Method for intelligently calculating and deducing generality service of enterprise-level power grid

Abstract

The invention provides a method for intelligently calculating and deducing generality service of an enterprise-level power grid, and belongs to the technical field of intelligent power grids. The method comprises the following steps: accessing and fusing typical scene multi-source heterogeneous data; carrying out enterprise-level power grid intelligent calculation deduction generality service; and carrying out pilot application verification of multiple types of typical scenes, including regional power grid state estimation and weak link analysis. The method can meet the deduction requirements of power grid operation state estimation, weak link analysis and multi-spatio-temporal trend prediction, and supports power grid operation state monitoring and lean management.

Description

Technical Field

[0001] The present invention relates to the field of smart grid technology, and in particular to a method for intelligent computing and deducing common services for enterprise-level power grids. Background Art

[0002] First of all, graph computing is data-centric and adopts a new computing model that closely combines analytical computing with data storage. It has natural advantages in data structure organization and complex association analysis, and has become a disruptive technology that breaks through data management and analytical computing. The graph model naturally supports complex association storage queries, which can effectively avoid the loading of redundant data in multiple relationship queries and support deep association analysis of entities up to 10 layers or more. In terms of storage and computing mode, the graph database query language can seamlessly connect data query and computing, eliminate the time overhead caused by data migration and conversion, and support node parallelism, layered parallelism and other storage and computing integrated computing methods, significantly improving the efficiency of analytical computing. In terms of application ecology, graph computing supports the relative separation of data source layer, data management and data application layer, and promotes the transformation of software development model from application-centric to data-centric, from isolated and closed to open and shared, and has achieved significant success in many fields such as the Internet.

[0003] Large-scale distributed graph computing can be divided into three major factions: MapReduce faction, Pregel faction and GraphLab faction. Most of the existing large-scale graph computing systems are global, batch processing systems, and are oriented to static graph structures. For processing massive graph data, it has good scalability, high fault tolerance and flexibility. In terms of system operation mode, the synchronous control system has high accuracy of operation results, while the restricted asynchronous system has fast execution speed. At present, graph computing systems have a large room for development in terms of different computing models, key technologies and research fields.

[0004] Neo4j is a popular graph database management system, known for its support for graph data structures. Unlike traditional relational databases (such as MySQL, PostgreSQL), Neo4j focuses on storing and managing the relationships between nodes (nodes) and edges (relationships). The core concept of graph databases is that they are particularly suitable for handling highly connected data and complex relational queries.

[0005] In Neo4j, nodes represent entities or objects, and each node can have a label to identify the type of node. The edges connecting nodes represent the relationship between nodes. Relationships are directional (unidirectional / bidirectional) and can have attributes. Nodes or relationships can have attributes to store metadata in the form of key-value pairs. Neo4j uses its own graph query language Cypher for data operations and queries. Cypher syntax is similar to SQL, but it is suitable for graph structures, for example, querying the shortest path between two nodes, finding neighbor nodes of a node, etc. Neo4j supports a series of graph algorithms, such as shortest path, community detection, PageRank, connected components, etc., which are suitable for processing social network analysis, recommendation systems, etc. Graph databases can easily adapt to structured or semi-structured data, and compared with relational databases, Neo4j performs better when processing complex connection queries (such as a friend's friend network). Neo4j supports ACID (atomicity, consistency, isolation, persistence) transactions, which means that it can ensure the reliability of data operations and is suitable for mission-critical applications.

[0006] Secondly, multi-source heterogeneous data processing technology refers to the ability to effectively process data from different data sources and different types on a single system or platform. Source heterogeneous data processing technology aims to integrate, transform, analyze and visualize these heterogeneous data to extract valuable information and insights from them. Data integration and transformation: Integrating data from different data sources into a unified data storage requires operations such as data format conversion, field mapping and data cleaning to ensure data consistency and availability. ETL (Extract, Transform, Load) process: used to extract data from source data, transform and process the data, and then load the data into the target data warehouse or database for further analysis and query. Data standardization: Standardize heterogeneous data so that similar data in different data sources can be compared and analyzed, including unit conversion, time format standardization, classification and coding unification, etc. Data mining and analysis: Use data mining and analysis techniques to discover patterns, trends and associations in data, which can help identify valuable information hidden in heterogeneous data. Multi-source heterogeneous data processing can help organizations obtain more comprehensive, accurate and meaningful information from different data sources, thereby supporting better business decisions and innovation.

