a dynamic adaptive analysis of the power grid data is performed, and a disturbance level control channel is configured based on the dynamic adaptive analysis result.

[0038] Specifically, based on the analysis results, adaptive analysis is performed on the power grid data, including data cleaning, feature conversion, and data standardization. This ensures that the data is suitable for collaborative analysis, adapting the data format and content to specific analysis requirements while also taking into account data privacy. Based on the results of dynamic adaptive analysis, a perturbation level control channel is configured, determining the amount of perturbation (e.g., noise) introduced into the data to protect data privacy while ensuring the accuracy of data analysis. By controlling the perturbation level, a balance is achieved between data privacy protection and data analysis accuracy.

[0039] In one possible implementation, after feeding the analysis results back to the data provision layer, dynamic adaptive analysis of power grid data is performed. Step S400 further includes step S410, where the data provision layer locates target power grid data. Specifically, a data index or query system is utilized to locate a specific power grid dataset based on task requirements and data type declarations, and the required power grid data is accurately found in the data provision layer.

[0040] Step S420: Utilizing the usage sensitivity analysis network of the data provision layer, a usage sensitivity calculation is performed based on the target power grid data and the data type usage statement, generating a first adaptation analysis result. Specifically, the usage sensitivity analysis network of the data provision layer is combined with the data type usage statement to calculate the sensitivity of the data and determine the risk level of the data under different usages.

[0041] Step S430: Invoke the reversible traceability network of the data provider layer, send the analysis logic summary to the reversible traceability network, and then analyze the original data restoration capability of the target power grid data in the collaborative analysis task scenario to establish a second adaptive analysis result. Specifically, the reversible traceability network of the data provider layer is invoked, the analysis logic summary is sent, and the reversibility of the data during the analysis process and the original data restoration capability are analyzed.

[0042] Step S440: Obtain the sensitivity mean of the multidimensional attribute vector set, and establish a third adaptation analysis result based on the sensitivity mean. Specifically, extract sensitivity data from the multidimensional attribute vector set and calculate its mean.

[0043] Step S450: Create the dynamic adaptation analysis result based on the first adaptation analysis result, the second adaptation analysis result, and the third adaptation analysis result. Specifically, based on the results of steps S420 to S440, the first, second, and third adaptation analysis results are integrated to form a comprehensive dynamic adaptation analysis result.

[0044] For example, suppose that grid data from two different regions (Region A and Region B) needs to be analyzed to predict future electricity demand and optimize grid operations. The data includes user IDs, timestamps, electricity usage, weather conditions, holidays, and device status. First, a data indexing system is used to locate relevant grid datasets for Regions A and B from the data provision layer based on the task requirements (e.g., predicting electricity demand) and data type declarations (e.g., electricity usage, weather conditions). The data provision layer's usage sensitivity analysis network, combined with the data type usage declarations, calculates the sensitivity of the data. For example, user IDs are labeled high, while weather conditions are labeled low. The data provision layer's reversible backtracking network is invoked, sending a summary of the analysis logic to analyze the reversibility of the data during the analysis process and the ability to restore the original data. For example, the analysis evaluates whether accurate electricity usage data can be restored without leaking user IDs. Sensitivity data is extracted from the multidimensional attribute vector set and its mean is calculated. For example, the average sensitivity of all attributes is calculated. These results are integrated to form a comprehensive dynamic adaptive analysis, which can be used to guide adjustments to data usage and protection policies. This implementation enhances data privacy protection by calculating usage sensitivity and analyzing the ability to restore original data, ensuring data security during analysis. By comprehensively considering the sensitivity mean of multi-dimensional attribute vectors, the reliability and accuracy of data analysis are improved.

[0045] In one possible implementation, configuring the perturbation hierarchy control channel based on the dynamic adaptive analysis results, step S400 further includes step S460, which establishes an adaptive game model for task efficiency and risk cost. Specifically, an adaptive game model is constructed that considers task execution efficiency (e.g., analysis speed and accuracy) and risk cost (e.g., the possibility of privacy leakage and the degree of data distortion). Relevant parameters are defined for the model, such as task type, data sensitivity, expected analysis result accuracy, and privacy protection level.

