Power grid data management system and method based on artificial intelligence data analysis

Abstract

The invention discloses a power grid data management system and method based on artificial intelligence data analysis in the technical field of power grid anomaly detection. The method comprises the steps of obtaining operation data of nodes in a power grid system; calculating an exception comprehensive score of the node based on the operation data by combining an exception detection model deployed by edge computing equipment; determining an abnormal response level of the node based on the abnormal comprehensive score; and triggering a corresponding response strategy based on the exception response level of the node, and performing exception processing on the node. According to the method, the lightweight anomaly detection model is deployed on the edge computing equipment, the power grid operation data is monitored in real time, the abnormal condition is quickly identified, the reliability and safety of power grid operation are improved, and the problems of response delay, network bandwidth limitation, single-point fault and the like of a traditional centralized anomaly detection system are solved.

Description

Technical Field

[0001] The present invention relates to a power grid data management system and method based on artificial intelligence data analysis, belonging to the technical field of power grid anomaly detection. Background Art

[0002] Anomaly detection in the power grid refers to the situation where the power grid is not attacked, but due to the failure of individual or partial components, other devices in the network show deviation from the balance state. Severe power grid anomalies can easily lead to large-scale power outages, resulting in huge economic losses. Therefore, analysts need to detect power grid anomalies quickly and accurately.

[0003] As the scale and complexity of power grids increase, traditional centralized anomaly detection systems have defects such as response delays, network bandwidth limitations, and single point failures, making it impossible to achieve real-time monitoring and rapid response to power grid operation data. Summary of the invention

[0004] The purpose of the present invention is to overcome the shortcomings of the prior art and provide a power grid data governance system and method based on artificial intelligence data analysis, which can realize real-time monitoring and rapid response of power grid operation data by deploying lightweight anomaly detection models on edge computing devices.

[0005] To achieve the above object, the present invention is implemented by adopting the following technical solutions:

[0006] In a first aspect, the present invention provides a power grid data governance method based on artificial intelligence data analysis, comprising:

[0007] Obtaining the operation data of nodes in the power grid system;

[0008] In combination with an anomaly detection model deployed by the edge computing device, a comprehensive anomaly score of the node is calculated based on the operation data;

[0009] Determining the abnormal response level of the node based on the abnormal comprehensive score;

[0010] A corresponding response strategy is triggered based on the abnormal response level of the node, and abnormal processing is performed on the node.

[0011] Furthermore, multiple edge computing devices are deployed at key nodes of the power grid system to form a distributed anomaly detection network. The edge computing devices in the anomaly detection network collaborate through high-speed communication links. When an abnormality is detected in a node, the edge computing device exchanges information with the edge computing device of the adjacent node to verify the authenticity of the abnormality. The anomaly comprehensive score calculation formula is:

[0012]

[0014] Where: S is the abnormal comprehensive score, n is the total number of monitoring parameters, x i is the actual value of the ith monitoring parameter, x i nom is the nominal value of the i-th parameter, Δx i is the allowable deviation threshold of the i-th parameter, w i is the weight coefficient of the i-th parameter.

[0015] Furthermore, the anomaly detection model identifies abnormal patterns and abnormal points in the operation data through machine learning and deep learning technologies, discovers nodes with abnormal attribute values ​​by clustering device attributes, and uses a time series model to detect abnormal changes in the operation status of nodes. When the anomaly detection model detects data anomalies, it uses the reasoning ability of the knowledge graph to infer possible correct values ​​or relationships. The credibility scoring mechanism of the anomaly detection model calculates the credibility score of the source of the operation data. The credibility score calculation formula is:

[0016] T S =β1·R acc +β2·F upd +β3·P exp

[0017] Where: T S R is the credibility score. acc is the historical accuracy of the data source, F upd is the update frequency of the data source, P exp is the professional score of the data source, β1, β2, and β3 are weight coefficients, satisfying β1+β2+β3=1.

[0018] Furthermore, the operating data is semantically modeled through knowledge graph technology, and a unified power grid data model is established to give clear semantics to all operating data. The power grid data model uses natural language processing and machine learning technology to identify entities and relationships in operating data from different systems and formats, maps the original data to the concepts and attributes corresponding to the ontology model, and uses the association relationship of the knowledge graph to fuse entities from different data sources.

[0019] Furthermore, the power grid system uses graph database and graph computing technology to model and store power grid topology, and updates relevant nodes and relationships in the knowledge graph in real time by subscribing to change events of data sources, and adopts incremental update strategies to optimize update efficiency, and provides semantic level queries based on the knowledge graph, and uses the knowledge graph to explore implicit relationships and patterns in operating data.

[0020] Furthermore, during the data fusion and update process, the power grid system uses the semantic constraints and consistency checks of the knowledge graph to detect anomalies and errors in the data. The semantic constraints include attribute constraints and relationship constraints, and the consistency checks include attribute value checks, relationship consistency checks, and time series checks.

[0021] Furthermore, the power grid system is configured with a dynamic visualization platform based on a knowledge graph, which sets different warning levels according to the degree of abnormality and prompts with different colors or icons, and uses the association relationship of the knowledge graph to analyze the equipment and areas that may be affected by the abnormality to provide chain reaction warnings, and uses a deep learning model to predict the probability of equipment failure through a combination of historical data and real-time data to arrange maintenance plans in advance, and when an abnormality is found, a scenario simulation is performed based on the knowledge graph to deduce possible development trends to assist in formulating response strategies.

[0022] In a second aspect, the present invention provides a power grid data management system based on artificial intelligence data analysis, comprising:

[0023] Data acquisition module: obtains the operation data of nodes in the power grid system;

[0024] Anomaly detection module: Combined with the anomaly detection model deployed by the edge computing device, the anomaly comprehensive score of the node is calculated based on the operation data;

[0025] Abnormality determination module: determines the abnormal response level of the node based on the abnormality comprehensive score;

[0026] Exception handling module: triggers the corresponding response strategy based on the abnormal response level of the node and performs exception handling on the node.