[0007] In addition, automated machine learning AutoML aims to simplify and automate the process of building and deploying machine learning models, especially to help non-machine learning experts or reduce the tedious work of experts in model development. The goal of AutoML is to lower the development threshold and accelerate the implementation of machine learning applications by automating steps such as feature engineering, model selection, hyperparameter tuning, and model evaluation. Before the emergence of AutoML, the construction of machine learning models mainly relied on expert knowledge and manual debugging. The traditional machine learning process includes data preprocessing, feature selection, model selection, hyperparameter tuning, model evaluation, and deployment, which is a complex and time-consuming process. In early machine learning research, automated attempts to optimize hyperparameters gradually emerged, such as grid search and random search. These two methods are the simplest hyperparameter optimization techniques, automatically trying multiple hyperparameter combinations, but they are less efficient, especially in high-dimensional space. Around 2010, more efficient hyperparameter optimization methods such as Bayesian Optimization became popular. Bayesian optimization infers the next possible optimal hyperparameter combination through a probabilistic model based on historical results (such as Gaussian process), which converges faster than grid search and random search. At this time, researchers began to realize the importance of automated model selection, feature selection, hyperparameter optimization and other tasks, and gradually formed the concept of AutoML. Early work on automated machine learning focused on reducing the tedious steps of manual parameter adjustment, and tools such as Hyperopt (launched in 2013, using a tree-structured Bayesian optimization algorithm) began to appear for automated hyperparameter tuning. The term AutoML became popular around 2014, and researchers began to formally define this field. AutoML is not limited to hyperparameter tuning, but also includes automation of multiple stages such as model selection, feature engineering, and data preprocessing. At this time, AutoML tools began to gradually emerge and support one-click automatic model development. Some of the earliest well-known tools include: auto-sklearn (2015), TPOT (2016). Since 2018, AutoML research has entered a new stage, especially in the field of deep learning. In addition to feature engineering and hyperparameter optimization, the automated design of neural network architecture has become one of the key directions of AutoML. At this time, some new technologies have begun to play an important role in the AutoML system. Among them, neural architecture search (NAS) is a technology for automated design of neural network architecture, which aims to automatically discover the optimal deep neural network structure using machine learning or evolutionary algorithms. NAS is one of the core breakthroughs of AutoML in the field of deep learning, and aims to replace the time-consuming and complex task of manually designing network architecture.In 2018, AutoKeras was launched as an open source project. It is an open source AutoML tool based on NAS, which is specifically used to automate the design of deep learning models. It aims to provide users with an out-of-the-box tool, especially those without a deep learning background. Summary of the invention

[0008] The purpose of the present invention is to provide a method for enterprise-level power grid intelligent computing deduction common services, which can meet the deduction needs of power grid operation status estimation, weak link analysis and multi-temporal and spatial trend prediction, and support power grid operation status monitoring and lean management.

[0009] In order to achieve the above object, the present invention provides a method for enterprise-level power grid intelligent computing and deducing common services, the method comprising: Access and integrate multi-source heterogeneous data from typical scenarios; Provide common services for enterprise-level power grid intelligent computing and simulation; Carry out pilot application verification of various typical scenarios, including regional power grid status estimation and weak link analysis.

[0010] Preferably, accessing and fusing multi-source heterogeneous data of typical scenarios includes: Acquire data, build a complete power grid diagram data model, and implement unified storage in the Neo4j graph database; at the same time, dynamically update data through the data incremental update mechanism; Based on the large amount of historical load data of each node, an initialization data pool is constructed, and the historical data in the data pool is used as samples for training. The real-time data of the remaining nodes are calculated by combining the training model with the current known measured cross-section real-time data; the historical load data is obtained based on the full amount of smart meter data to obtain the load density curve; the load density curve is sampled by Monte Carlo to obtain the active and reactive power of each node, and the voltage amplitude, phase angle and state quantity of each node are obtained through power flow calculation; The corresponding part of the known quantity measurement is extracted from the calculation results as the training input, and the voltage amplitude and phase angle of each node obtained by calculation are used as the training output to train the deep neural network model. After the training is completed, the model is released.