[0046] In step S470, the dynamic adaptation analysis results are synchronized with the adaptation game model to execute a balancing game. Specifically, the dynamic adaptation analysis results are input into the adaptation game model, which is then run to simulate different strategies and parameter settings to find the optimal or satisfactory solution. This solution maximizes task execution efficiency while ensuring data privacy.

[0047] Step S480, configure the perturbation level control channel using the balance game result. Specifically, according to the balance game result obtained in step S470, configure the perturbation level control channel, including adjusting the encryption strength, the level of perturbation injection, etc. Apply the configured perturbation level control channel to the data encryption and analysis process to ensure that data privacy can be effectively protected when performing collaborative analysis tasks. This implementation method can find the best balance between protecting data privacy and improving task execution efficiency by adapting the game model. Through dynamic adaptive analysis and balanced game, the system can adapt to different task requirements and data characteristics, and provide a more flexible privacy protection strategy.

[0048] In one possible implementation, configuring the disturbance level control channel based on the dynamic adaptation analysis results, step S400 further includes step S490 of generating a basic trust level for the target power grid data based on the multidimensional attribute vector set. Specifically, a basic trust level is generated for each data field or data set by analyzing factors such as the data source, historical accuracy, and acquisition method. The basic trust level is an initial trust level generated for the target power grid data based on the multidimensional attribute vector set of the data, and this level reflects the inherent trustworthiness of the data.

[0049] Step S4100: Establish a usage trust level based on the data type usage statement. Specifically, based on the data type usage statement, evaluate the applicability and credibility of the data for that usage, and establish a usage trust level that reflects the credibility of the data for that specific usage.

[0050] Step S4110, using the basic trust level and the purpose trust level to construct a field decision constraint, and using the field decision constraint to correct the disturbance level control channel. Specifically, the basic trust level and the purpose trust level are used to construct field decision constraints, which are used to guide how to adjust the disturbance level control channel to ensure the privacy and availability of the data. According to the field decision constraints, the disturbance level control channel is corrected to ensure that while protecting the privacy of the data, the availability of the data and the accuracy of the analysis are not affected. This implementation method can more accurately control the disturbance level of the data by establishing trust levels and field decision constraints, thereby reducing the impact on the accuracy of data analysis while protecting privacy.

[0051] For example, suppose you are processing a cross-regional power grid dataset containing user electricity usage information. The goal is to analyze this data to predict electricity demand while protecting user privacy. First, evaluate the base trust level of each field in the dataset. For example, user ID: Due to personal privacy implications, the base trust level is low (e.g., 2 / 10); electricity usage: This is a relatively insensitive metric, so the base trust level is high (e.g., 8 / 10); timestamp: Although the timestamp itself is not sensitive, it is associated with user behavior, so the trust level is medium (e.g., 5 / 10). Next, adjust the trust level based on the data's stated purpose. For example: User ID: When used to predict electricity demand, the use trust level is further reduced (e.g., 1 / 10) due to privacy risks; electricity usage: When used directly for predictive analysis, the use trust level remains unchanged or slightly increased (e.g., 8 / 10 or 9 / 10); timestamp: When used to analyze cyclical fluctuations in electricity demand, the use trust level is increased (e.g., 6 / 10). Finally, field decision constraints are constructed based on the base trust level and the use trust level, and the perturbation-level control channel is calibrated accordingly. For example: User ID: Due to the low trust level, more disturbances need to be introduced (for example, increasing noise) to protect user privacy; Power usage: Due to the high trust level, fewer disturbances can be introduced to maintain the accuracy of analysis; Timestamp: The medium trust level means that it is necessary to balance disturbance and accuracy, and moderate disturbance can be selected.