[0027] In a third aspect, the present invention provides a power grid data management device based on artificial intelligence data analysis, including a processor and a storage medium;

[0028] The storage medium is used to store instructions;

[0029] The processor is used to operate according to the instructions to execute the steps of any of the methods described above.

[0030] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, which implements the steps of any of the above methods when executed by a processor.

[0031] Compared with the prior art, the present invention has the following beneficial effects:

[0032] 1. The present invention deploys a lightweight anomaly detection model on edge computing devices to monitor power grid operation data in real time, quickly identify abnormal situations, improve the reliability and safety of power grid operation, and solve the problems of response delay, network bandwidth limitation and single point failure in traditional centralized anomaly detection systems;

[0033] 2. The present invention uses knowledge graph technology to semantically model the data of power grid equipment, topology, operating status, etc., establish a unified power grid data model, and realize the standardization and fusion of heterogeneous data from different sources and formats. The knowledge graph technology includes semantic annotation of entities and relationships of power grid equipment, topology and operating status, forming a knowledge graph of the power grid, realizing the semantic layer fusion of data, and integrating power grid data from different data sources through the association relationship of the knowledge graph to solve the problems of data islands and information fragmentation.

[0034] 3. The present invention utilizes the semantic constraints and consistency checks of the knowledge graph to automatically detect and repair anomalies and errors in the data, thereby improving the accuracy and completeness of the data; based on the knowledge graph and real-time data analysis results, it provides a visual display of the grid operation status and decision support information, thereby improving the grid operation efficiency and management level. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] The accompanying drawings constituting a part of the present invention are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the accompanying drawings:

[0036] Figure 1 A flowchart of a power grid data management method based on artificial intelligence data analysis provided in Example 1 of the present invention. DETAILED DESCRIPTION

[0037] The present invention will be described in detail below with reference to the accompanying drawings and in combination with embodiments. It should be noted that the embodiments and features in the embodiments of the present invention can be combined with each other without conflict.

[0038] The following detailed description is an exemplary description, which is intended to provide further detailed description of the present invention. Unless otherwise specified, all technical terms used in the present invention have the same meaning as those generally understood by those skilled in the art to which the present invention belongs. The terms used in the present invention are only for describing specific embodiments, and are not intended to limit the exemplary embodiments according to the present invention.

[0039] Embodiment 1:

[0040] As the scale and complexity of power grids increase, traditional centralized anomaly detection systems face challenges such as response delays, network bandwidth limitations, and single point failures. In order to improve the reliability and security of power grid operation, this paper innovatively proposes a power grid data governance method based on artificial intelligence data analysis. By deploying a lightweight anomaly detection model on edge computing devices, real-time monitoring and rapid response to power grid operation data are achieved. Specifically, the following steps are included:

[0041] 1. Design and optimization of lightweight anomaly detection model

[0042] Adaptive model compression: To address the problem of limited computing resources on edge devices, we use model compression and pruning technology to optimize deep learning models and generate lightweight anomaly detection models. Through automated tools, we adaptively adjust the complexity of the model based on the hardware configuration and real-time load of the device to ensure high detection accuracy even when resources are limited.

[0043] Model compression method: By removing redundant network layers and neurons, the number of model parameters and the amount of computation can be reduced, thereby improving the running efficiency of the model on edge devices.

[0044] Multimodal data fusion: Grid operation data includes multiple parameters such as voltage, current, frequency, phase, etc. Traditional methods often only focus on a single indicator and have difficulty capturing complex abnormal patterns. This invention improves the ability to identify abnormal situations by fusing multimodal data and using advanced feature extraction algorithms.

[0045] Feature fusion: Different types of data are uniformly converted into feature vectors and input into the anomaly detection model to enhance the model's ability to perceive complex anomalies.

[0046] 2. Edge intelligence collaboration mechanism

[0047] Distributed anomaly detection network: Edge computing devices are deployed at each key node of the power grid to form a distributed anomaly detection network. The devices collaborate through high-speed communication links. When a node detects an anomaly, it quickly exchanges information with neighboring nodes to verify the authenticity of the anomaly and reduce the false alarm rate.

[0048] Dynamic load balancing: To address the load imbalance problem that may occur on edge devices, a dynamic task allocation algorithm is designed. When a device is overloaded, some detection tasks can be assigned to neighboring devices to ensure efficient operation of the system.

[0049] 3. Development and implementation of real-time response strategies

[0050] Local fast response: Once the edge device detects an anomaly, it immediately triggers the local preset response strategy, such as automatically cutting off the faulty line, starting the backup power supply, adjusting the load distribution, etc. This localized fast response mechanism greatly shortens the processing time and prevents the spread of anomalies.

[0051] Multi-level response mechanism: According to the severity and impact of the anomaly, a multi-level response plan is formulated. For general anomalies, the edge device handles them by itself; for serious anomalies, it is immediately reported to the regional control center to initiate high-level emergency measures.

[0052] 4. Autonomous learning and updating of abnormal knowledge

[0053] Online learning capability: Edge devices have online learning capabilities and can continuously update anomaly detection models based on newly collected data, allowing the system to adapt to changes in the power grid environment and the emergence of new abnormal patterns.

[0054] Knowledge sharing platform: Establish an abnormal knowledge sharing platform between edge devices. Each device will upload the detected new abnormal patterns and processing experience to the platform, and other devices can download and update them, forming a collaborative abnormal response capability across the entire network.

[0055] 5. Secure and reliable edge computing environment

[0056] Lightweight security protocol: To ensure the security of communication between edge devices, lightweight encryption and authentication protocols are designed to ensure the confidentiality and integrity of communication data under limited resources.

[0057] Fault tolerance and recovery mechanism: Considering the possibility of failure of edge devices, the system is designed with a fault tolerance mechanism. When a device fails, other devices automatically take over its tasks. After the failed device recovers, it automatically synchronizes the latest data and models and rejoins the system.