[0011] Preferably, acquiring data includes: acquiring static grid ledger data from the resource center and asset center of the grid resource business center, acquiring telemetry and telesignaling data of equipment from the real-time measurement center, and acquiring historical measurement data from the data center.

[0012] Preferably, the state quantity includes line transmission power and current.

[0013] Preferably, the common services for enterprise-level power grid intelligent computing deduction include: Based on the highly available Neo4j graph database, combined with the power system standard CIM / E data model and multi-source heterogeneous business data, a unified deduced static topology model is designed; at the same time, a new label relationship representing dynamic topological connections that is different from the static physical connection labels is created; the graph data science library GDS graph algorithm is used to selectively load the content in the graph database into memory; Based on the Autokeras automatic machine learning framework and the Spark batch feature calculation engine, combined with the model parameter search algorithm library including PBT and grid search, an automatic iteration architecture of AutoML and neural architecture search NAS is built. By creating feature engineering tasks, model generation and parameter search tasks, a pipeline is formed from data feature calculation and model generation space limitation to automatic model parameter search. Based on the microservice architecture specifications, Docker lightweight container virtual technology is used to independently encapsulate graph database data services, graph computing services, intelligent inference algorithm services, and inference business application services. Container orchestration tools are used to orchestrate containers. At the same time, clustered deployment and resource scheduling functions are provided for applications based on container orchestration services.

[0014] Preferably, the container orchestration tools include Kubernetes, Docker Swarm and Marathon.

[0015] Preferably, pilot application verification of multiple typical scenarios should include: Select a state estimation scenario, substitute the real-time measurement data into the static graph model, estimate the state of the power grid based on the graph model integrated intelligent computing model and acceleration method, and obtain the current node voltage, phase angle and power distribution of each branch; Select a weak link location scenario, perform multi-time scale power forecasts on all source and load nodes of the power grid based on a multi-space-time deduction model, substitute the single-point prediction results into the state estimation model and simulate random N-1 faults, perform repeated calculations and deductions on the future state of the power flow distribution, and perform safety checks to locate weak links with potential operational risks; Select the power supply path search scenario, perform topology analysis and power supply path analysis on the current state graph model, search for backup power supply paths on the future state graph model, verify safety factors, and select feasible backup power supply paths; Select random fault handling scenarios, sort feasible backup power supply paths based on historical fault handling plans, and intelligently generate random fault handling plans based on the load limits of each component and the located operational weak points.

[0016] Preferably, the safety factors include repeated power outages of equipment, over-limit of dispatching capacity, conflict of power supply tasks, excessive number of households during a single power outage, and repeated power outages of users.

[0017] According to the above technical solution, the large-scale and efficient service of the intelligent computing and deduction model of the power grid is realized through container orchestration technology, supporting multi-business and multi-scenario applications, and providing basic common power grid computing and deduction service capabilities. By uniformly modeling the accessed multi-source heterogeneous data in the graph database, supporting parallel computing of custom algorithms, and improving the overall data storage efficiency and computing performance of the system. By introducing computing components such as the automatic machine learning framework and the model parameter search algorithm library, a model automatic training, iteration and update architecture is constructed to provide common computing services for large-scale model optimization and iteration.

[0018] Other features and advantages of the embodiments of the present invention will be described in detail in the subsequent detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The accompanying drawings are used to provide a further understanding of the embodiments of the present invention and constitute a part of the specification. Together with the following specific embodiments, they are used to explain the embodiments of the present invention, but do not constitute a limitation on the embodiments of the present invention. In the accompanying drawings: Figure 1 It is a flow chart of a method for enterprise-level power grid intelligent computing and deducing common services provided by the present invention. DETAILED DESCRIPTION

[0020] The specific implementation of the embodiment of the present invention is described in detail below in conjunction with the accompanying drawings. It should be understood that the specific implementation described here is only used to illustrate and explain the embodiment of the present invention, and is not used to limit the embodiment of the present invention.