[0052] In one possible implementation, the use sensitivity analysis network of the data provision layer performs use sensitivity calculation based on the target power grid data and the data type use statement, and step S420 further includes step S421 of converting the data type use statement into a function chain structure using a preprocessing layer of the use sensitivity analysis network. Specifically, the preprocessing layer of the use sensitivity analysis network is used to convert the data type use statement into a function chain structure, that is, to decompose the data use into a series of function modules and the relationships between them, so as to better understand and analyze the data use.

[0053] Step S422, map the target power grid data to the modules involved in each functional chain, and identify directly used fields, derived reasoning fields, and indirectly affected fields. Specifically, map the target power grid data to the modules involved in each functional chain, and identify different usage modes of data fields. Directly used fields are data fields directly used in the analysis, such as electricity consumption. Derived reasoning fields are data fields inferred by analyzing directly used fields, such as user behavior patterns. Indirectly affected fields are data fields that have an indirect impact on the analysis results, such as weather conditions.

[0054] Step S423: Calculate the usage path tension for each identified field using the calculation layer of the usage sensitivity analysis network, and complete the usage sensitivity calculation based on the usage path tension calculation result. The calculation layer is as follows: ; in, Characterize the results of sensitivity calculations for usage, is the total number of fields, Represents any field, Characterization The basic sensitivity weight of each field, Characterization Whether the field is directly used by the current purpose task, Characterization Whether the field is indirectly used by the task, Characterization The field usage and control conflict of each field, 、 are the weighting factors for the indirect use component and the use control conflict component, respectively. Specifically, the use path tension for each field is calculated to complete the use sensitivity calculation. This calculation is performed using the computational layer of the use sensitivity analysis network according to the formula.

[0055] For example, consider a power grid dataset containing the following fields: user ID, electricity usage, timestamp, and weather conditions. This data needs to be analyzed to predict electricity demand while protecting user privacy. Transform the data type's purpose statement into a functional chain structure. In this example, the purpose statement is "analyze electricity demand." This statement can be transformed into the following functional chain structure: Functional Module 1: Data Collection (collecting user ID, electricity usage, timestamp, and weather conditions); Functional Module 2: Data Preprocessing (cleaning and formatting data); Functional Module 3: Feature Extraction (extracting hour and date information from timestamps); Functional Module 4: Model Training (training a prediction model using electricity usage and weather conditions); Functional Module 5: Predicting Electricity Demand (using the trained model for prediction). Map the target power grid data to each module involved in the functional chain, identifying directly used fields, derived inference fields, and indirectly affected fields. In this example, the directly used field is electricity usage (directly used in model training and prediction), the derived inference field is None (in this simple example, no data fields are derived through inference), and the indirectly affected field is weather conditions. The calculation layer of the usage sensitivity analysis network is used to calculate the usage path tension of each identified field, and the usage sensitivity calculation is completed based on the usage path tension calculation results. The example calculation process is as follows, for example, for user ID: basic sensitivity weight = 0.8 (assuming user ID is very sensitive), use directly =0 (user ID is not used directly for prediction), indirectly used =1 (user ID indirectly affects prediction), usage and control conflict =0.5 (assuming there is some conflict in the use of user IDs), weight factor =0.3, =0.2, usage sensitivity =0.8×(0+0.3×1+0.2×0.5)=0.32. This implementation method can more accurately calculate the data's usage sensitivity by analyzing the different usage patterns of data fields in detail. By identifying and calculating usage sensitivity, sensitive data can be more effectively protected, reducing the risk of privacy leaks.

[0056] In one possible implementation, after sending the analysis logic summary to the reversible traceability network, an analysis of the original data restoration capability of the target power grid data in the collaborative analysis task scenario is performed. Step S430 further includes step S431, extracting the data dependency structure in the analysis logic summary and converting it into a data access graph. Specifically, natural language processing (NLP) technology is used to parse the analysis logic summary, extracting the data dependency structure from the analysis logic summary, including direct and indirect dependencies between data fields. These dependencies are then converted into a data access graph using a graph database (such as Neo4j) or a graph processing library (such as GraphX), where nodes in the graph represent data fields and edges represent dependencies between fields.