[0058] 6. Efficient data management and communication optimization

[0059] Data preprocessing and compression: Preprocess the raw data on the edge device to extract key features and reduce the amount of data. Use efficient compression algorithms to reduce the bandwidth requirements for data transmission.

[0060] Event-driven communication mode: Adopting an event-driven communication strategy, communication is performed only when an anomaly is detected or an update is required, reducing unnecessary data exchange and improving communication efficiency. In order to assess the severity of anomalies and guide the formulation of real-time response strategies, this paper proposes an innovative anomaly comprehensive scoring formula:

[0061]

[0062] Where: S is the abnormal comprehensive score, n is the total number of monitoring parameters, xi is the actual value of the ith monitoring parameter (such as voltage, current, etc.), x i nom is the nominal value (normal value) of the i-th parameter, Δx i is the allowable deviation threshold of the i-th parameter, w i is the weight coefficient of the i-th parameter, reflecting the impact of this parameter on the severity of the abnormality.

[0063] Relative deviation calculation: Calculate the deviation between the actual value and the nominal value of each parameter and normalize it relative to the allowable deviation threshold of the parameter;

[0064] Weighted summation: multiply the normalized deviation of each parameter by the corresponding weight coefficient and then sum them to obtain the abnormal comprehensive score S;

[0065] Scoring application: According to the value of S, different response level thresholds are set to trigger corresponding response strategies.

[0066] In power grid data governance, the heterogeneity, complexity and dynamism of data are the core issues that need to be solved urgently. Traditional data processing methods are difficult to effectively unify and integrate power grid data from different systems and formats, resulting in data silos and information fragmentation, which seriously restricts the ability of intelligent operation and decision support of power grids. To this end, the present invention innovatively uses knowledge graph technology to perform semantic modeling on data such as equipment, topology, and operating status of the power grid, establish a unified power grid data model, and realize the standardization and integration of heterogeneous data, including:

[0067] Power grid ontology construction and semantic annotation: Based on the professional knowledge and standards of the power industry, an ontology model (Ontology) in the power grid field is constructed to clearly define the concepts and attributes of various types of equipment in the power grid (such as transformers, switches, and lines), topological relationships (such as connections and dependencies), and operating states (such as voltage, current, and fault states). Through the ontology model, all data are given clear semantics, so that data from different sources can be interpreted and processed under a unified semantic framework.

[0068] Heterogeneous data mapping and conversion: Develop intelligent data mapping tools for data from different systems (such as SCADA, GIS, OMS) and different formats (structured, semi-structured, unstructured). Use natural language processing and machine learning technology to automatically identify entities and relationships in the data, map the original data to the concepts and attributes corresponding to the ontology model, and achieve data semantic unification. The problem of inconsistent data semantics caused by different naming and definitions of the same concept in different systems is solved through a unified ontology model.

[0069] Multi-source data fusion and association: After completing data mapping, the association relationship of the knowledge graph is used to merge entities from different data sources. Through entity disambiguation and relationship reasoning, the problem of inconsistent naming and incomplete attributes of the same entity in different systems is solved, and a globally consistent network of power grid equipment and topological relationships is established. Through automated data mapping and fusion tools, the technical differences between different data sources are overcome and seamless data integration is achieved.

[0070] Modeling of complex topological relationships: For complex topological structures in power grids (such as ring networks and grid networks), knowledge graphs can flexibly represent many-to-many relationships and multi-level structures. Using graph databases and graph computing technologies, the topology of power grids can be accurately modeled and stored, supporting efficient query and analysis of topological structures.

[0071] Dynamic data update and synchronization: The operating status of the power grid changes in real time. For this reason, a dynamic update mechanism for the knowledge graph is designed. By subscribing to the change events of the data source, the relevant nodes and relationships in the knowledge graph are updated in real time to ensure the consistency and timeliness of the data. At the same time, an incremental update strategy is adopted to optimize the update efficiency and reduce the impact on system performance. The real-time dynamic update mechanism ensures the consistency and synchronization of data between various systems and avoids decision-making errors caused by data lag.

[0072] Semantic query and intelligent retrieval: Based on the constructed knowledge graph, a semantic query interface is provided. Users can query through natural language or professional terms. The system uses semantic analysis and reasoning to accurately understand the user's intention and return highly relevant and complete results. This greatly improves the accessibility and utilization efficiency of data.

[0073] Implicit relationship discovery and knowledge reasoning: Use the reasoning ability of knowledge graphs to discover implicit relationships and patterns in data. For example, through path reasoning, potential fault propagation paths between devices can be discovered; through rule reasoning, conflicts between operating status and safety rules can be detected. This provides important support for the safe operation and preventive maintenance of power grids. Traditional data models are difficult to effectively represent the complex topology of power grids, and the multi-relationship, multi-level structure of knowledge graphs successfully solves this problem.

[0074] Data quality control and anomaly detection: In the process of data fusion and update, the semantic constraints and consistency checks of the knowledge graph are used to automatically detect anomalies and errors in the data. For example, isolated nodes and loop errors are detected using topological rules; device parameter anomalies are discovered using attribute constraints. Improve data accuracy and reliability through automated quality control. Automated data quality detection and anomaly handling are achieved using the semantic constraints and reasoning capabilities of the knowledge graph, improving data reliability.

[0075] Cross-domain data integration and expansion: Grid operation needs to consider a variety of external factors, such as weather, geographical environment, user needs, etc. By expanding the grid ontology model and incorporating concepts and relationships from other fields, cross-domain data integration can be achieved. This provides data support for multi-dimensional grid analysis and optimization.

[0076] Scalable architecture design: In order to adapt to the continuous development of power grid business, the architecture design of the knowledge graph has good scalability. New concepts, attributes and relationships can be dynamically added as needed to support the introduction of new businesses and new technologies and ensure the continued effectiveness of the data model.