[0057] In step S432, a field reconstruction chain is constructed using the data access graph. The field reconstruction chain represents the reverse restoration path. Specifically, starting from the output result node, a graph traversal algorithm (such as depth-first search (DFS) or breadth-first search (BFS)) is used to reversely traverse the data access graph. During the traversal, the path from the output result to each original data field is recorded. These paths constitute the field reconstruction chain, which represents the path from the output result back to the original data field.

[0058] In step S433, the reversibility index of each original data field is calculated using the field reconstruction chain. The calculated feature dimensions include the shortest path length from the original data field to the output result, the type of computational features in the path, the frequency of the original data field's participation, the statistical feature significance, and the stability of the field reconstruction chain. Restoration capability analysis is performed based on the reversibility index. Specifically, a graph algorithm (such as the Dijkstra algorithm) is used to calculate the shortest path length from the original data field to the output result. The computation types involved in the path (such as addition, multiplication, and aggregation functions) are analyzed by parsing the computational operations in the analysis logic summary or by recording the computational nodes along the path during graph traversal. The number of occurrences of the original data field in all field reconstruction chains is counted as its frequency of participation. Statistical methods (such as correlation analysis and feature importance scoring) are used to assess the importance of the field in the statistical analysis. The stability of the field reconstruction chain, i.e., the reliability of the chain under different scenarios, is evaluated by simulating different analysis scenarios. Based on these feature dimensions, a reversibility index is calculated for each field, helping to understand which data fields are most likely to be restored in collaborative analysis tasks and enabling appropriate privacy protection measures to be implemented. This approach can more accurately identify and protect sensitive data by analyzing the reversibility of data fields in detail. By evaluating the restore capability of data fields, it can better balance data privacy protection and data availability.

[0059] In one possible implementation, the disturbance level control channel is configured using the equilibrium game result, and step S480 further includes step S481, creating a disturbance mapping table, and the disturbance mapping table is mapped one-to-one with the task value. Specifically, the evaluation criteria for task value are defined, including the urgency of the task, the impact on decision-making, the sensitivity of the data, etc. A disturbance level is defined for each task value, and the disturbance level determines the degree of data disturbance. For example, high-value tasks correspond to lower disturbance levels, and low-value tasks correspond to higher disturbance levels. Use a database or configuration file to store these mapping relationships for quick search when executing tasks.

[0060] In step S482, after evaluating the collaborative analysis task value, the perturbation mapping table is used to establish a base perturbation. Specifically, the task value is assessed using a predefined evaluation model or algorithm, such as an expert system, machine learning model, or simple rule engine. Based on the evaluation result, the corresponding perturbation level is searched in the perturbation mapping table and used as the base perturbation.

[0061] Step S483, configure the random perturbation factor, and use the basic perturbation, random perturbation factor and balanced game results to configure the perturbation level control channel. Specifically, in order to increase the randomness of data protection, a random perturbation factor is introduced by generating random numbers or using a random process (such as Gaussian noise). The balanced game results (such as the optimal or satisfactory solution) are applied to the configuration of the perturbation level control channel to ensure that the efficiency of task execution is maximized while protecting privacy. Combining the basic perturbation, random perturbation factor and balanced game results, the perturbation level control channel is configured, including adjusting the data encryption strength, the level of perturbation injection, etc. This implementation method can flexibly protect data privacy according to the different needs and values ​​of the tasks by dynamically adjusting the perturbation level.

[0062] Step S500: After the power grid data is encrypted and encapsulated using the encryption encapsulation solution, the disturbance level control channel is used to perform call disturbance control on the encapsulated data.

[0063] Specifically, the configured perturbation level control channel is used to perform perturbation control on the encrypted and encapsulated data, including introducing a certain degree of random noise into the data through data perturbation algorithms, such as adding Gaussian noise or Laplace noise, to increase the protection of data privacy without significantly affecting the data analysis results.

[0064] Step S600: Generate a collaborative risk prediction of the power grid based on the result of calling disturbance control, and perform user feedback management on the collaborative risk prediction.