[0077] The data sources of the power grid are extensive and heterogeneous, involving monitoring systems (SCADA), geographic information systems (GIS), metering systems (AMI), customer information systems (CIS), asset management systems (AMS), etc. These systems are usually developed by different suppliers, using their own technical standards and data formats, forming "information islands". This phenomenon of data fragmentation seriously hinders the comprehensive utilization and intelligent analysis of power grid data. In order to solve this problem, the present invention innovatively uses the association relationship of the knowledge graph to achieve deep fusion and association of power grid data from different data sources. The specific method is as follows:

[0078] 1. Unified semantic modeling of multi-source heterogeneous data

[0079] 1.1 Semantic Analysis and Ontology Construction

[0080] Core entity extraction: Perform in-depth semantic analysis on each data source to extract core entities, such as transformers, switches, lines, substations, users, etc. These entities may have different names and attribute descriptions in different systems.

[0081] Relationship extraction: Identify the relationships between entities, including connection relationships, geographic locations, regions, business associations, etc.

[0082] Ontology model construction: Construct an ontology model in the power grid field to map the concepts of different data sources into a unified semantic space. By defining unified concepts, attributes, and relationships, the semantic inconsistency caused by naming differences and structural differences can be eliminated.

[0083] Cross-domain semantic mapping: It not only unifies the concepts within the power grid, but also semantically maps the concepts of the power grid with those of external fields (such as meteorology and geology) to form a broader knowledge network.

[0084] Adaptive ontology expansion: The ontology model can automatically expand according to the addition of new data sources, support dynamic updates and self-improvement, and solve the problem that traditional static ontology models are difficult to adapt to changes.

[0085] 2. Intelligent Entity Alignment and Disambiguation

[0086] 2.1 Multi-dimensional feature similarity calculation

[0087] Feature extraction: For each entity, multi-dimensional features are extracted, including name, geographic location (latitude and longitude coordinates), technical parameters (such as capacity, voltage level), association relationships (upstream and downstream equipment), etc.

[0088] Similarity calculation: Use machine learning algorithms (such as cosine similarity, Euclidean distance, Mahalanobis distance) to calculate the similarity between entities and evaluate whether they represent the same actual object.

[0089] 2.2 Entity Alignment and Disambiguation

[0090] Entity alignment: Based on the preset similarity threshold, entities with similarity higher than the threshold are aligned as the same entity to achieve entity unification across data sources.

[0091] Context analysis: For entities with the same name but different actual entities (such as different substations with the same name), accurate entity disambiguation is performed by analyzing their association relationships and context information to avoid confusion.

[0092] Dynamic weight adjustment: Dynamically adjust feature weights according to different data sources and entity types to improve the accuracy of similarity calculation and adapt to the needs of different scenarios.

[0093] Application of deep learning models: Use deep learning models (such as neural networks, GNN) to capture complex relationships between entities, improve the accuracy of entity alignment and disambiguation, and break through the limitations of traditional methods.

[0094] 3. Association and fusion of complex relationships

[0095] 3.1 Multi-level relationship modeling

[0096] Physical relationships: such as electrical connections and topological structures, which represent the actual electrical connection methods and paths between devices.

[0097] Spatial relationships: such as geographic location and spatial proximity, which indicate the distribution and relative positions of devices in physical space.

[0098] Business relationships: such as maintenance responsibility, scheduling control, and customer service, indicating the relationship and role of equipment in the business process.

[0099] 3.2 Relationship Fusion

[0100] Unified relationship model: Different types of relationships are uniformly modeled in the knowledge graph to form a multi-level, multi-dimensional relationship network that fully reflects the relationship between devices.

[0101] Relational reasoning: Use the reasoning ability of the knowledge graph to discover new relationships and dependencies. For example, based on spatial proximity and electrical connection relationships, potential fault propagation paths can be inferred to improve fault warning capabilities.

[0102] Relationship Weight and Reliability: Assign weights and reliability scores to different relationships to reflect their importance and trustworthiness, supporting more sophisticated and accurate analysis.

[0103] Complex network analysis: Apply complex network theory to conduct community division and centrality analysis on relationship networks, identify key equipment and vulnerable links, and improve the safety and stability of the power grid.

[0104] 4. Dynamic Fusion of Spatiotemporal Data

[0105] 4.1 Introduction of the time dimension

[0106] Timestamp management: Add an accurate timestamp to each data record to support timing analysis and historical tracing of data.

[0107] Historical data management: Save historical versions of data, support tracking of data changes and trend analysis, and provide a basis for prediction and decision-making.

[0108] 4.2 Introduction of spatial dimension

[0109] Unification of spatial coordinates: Standardize the geographic location information of different data sources into a unified coordinate system to facilitate spatial analysis.

[0110] Spatial relationship calculation: Use GIS technology to calculate the spatial relationship between devices, such as distance and direction, to provide support for spatial correlation analysis.

[0111] Spatiotemporal correlation analysis: Establish the correlation between time and space in the knowledge graph, support the spatiotemporal trajectory analysis of events, and improve the ability to perceive dynamic changes.

[0112] Time series prediction model: Combine time series data with deep learning models (such as LSTM and Transformer) to predict load demand, failure rate, etc. to support forward-looking decision-making.

[0113] 5. Data quality assessment and automatic repair

[0114] 5.1 Data Conflict Detection

[0115] Attribute conflict identification: When the same entity has inconsistent attribute values ​​in different data sources, the system automatically marks the conflict and prompts that it needs to be processed.

[0116] Relationship conflict detection: If there are conflicts between the relationships between entities in different data sources (such as inconsistent connection relationships), the system will mark and record them.

[0117] 5.2 Automatic repair strategy

[0118] Rule engine application: Establish a data repair rule library to automatically repair common data errors based on business logic and experience to improve efficiency.

[0119] Machine learning model: Train the data repair model to automatically predict and correct incorrect attribute values ​​and relationships using historical data and expert annotations.

[0120] Credibility fusion model: It comprehensively considers the credibility of the data source, the freshness of the data, and the historical accuracy, calculates the comprehensive credibility of the attribute value, and selects the most credible value for repair.