[0065] Specifically, based on the data after disturbance control, appropriate analytical models (such as machine learning models) are used to generate collaborative risk forecasts for the power grid, including forecasts of power demand and grid failure risks. The collaborative risk forecast results are fed back to users, and user feedback is collected through user interface design or feedback collection tools to ensure that users can understand and utilize the forecast results. At the same time, the forecast model and analysis process are continuously optimized based on user feedback.

[0066] The embodiment of the present application adopts multi-dimensional attribute vector extraction and encryption graph mapping matching technology. It extracts attribute vectors from power grid data from multiple dimensions such as data sensitivity, historical time series correlation, causal correlation, predicted target variables, and control intervention, and establishes a vector set, and then maps and matches to generate an encryption encapsulation scheme. At the same time, based on the analysis of collaborative analysis tasks, it combines dynamic adaptation analysis to configure a disturbance level control channel, and uses this channel to call disturbance control and other technical means on the encrypted and encapsulated data. It solves the technical problem of the existing cross-regional power grid data privacy protection collaborative analysis that is difficult to balance data privacy protection and data availability, and achieves the technical effect of improving the availability of data in the collaborative analysis process while ensuring data privacy security.

[0067] In the above, refer to Figure 1 The cross-regional power grid data privacy protection collaborative analysis method according to an embodiment of the present invention is described in detail. Figure 2 A cross-regional power grid data privacy protection collaborative analysis system according to an embodiment of the present invention is described.

[0068] The cross-regional power grid data privacy protection collaborative analysis system according to an embodiment of the present invention is designed to address the technical problem of balancing data privacy protection and data availability in existing cross-regional power grid data privacy protection collaborative analysis, thereby achieving the technical effect of improving data availability during the collaborative analysis process while ensuring data privacy security. The cross-regional power grid data privacy protection collaborative analysis system includes: a multidimensional attribute vector extraction module 10, an encryption encapsulation scheme establishment module 20, a collaborative analysis task parsing module 30, a disturbance level control channel configuration module 40, a disturbance control call module 50, and a collaborative risk prediction generation module 60.

[0069] The multidimensional attribute vector extraction module 10 is used to perform multidimensional attribute vector extraction of power grid data and establish a multidimensional attribute vector set. The dimensions of the multidimensional attribute vector extraction include data sensitivity dimension, historical time series correlation dimension, causal correlation dimension, prediction target variable dimension, and control intervention dimension. The encryption encapsulation scheme establishment module 20 is used to perform mapping and matching of the encryption graph using the multidimensional attribute vector set to establish an encryption encapsulation scheme. The collaborative analysis task parsing module 30 is used to parse the collaborative analysis task after receiving it and establish a parsing result. The parsing result includes an analysis logic summary and a data type usage statement. The disturbance level control channel configuration module 40 is used to feed the parsing result back to the data provision layer, perform dynamic adaptive analysis of the power grid data, and configure the disturbance level control channel based on the dynamic adaptive analysis result. The call disturbance control module 50 is used to encrypt and encapsulate the power grid data using the encryption encapsulation scheme and then use the disturbance level control channel to perform call disturbance control of the encapsulated data. The collaborative risk prediction generation module 60 is used to generate a collaborative risk prediction of the power grid based on the call disturbance control result and perform user feedback management on the collaborative risk prediction.

[0070] The specific configuration of the disturbance level control channel configuration module 40 will be described in detail below. As described above, after the analysis result is fed back to the data provision layer, a dynamic adaptive analysis of the power grid data is performed, and the disturbance level control channel is configured based on the dynamic adaptive analysis result. The disturbance level control channel configuration module 40 may further include: a target power grid data positioning unit for locating the target power grid data using the data provision layer; a usage sensitivity calculation unit for performing usage sensitivity calculation based on the target power grid data and the data type usage declaration using the usage sensitivity analysis network of the data provision layer to generate a first adaptive analysis result; an original data restoration capability analysis unit for calling the reversible backtracking network of the data provision layer, sending the analysis logic summary to the reversible backtracking network, and then performing an original data restoration capability analysis of the target power grid data in the collaborative analysis task scenario to establish a second adaptive analysis result; a sensitivity mean acquisition unit for acquiring the sensitivity mean of the multidimensional attribute vector set and establishing a third adaptive analysis result based on the sensitivity mean; and a dynamic adaptive analysis result establishment unit for establishing the dynamic adaptive analysis result based on the first adaptive analysis result, the second adaptive analysis result, and the third adaptive analysis result.