[0121] Interactive repair: For complex or high-risk data repairs, the system prompts the data administrator to conduct manual review and provide repair suggestions to ensure the accuracy and reliability of data repair.

[0122] 6. Mining of Implicit Relationships

[0123] 6.1 Association Rule Mining

[0124] Data mining algorithm: Use association rule mining algorithms (such as Apriori and FP-Growth) to discover potential associations between devices and identify frequent item sets.

[0125] Frequent pattern recognition: By analyzing historical data, we can discover frequently occurring patterns and mine implicit business rules and abnormal patterns.

[0126] 6.2 Knowledge Graph Reasoning

[0127] Inference rule definition: Define inference rules in the knowledge graph, use logical reasoning to discover new relationships and dependencies, and improve the intelligence level of the system.

[0128] Fault correlation analysis: By analyzing equipment operation data and environmental data, possible fault correlations can be discovered, and early warnings can be issued to prevent accidents.

[0129] Multi-dimensional data fusion mining: Combine power grid data, environmental data, and social data to conduct multi-dimensional correlation mining to discover deeper correlations and influencing factors.

[0130] Adaptive reasoning engine: The reasoning engine can automatically update reasoning rules based on new data, continuously improve reasoning capabilities, and adapt to the ever-changing environment.

[0131] 7. Integration and expansion of cross-domain data

[0132] 7.1 Introduction of External Data

[0133] Meteorological data fusion: Introduce meteorological data, such as temperature, humidity, wind speed, rainfall, etc., to analyze the impact of weather on power grid operation and support disaster prevention and mitigation.

[0134] Geological data fusion: Introduce geological information, such as topography and soil type, to evaluate the impact of geological conditions on equipment site selection and line planning, and improve project quality.

[0135] 7.2 Cross-domain association modeling

[0136] Establishing association relationships: In the knowledge graph, establish association relationships between power grid entities and external data entities to achieve cross-domain association and collaboration.

[0137] Comprehensive analysis application: Utilize integrated data to carry out load forecasting, disaster warning, market analysis and other applications to improve the intelligence level of power grid operation.

[0138] Dynamic environmental perception: Real-time acquisition of external environmental data, combined with grid operation data, to achieve dynamic perception and response to grid status and enhance adaptability.

[0139] Intelligent decision support: Based on the integration of cross-domain data, it provides intelligent decision support for power grid planning, dispatching, operation and maintenance, and improves management efficiency

[0140] 8. Flexible query and analysis support

[0141] 8.1 Diverse query methods

[0142] Natural language query: Users can use natural language to make query requests, and the system uses natural language processing technology to parse query intent and lower the usage threshold.

[0143] Graphical query: Provides a visual query interface where users can construct query conditions by dragging, clicking, etc., which is intuitive and convenient.

[0144] 8.2 Efficient Query Processing

[0145] Index optimization: Establish an efficient index structure in the knowledge graph to improve query response speed and meet real-time requirements.

[0146] Parallel query processing: Utilizes a distributed computing framework to support fast query and analysis of large-scale data and improve system performance.

[0147] Intelligent question-answering system: Build an intelligent question-answering system based on knowledge graph that can understand complex business problems, provide accurate answers, and support decision-making.

[0148] Personalized recommendations: Provide personalized data views and analysis results based on the user's query history and role to improve user experience and work efficiency.

[0149] 9.Automated data update and synchronization

[0150] 9.1 Data Source Monitoring Mechanism

[0151] Real-time data capture: Establish a monitoring mechanism for the data source to capture data addition, modification, and deletion operations in real time to ensure the timeliness of the data.

[0152] Event-driven update: Based on data change events, the update process of the knowledge graph is triggered to keep the data consistent and up-to-date.

[0153] 9.2 Incremental Update Strategy

[0154] Minimize update scope: Update only the affected entities and relationships to reduce system overhead and improve update efficiency.

[0155] Conflict detection and resolution: During the update process, data conflicts are detected and handled using the aforementioned data quality assessment and automatic repair strategies to ensure data quality.

[0156] Version management and backtracking: Support version management of knowledge graphs, record the content and reasons of each update, meet audit and backtracking requirements, and enhance system transparency.

[0157] Consistency guarantee mechanism: In a distributed environment, a consistency protocol is used to ensure the atomicity and consistency of data updates and prevent data disorder.

[0158] 10. Security and permission control mechanism

[0159] 10.1 Fine-grained permission control

[0160] Role-based access control (RBAC): Grant corresponding data access rights according to the user's role and responsibilities to prevent unauthorized access.

[0161] Attribute-level permission settings: You can set access permissions for specific attributes of an entity to protect sensitive information and meet security requirements.

[0162] In the process of power grid data governance, the quality of data directly affects the safety and efficiency of power grid operation. However, the sources of power grid data are diverse, involving various types such as equipment ledgers, operating status, topological structures, etc., and the amount of data is huge and updated frequently. There may be errors, redundancies, omissions and inconsistencies in the data. Traditional data quality management methods often rely on manual inspection and simple rule verification, which cannot meet the requirements of modern power grids for data accuracy and real-time performance. To this end, the present invention innovatively uses the semantic constraints and consistency verification of knowledge graphs to automatically detect and repair anomalies and errors in the data, thereby improving the accuracy and integrity of the data. The specific method includes the following steps:

[0163] 1. Build a data quality constraint model

[0164] 1.1 Semantic Constraint Definition

[0165] In the ontology model of the knowledge graph, the attribute constraints and relationship constraints of each entity in the power grid are clearly defined. These constraints not only cover the format and scope of the data, but also involve deeper business logic and physical laws. For example:

[0166] Property constraint: The transformer capacity C must be within the rated range, that is, C min ≤C≤C max , where C min is the minimum rated capacity of the transformer, C max It is the maximum rated capacity of the transformer.

[0167] Relationship constraint: Both ends of a line must connect valid nodes, i.e., if line L connects nodes N1 and N2, then both N1 and N2 should exist and be valid.