[0071] Among them, the use sensitivity analysis network of the data providing layer is used to perform use sensitivity calculation according to the target power grid data and the data type use statement, and the use sensitivity calculation unit may further include: a preprocessing subunit for converting the data type use statement into a function chain structure by using the preprocessing layer of the use sensitivity analysis network; a field identification subunit for mapping the target power grid data to each module participating in the function chain, and identifying directly used fields, derived reasoning fields, and indirectly affected fields; a use path tension calculation subunit for performing use path tension calculation on each identified field by using the calculation layer of the use sensitivity analysis network, and completing the use sensitivity calculation according to the use path tension calculation result.

[0072] The usage path tension calculation subunit may further include: the calculation layer is as follows: ;in, Characterize the results of sensitivity calculations for usage, is the total number of fields, representing any field, Characterization The basic sensitivity weight of each field, Characterization Whether the field is directly used by the current purpose task, Characterization Whether the field is indirectly used by the task, Characterization The field usage and control conflict of each field, 、 are the weight factors for the indirect use component and the use control conflict component respectively.

[0073] Among them, the disturbance level control channel is configured based on the dynamic adaptation analysis results, and the disturbance level control channel configuration module 40 can further include: a basic trust level generation unit for generating a basic trust level of the target power grid data according to the multidimensional attribute vector set; a usage trust level establishment unit for establishing a usage trust level according to the data type usage declaration; a disturbance level control channel correction unit for constructing a field decision constraint using the basic trust level and the usage trust level, and correcting the disturbance level control channel using the field decision constraint.

[0074] Among them, after the analysis logic summary is sent to the reversible backtracking network, the original data restoration capability analysis of the target power grid data in the collaborative analysis task scenario is performed, and the original data restoration capability analysis unit may further include: a data access graph conversion subunit for extracting the data dependency structure in the analysis logic summary and converting it into a data access graph; a field reconstruction chain construction subunit for using the data access graph to construct a field reconstruction chain, and the field reconstruction chain represents the reverse restoration path; a reversibility index calculation subunit for using the field reconstruction chain to calculate the reversibility index of each original data field, and the calculated feature dimensions include the shortest path length from the original data field to the output result, the calculation feature type in the path, the participation frequency of the original data field, the statistical feature significance, and the stability of the field reconstruction chain, and the restoration capability analysis is completed based on the reversibility index.

[0075] The specific configuration of the collaborative analysis task parsing module 30 will be described in detail below. As described above, after receiving a collaborative analysis task, the collaborative analysis task parsing module 30 may further include: a task intent deep semantic modeling unit for activating a task ontology recognition mechanism and utilizing the task ontology recognition mechanism to perform deep semantic modeling of task intent based on the collaborative analysis task; a conflict detection unit for utilizing the results of the task intent deep semantic modeling to perform conflict detection, including conflict detection for use violations and use drift; and a conflict point prompt warning output unit for utilizing conflict detection to generate conflict point prompt warnings.

[0076] Among them, the disturbance level control channel is configured based on the dynamic adaptation analysis results, and the disturbance level control channel configuration module 40 can further include: an adaptation game model establishment unit for establishing an adaptation game model of task efficiency and risk cost; a balance game unit for synchronizing the dynamic adaptation analysis results to the adaptation game model and executing a balance game; and a disturbance level control channel configuration unit for configuring the disturbance level control channel using the balance game results.