[0168] 1.2 Rules and logical reasoning

[0169] Use description logic and rule languages ​​(such as OWL and SWRL) to define complex business rules and reasoning logic. For example:

[0170] Overload detection rule: If the current I of line L L Exceeding its rated current I rated , an overload warning is triggered.

[0171] If I L >I rated → Trigger overload warning

[0172] Geographical consistency: If there is an electrical connection between two devices, their geographical locations should meet certain distance conditions:

[0173] If connection (E1, E2) → distance (E1, E2) ≤ maximum distance dmax of geographic location

[0174] 2. Automated anomaly detection mechanism

[0175] 2.1 Multi-dimensional consistency check

[0176] Design corresponding consistency verification algorithms for different dimensions of data:

[0177] Attribute value verification: check whether the parameters are within a reasonable range; relationship consistency verification: check whether the topological connection is logical; time series verification: check whether the timestamp of the data is continuous and reasonable.

[0178] 2.2 Abnormal pattern recognition

[0179] Using machine learning and deep learning techniques, we train anomaly detection models to identify abnormal patterns and outliers in data. For example:

[0180] Cluster analysis: cluster device attributes to find devices with abnormal attribute values; time series model: use LSTM and other time series models to detect abnormal changes in operating status.

[0181] 3. Intelligent data repair strategy

[0182] 3.1 Reasoning-based automatic repair

[0183] When data anomalies are detected, the reasoning ability of the knowledge graph is used to automatically infer possible correct values ​​or relationships. For example:

[0184] Geographic location repair: If the geographic location of device E is found to be abnormal, a reasonable location P can be inferred based on its connection relationship with neighboring devices. correct .

[0185] 3.2 Data source credibility assessment

[0186] Introduce a credibility scoring mechanism for data sources. For conflicts in the same data item, the more credible data source information is used first. The credibility score can be calculated based on factors such as the historical accuracy, update frequency, and professionalism of the data source:

[0187] 3.3 Interactive repair and feedback

[0188] For complex anomalies that cannot be automatically repaired, the system generates repair suggestions and prompts data managers to manually confirm and process them. The repair results will be fed back to the system to further optimize the anomaly detection and repair model.

[0189] 4. Data version control and traceability

[0190] 4.1 Fine-grained data version management

[0191] A version record is kept for each modification of the data, including: modification time t, modifier u, modification content △d, and modification reason r. When data problems occur, the evolution of the data can be traced back to facilitate problem location and responsibility division.

[0192] 4.2 Data traceability mechanism

[0193] Through the relationship tracking capability of the knowledge graph, we can understand the source, references and impact scope of each data item. For example, if the device parameter p is modified, we can immediately know the affected topological relationship and operating status evaluation.

[0194] 5. Dynamic data quality monitoring

[0195] 5.1 Real-time monitoring and early warning

[0196] The system monitors key data indicators in real time and sets thresholds θ and alarm rules. Once the data deviates from the normal range, an early warning is triggered immediately to prevent decision-making errors caused by erroneous data.

[0197] 5.2 Data Quality Indicator Evaluation

[0198] Establish a data quality evaluation system that includes the following indicators: completeness C, accuracy A, consistency S, and timeliness T. Generate data quality reports regularly to help managers fully understand the health of the data.

[0199] 6. Self-learning and continuous optimization

[0200] 6.1 Model Adaptive Update

[0201] The system automatically updates anomaly detection and repair models based on new data and feedback, improving detection accuracy and repair success rate.

[0202] 6.2 Construction of abnormal pattern library

[0203] Collect and organize discovered abnormal cases and build an abnormal pattern library. New data can be matched with the pattern library to quickly identify known anomalies and improve detection efficiency.

[0204] 7. Collaborative data quality governance

[0205] 7.1 Cross-departmental collaboration mechanism

[0206] Establish a collaborative process for handling data quality issues, involving multiple departments such as equipment maintenance, operation scheduling, and data management. Share data quality information through knowledge graphs to promote collaborative governance among departments.

[0207] 7.2 Data Governance with User Participation

[0208] Some data quality issues are open to business users for feedback. For example, field maintenance personnel can report abnormal equipment information through mobile terminals, and the system will incorporate the information into the data quality management process. In order to comprehensively evaluate the credibility of data sources, the following data source credibility scoring formula is proposed:

[0209] T S =β1·R acc +β2·F upd +β3·P exp

[0210] Where: T S R is the credibility score. acc is the historical accuracy of the data source, reflecting the accuracy of the data provided by the data source in the past. updis the update frequency of the data source, indicating the timeliness and activity of the data source update. exp Score the professionalism of the data source to evaluate the professional level and credibility of the data source. β1, β2, and β3 are weight coefficients, satisfying β1+β2+β3=1. When dealing with data conflicts, T is preferred. S The information of data sources with higher value ensures data quality.

[0211] In modern power grid operation, facing massive multi-source heterogeneous data, how to intuitively present complex information to decision makers has become the key to improving power grid operation efficiency and management level. However, traditional data visualization methods often cannot fully reveal the complex relationships within the power grid, and decision support systems also lack the ability to conduct in-depth analysis of real-time data. To this end, the present invention innovatively provides a visual display of the power grid operation status and decision support information based on knowledge graphs and real-time data analysis. The specific method is as follows:

[0212] 1. Build a dynamic visualization platform based on knowledge graph

[0213] 1.1 Multidimensional Data Fusion and Semantic Modeling

[0214] Unified data model: Use knowledge graph technology to perform semantic modeling on multi-source heterogeneous data from SCADA, GIS, AMI and other systems, establish a unified data model, and solve the problems of inconsistent data formats and semantics.

[0215] Multi-dimensional data association: Through the knowledge graph, multi-dimensional data such as equipment attributes, topological structure, geographic location, and operating status are associated to form a comprehensive power grid knowledge network.

[0216] 1.2 Dynamic Visualization

[0217] Adaptive visualization interface: Design a visualization interface that can automatically adjust according to data changes to achieve real-time display of power grid operation status.