[0077] Among them, the disturbance level control channel is configured using the balance game result, and the disturbance level control channel configuration unit can further include: a disturbance mapping table creation subunit is used to create a disturbance mapping table, and the disturbance mapping table is mapped one-to-one with the task value; a basic disturbance establishment subunit is used to call the disturbance mapping table and establish the basic disturbance after evaluating the task value of the collaborative analysis task; the disturbance level control channel configuration subunit is used to configure the random disturbance factor, and configure the disturbance level control channel using the basic disturbance, the random disturbance factor and the balance game result.

[0078] The cross-regional power grid data privacy protection collaborative analysis system provided by the embodiment of the present invention can execute the cross-regional power grid data privacy protection collaborative analysis method provided by any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0079] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules may be used and run on the user terminal and / or server, and the various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other and are not used to limit the scope of protection of the present invention.

[0080] The above specific embodiments do not constitute a limitation to the scope of protection of this application. It should be understood by those skilled in the art that various modifications, combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent replacements and improvements made within the spirit and principles of this application should be included in the scope of protection of this application. In some cases, the actions or steps recorded in this application can be performed in an order different from that in the embodiments and can still achieve the desired results. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. A collaborative analysis method for cross-regional power grid data privacy protection, characterized by: The method comprises: Perform multidimensional attribute vector extraction on power grid data and establish a multidimensional attribute vector set. The dimensions of the multidimensional attribute vector extraction include data sensitivity, historical time series correlation, causal correlation, predicted target variable, and control intervention. Performing mapping matching of the encrypted graph using the multidimensional attribute vector set to establish an encryption encapsulation scheme; After receiving the collaborative analysis task, the collaborative analysis task is parsed to create a parsing result, wherein the parsing result includes an analysis logic summary and a data type usage statement; After feeding back the analysis results to the data provision layer, performing dynamic adaptive analysis of the power grid data, and configuring the disturbance level control channel based on the dynamic adaptive analysis results; After the power grid data is encrypted and encapsulated using the encryption encapsulation solution, the call disturbance control of the encapsulated data is performed using the disturbance level control channel; A collaborative risk prediction of the power grid is generated according to the result of calling disturbance control, and the collaborative risk prediction is managed for user feedback.

2. The cross-regional power grid data privacy protection collaborative analysis method according to claim 1, characterized in that: After the analysis result is fed back to the data providing layer, a dynamic adaptation analysis of the power grid data is performed, and a disturbance level control channel is configured based on the dynamic adaptation analysis result, including: Locating target power grid data using the data providing layer; Utilizing the usage sensitivity analysis network of the data providing layer to perform usage sensitivity calculation according to the target power grid data and the data type usage statement, to generate a first adaptation analysis result; Invoking the reversible traceability network of the data provision layer, sending the analysis logic summary to the reversible traceability network, performing original data restoration capability analysis of the target power grid data in the collaborative analysis task scenario, and establishing a second adaptation analysis result; Obtaining a sensitivity mean of the multidimensional attribute vector set, and establishing a third adaptation analysis result according to the sensitivity mean; The dynamic adaptation analysis result is established based on the first adaptation analysis result, the second adaptation analysis result, and the third adaptation analysis result.

3. The cross-regional power grid data privacy protection collaborative analysis method according to claim 2, characterized in that: The use sensitivity analysis network using the data providing layer performs use sensitivity calculation according to the target power grid data and the data type use statement, including: Using a preprocessing layer of a usage-sensitive analysis network to convert the data type usage declaration into a function chain structure; Mapping the target power grid data to the modules involved in each functional chain, identifying directly used fields, derived reasoning fields, and indirectly affected fields; The calculation layer of the use sensitivity analysis network is used to calculate the use path tension of each identified field, and the use sensitivity calculation is completed based on the use path tension calculation results.

4. The cross-regional power grid data privacy protection collaborative analysis method according to claim 3, characterized in that: The computational layers are as follows: ； in, Characterize the results of sensitivity calculations for usage, is the total number of fields, Represents any field, Characterization The basic sensitivity weight of each field, Characterization Whether the field is directly used by the current purpose task, Characterization Whether the field is indirectly used by the task, Characterization The field usage and control conflict of each field, 、 are the weight factors for the indirect use component and the use control conflict component respectively.