[0218] Multi-level information presentation: Provides multi-level visualization such as global view, regional view, and device view. Users can zoom in or out the viewing range as needed to gain in-depth understanding of specific information.

[0219] Knowledge-driven visual mapping: Utilize the semantic relationships in the knowledge graph to dynamically generate visual mapping rules, making data presentation more intelligent and intuitive.

[0220] Intelligent layout algorithm: Develop a layout algorithm based on topology and geographic information to automatically optimize the visual location of equipment and lines, reduce visual clutter, and improve the efficiency of information transmission.

[0221] 2. Real-time data analysis and intelligent early warning

[0222] 2.1 Real-time data stream processing

[0223] High-performance data processing engine: Introducing real-time stream processing technologies, such as Spark Streaming and Flink, to process real-time data of the power grid at high speed to ensure the timeliness of the data.

[0224] Anomaly detection and analysis: Build a real-time anomaly detection model and use machine learning algorithms to identify voltage fluctuations, load anomalies, and other situations to discover potential problems in a timely manner.

[0225] 2.2 Intelligent early warning mechanism

[0226] Multi-level warning system: Different warning levels (such as general, serious, and urgent) are set according to the degree of abnormality, and are prompted with different colors or icons in the visual interface.

[0227] Correlation warning: Utilize the correlation relationship of the knowledge graph to analyze the equipment and areas that may be affected by the anomaly, provide chain reaction warning, and help decision makers comprehensively assess risks.

[0228] Predictive maintenance: By combining historical data with real-time data and using deep learning models to predict the probability of equipment failure, maintenance plans can be arranged in advance to reduce sudden failures.

[0229] Scenario simulation and deduction: When an anomaly is discovered, the system can perform scenario simulation based on the knowledge graph, deduce possible development trends, and assist in formulating response strategies.

[0230] 3. Intelligent decision support system

[0231] 3.1 Personalized decision-making recommendations

[0232] Role customization: Provide customized decision support information based on user roles (such as dispatcher, operation and maintenance engineer, and management) to meet the needs of different levels.

[0233] Situational awareness: Combines real-time data and external environmental information (such as weather and load forecasts) to provide the best action recommendations that match the current situation.

[0234] 3.2 Intelligent evaluation of multiple solutions

[0235] Decision model library: Establish a decision model library covering load scheduling, fault handling, maintenance planning, etc. to support multiple business scenarios.

[0236] Solution optimization and evaluation: Use optimization algorithms and simulation technology to automatically generate multiple feasible solutions, evaluate their effects and risks, and assist decision makers in selecting the optimal solution.

[0237] Reasoning engine based on knowledge graph: Develop a reasoning engine that uses the semantics and logical relationships of the knowledge graph to automatically deduce complex decisions and improve the level of intelligent decision-making.

[0238] Collaborative decision support: Supports multi-user online collaboration, and enables cross-departmental collaborative decision-making and improved response speed by sharing visual information and analysis results.

[0239] 4. Application of advanced visualization technology

[0240] 4.1 3D Visualization and Virtual Reality

[0241] 3D power grid model: Build a three-dimensional model of the power grid, including substations, lines, equipment, etc., to provide a more intuitive spatial display.

[0242] Virtual reality (VR) technology: VR devices are used to allow users to immersively experience the power grid operating environment, support remote inspection and training, and improve work efficiency.

[0243] 4.2 Augmented Reality (AR) Applications

[0244] On-site information overlay: On mobile devices, the camera captures the on-site images and overlays information such as equipment operating status and maintenance records to assist on-site staff in operations.

[0245] Real-time data interaction: Staff can interact with the system through gestures, voice, etc. to obtain the required information, improving the convenience of information acquisition.

[0246] Multimodal interaction: Integrates multiple interaction methods such as voice, gesture, and touch to provide a more natural and efficient human-computer interaction experience.

[0247] Visual customization tool: Develop a visual configuration tool that allows users to customize the visual interface and data display method according to their own needs to meet personalized requirements.

[0248] In summary, this method uses knowledge graph technology to perform semantic modeling on data such as equipment, topological structure, and operating status of the power grid, establish a unified power grid data model, and realize the standardization and integration of heterogeneous data from different sources and formats. Knowledge graph technology includes semantic annotation of entities and relationships of power grid equipment, topological relationships, and operating status to form a knowledge graph of the power grid and realize semantic layer integration of data; through the association relationship of the knowledge graph, the power grid data from different data sources are integrated to solve the problems of data islands and information fragmentation; a lightweight anomaly detection model is deployed on the edge computing device to monitor the power grid operation data in real time, quickly identify abnormal situations, and improve the reliability and safety of the power grid operation; the semantic constraints and consistency checks of the knowledge graph are used to automatically detect and repair anomalies and errors in the data to improve the accuracy and completeness of the data; based on the knowledge graph and real-time data analysis results, a visual display of the power grid operation status and decision support information are provided to improve the operation efficiency and management level of the power grid.

[0249] Embodiment 2:

[0250] A power grid data management system based on artificial intelligence data analysis can implement a power grid data management method based on artificial intelligence data analysis described in Example 1, including:

[0251] Data acquisition module: obtains the operation data of nodes in the power grid system;

[0252] Anomaly detection module: Combined with the anomaly detection model deployed by the edge computing device, the anomaly comprehensive score of the node is calculated based on the operation data;

[0253] Abnormality determination module: determines the abnormal response level of the node based on the abnormality comprehensive score;

[0254] Exception handling module: triggers the corresponding response strategy based on the abnormal response level of the node and performs exception handling on the node.