5. The cross-regional power grid data privacy protection collaborative analysis method according to claim 2, characterized in that: The configuring of the disturbance level control channel based on the dynamic adaptation analysis result includes: generating a basic trust level of target power grid data according to the multidimensional attribute vector set; Establishing a usage trust level based on the data type usage statement; A field decision constraint is constructed using the basic trust level and the purpose trust level, and the disturbance level control channel is corrected using the field decision constraint.

6. The cross-regional power grid data privacy protection collaborative analysis method according to claim 2, characterized in that: After sending the analysis logic summary to the reversible traceability network, performing an analysis of the original data restoration capability of the target power grid data in the collaborative analysis task scenario includes: Extracting the data dependency structure in the analysis logic summary and converting it into a data access graph; Constructing a field reconstruction chain using the data access graph, wherein the field reconstruction chain represents a reverse restoration path; The reversibility index of each original data field is calculated using the field reconstruction chain. The calculated feature dimensions include the shortest path length from the original data field to the output result, the type of calculated features in the path, the participation frequency of the original data field, the statistical feature significance, and the stability of the field reconstruction chain. The reduction ability analysis was performed based on the reversibility index.

7. The cross-regional power grid data privacy protection collaborative analysis method according to claim 1, characterized in that: After receiving the collaborative analysis task, the steps include: Activate the task ontology recognition mechanism and use the task ontology recognition mechanism to perform deep semantic modeling of task intent based on collaborative analysis tasks; Utilize the deep semantic modeling results of task intent to perform conflict detection, including use exceeding authority and use drift; Use conflict detection to report conflict points and provide early warning.

8. The cross-regional power grid data privacy protection collaborative analysis method according to claim 1, characterized in that: The configuration of the disturbance level control channel based on the dynamic adaptation analysis result further includes: Establish an adaptive game model of task efficiency and risk cost; Synchronizing the dynamic adaptation analysis results to the adaptation game model to execute the equilibrium game; The equilibrium game results are used to configure the perturbation hierarchy control channel.

9. The cross-regional power grid data privacy protection collaborative analysis method according to claim 8, characterized in that: The configuration of the disturbance level control channel using the equilibrium game result includes: Creating a disturbance mapping table, wherein the disturbance mapping table is mapped one-to-one with the task value; After evaluating the task value of the collaborative analysis task, calling the perturbation mapping table to establish a basic perturbation; A random perturbation factor is configured, and a perturbation level control channel is configured using the basic perturbation, the random perturbation factor, and the equilibrium game result.

10. A cross-regional power grid data privacy protection collaborative analysis system, characterized by: The system is used to implement the cross-regional power grid data privacy protection collaborative analysis method according to any one of claims 1 to 9, and the system includes: A multidimensional attribute vector extraction module is used to extract multidimensional attribute vectors from power grid data and establish a multidimensional attribute vector set. The dimensions of the multidimensional attribute vector extraction include data sensitivity, historical time series correlation, causal correlation, predicted target variable, and control intervention. An encryption encapsulation scheme establishing module, used to perform mapping matching of the encryption graph using the multi-dimensional attribute vector set to establish an encryption encapsulation scheme; A collaborative analysis task parsing module is used to parse the collaborative analysis task after receiving it and create a parsing result, wherein the parsing result includes an analysis logic summary and a data type usage statement; A disturbance level control channel configuration module is configured to feed back the analysis results to the data provision layer, perform dynamic adaptation analysis of the power grid data, and configure the disturbance level control channel based on the dynamic adaptation analysis results; A calling disturbance control module is used to perform calling disturbance control of the encapsulated data by using the disturbance level control channel after encrypting and encapsulating the power grid data by using the encryption encapsulation solution; The collaborative risk prediction generation module is used to generate a collaborative risk prediction of the power grid according to the result of calling disturbance control, and perform user feedback management on the collaborative risk prediction.

Images