[0255] Embodiment three:

[0256] The embodiment of the present invention further provides a power grid data management device based on artificial intelligence data analysis, which can implement the power grid data management method based on artificial intelligence data analysis described in the first embodiment, including a processor and a storage medium;

[0257] The storage medium is used to store instructions;

[0258] The processor is used to operate according to the instructions to perform the steps of the following method:

[0259] Obtaining the operation data of nodes in the power grid system;

[0260] In combination with an anomaly detection model deployed by the edge computing device, a comprehensive anomaly score of the node is calculated based on the operation data;

[0261] Determining the abnormal response level of the node based on the abnormal comprehensive score;

[0262] A corresponding response strategy is triggered based on the abnormal response level of the node, and abnormal processing is performed on the node.

[0263] Embodiment 4:

[0264] The embodiment of the present invention further provides a computer-readable storage medium, which can implement the power grid data management method based on artificial intelligence data analysis described in the first embodiment, and a computer program is stored thereon, and when the program is executed by a processor, the steps of the following method are implemented:

[0265] Obtaining the operation data of nodes in the power grid system;

[0266] In combination with an anomaly detection model deployed by the edge computing device, a comprehensive anomaly score of the node is calculated based on the operation data;

[0267] Determining the abnormal response level of the node based on the abnormal comprehensive score;

[0268] A corresponding response strategy is triggered based on the abnormal response level of the node, and abnormal processing is performed on the node.

[0269] It is known from common technical knowledge that the present invention can be implemented by other embodiments that do not deviate from its spirit or essential features. Therefore, the above disclosed embodiments are only illustrative in all respects and are not exclusive. All changes within the scope of the present invention or within the scope equivalent to the present invention are included in the present invention.

[0270] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0271] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems) and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the process in the flowchart. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0272] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0273] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0274] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A power grid data governance method based on artificial intelligence data analysis, characterized in that: include: Obtaining the operation data of nodes in the power grid system; In combination with an anomaly detection model deployed by the edge computing device, a comprehensive anomaly score of the node is calculated based on the operation data; Determining the abnormal response level of the node based on the abnormal comprehensive score; A corresponding response strategy is triggered based on the abnormal response level of the node, and abnormal processing is performed on the node.

2. The power grid data management method based on artificial intelligence data analysis according to claim 1 is characterized in that: Multiple edge computing devices are deployed at key nodes of the power grid system to form a distributed anomaly detection network. The edge computing devices in the anomaly detection network collaborate through high-speed communication links. When a node anomaly is detected, the edge computing device exchanges information with the edge computing devices of adjacent nodes to verify the authenticity of the anomaly. The anomaly comprehensive score calculation formula is: Where: S is the abnormal comprehensive score, n is the total number of monitoring parameters, x i is the actual value of the ith monitoring parameter, x i nom is the nominal value of the i-th parameter, Δx i is the allowable deviation threshold of the i-th parameter, w i is the weight coefficient of the i-th parameter.

3. The power grid data management method based on artificial intelligence data analysis according to claim 1 is characterized in that: The anomaly detection model uses machine learning and deep learning technologies to identify abnormal patterns and abnormal points in operating data, clusters device attributes to find nodes with abnormal attribute values, and uses a time series model to detect abnormal changes in node operating status. When the anomaly detection model detects data anomalies, it uses the reasoning ability of the knowledge graph to infer possible correct values ​​or relationships. The credibility scoring mechanism of the anomaly detection model calculates the credibility score of the source of the operating data. The credibility score calculation formula is: T S =β1·R acc +β2·F upd +β3·P exp Where: T S R is the credibility score. acc is the historical accuracy of the data source, F upd is the update frequency of the data source, P exp is the professional score of the data source, β1, β2, and β3 are weight coefficients, satisfying β1+β2+β3=1.

4. The power grid data management method based on artificial intelligence data analysis according to claim 1 is characterized in that: The operation data is semantically modeled through knowledge graph technology, and a unified power grid data model is established to give clear semantics to all operation data. The power grid data model uses natural language processing and machine learning technology to identify entities and relationships in operation data of different systems and formats, maps the original data to the concepts and attributes corresponding to the ontology model, and uses the association relationship of the knowledge graph to fuse entities from different data sources.

5. The power grid data management method based on artificial intelligence data analysis according to claim 1 is characterized in that: The power grid system uses a graph database and graph computing technology to model and store the power grid topology, and updates the relevant nodes and relationships in the knowledge graph in real time by subscribing to change events of the data source, and uses an incremental update strategy to optimize the update efficiency, and provides semantic level queries based on the knowledge graph, and uses the knowledge graph to explore implicit relationships and patterns in the operating data.

6. The power grid data management method based on artificial intelligence data analysis according to claim 1 is characterized in that: During the data fusion and update process, the power grid system uses the semantic constraints and consistency checks of the knowledge graph to detect anomalies and errors in the data. The semantic constraints include attribute constraints and relationship constraints, and the consistency checks include attribute value checks, relationship consistency checks, and time series checks.

7. The power grid data management method based on artificial intelligence data analysis according to claim 1 is characterized in that: The power grid system is configured with a dynamic visualization platform based on a knowledge graph. The dynamic visualization platform sets different warning levels according to the degree of abnormality and prompts with different colors or icons. It also uses the association relationship of the knowledge graph to analyze the equipment and areas that may be affected by the abnormality to provide chain reaction warnings. It also uses a deep learning model to predict the probability of equipment failure through a combination of historical data and real-time data to arrange maintenance plans in advance. In addition, when an abnormality is found, scenario simulation is performed based on the knowledge graph to deduce possible development trends and assist in formulating response strategies.

8. A power grid data management system based on artificial intelligence data analysis, characterized by: include: Data acquisition module: obtains the operation data of nodes in the power grid system; Anomaly detection module: Combined with the anomaly detection model deployed by the edge computing device, the anomaly comprehensive score of the node is calculated based on the operation data; Abnormality determination module: determines the abnormal response level of the node based on the abnormality comprehensive score; Exception handling module: triggers the corresponding response strategy based on the abnormal response level of the node and performs exception handling on the node.

9. A power grid data management device based on artificial intelligence data analysis, characterized in that: including processor and storage medium; The storage medium is used to store instructions; The processor is configured to operate according to the instructions to execute the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images