[0021] It should be noted that the acquisition, transmission, storage, use, and processing of data in the technical solution of this application are in compliance with the relevant provisions of national laws and regulations. In the embodiments of this application, some existing solutions in the industry such as certain software, components, and models may be mentioned, which should be considered as exemplary. Their purpose is only to illustrate the feasibility of implementing the technical solution of this application, but it does not mean that the applicant has or will necessarily use the solution.

[0022] See also Figure 1 The present invention provides a method for enterprise-level power grid intelligent computing and deducing common services, the method comprising: Step 1: Access and integrate multi-source heterogeneous data from typical scenarios; Step 2: Provide common services for enterprise-level power grid intelligent computing deduction; Step 3: Carry out pilot application verification of various typical scenarios, including regional power grid status estimation and weak link analysis.

[0023] Specifically, in this embodiment, step 1 includes: Step 1.1: Obtain the static ledger data of the power grid from the resource center and asset center of the power grid resource business platform, obtain the telemetry and telesignaling data of the equipment from the real-time measurement center, and obtain the historical measurement data from the data platform, build a complete power grid graph data model, and realize unified storage in the Neo4j graph database; at the same time, realize dynamic data update through the data incremental update mechanism, and provide data support for intelligent computing and deduction of multi-business links and multi-scenario linkage of the power grid.

[0024] Step 1.2: First, for a large amount of historical load data of each node, an initialization data pool is constructed; the historical data in the data pool is used as samples for training, and the real-time data of the remaining nodes are calculated by combining the training model with the current known measured cross-section real-time data; secondly, the historical load data is obtained based on the full amount of smart meter data to obtain the load density curve; finally, the load density curve is sampled by Monte Carlo to obtain the active and reactive power of each node, and the voltage amplitude and phase angle of each node and the state quantities such as line transmission power and current are obtained through power flow calculation; Step 1.3: Extract the corresponding part of the known quantity measurement from the calculation results as the training input, and use the calculated voltage amplitude and phase angle of each node as the training output. Train the deep neural network model and issue the model after training. In actual operation, the voltage amplitude and phase angle of each node and attribute classification can be calculated by the deep neural network based on the known measurement.

[0025] In this embodiment, step 2 includes: Step 2.1: Based on the highly available Neo4j graph database, combined with the power system standard CIM / E data model and multi-source heterogeneous business data, a unified derivation static topology model is designed; at the same time, a new label relationship representing the dynamic topology connection, which is different from the static physical connection label, is created to realize the monitoring of the switch status and telesignaling information, and realize the construction of a complete dynamic model; the graph data science library GDS graph algorithm is used to selectively load the content in the graph database into the memory, providing public data service support for batch parallel and hybrid computing; Step 2.2: Based on the Autokeras automatic machine learning framework and the Spark batch feature computing engine, combined with the model parameter search algorithm library including PBT and grid search, build an automatic iteration architecture for AutoML and neural architecture search NAS. By creating feature engineering tasks, model generation, and parameter search tasks, a pipeline is formed from data feature calculation and model generation space limitation to automatic model parameter search, supporting model training, iteration, and online incremental update, and providing public computing services for the optimization and iteration of large-scale intelligent computing deduction models. Step 2.3: Based on the microservice architecture specification, the graph database data service, graph computing service, intelligent deduction algorithm service, and deduction business application service are independently encapsulated through Docker lightweight container virtual technology to provide common services with highly decoupled and independently running basic capabilities; use Kubernetes, Docker Swarm, Marathon and other container orchestration tools to orchestrate containers to achieve service lifecycle management; at the same time, based on the container orchestration service, provide clustered deployment and resource scheduling functions for applications, and provide highly available, load-balanced common intelligent algorithms and business application services for intelligent computing deduction.

[0026] In this embodiment, step 3 includes: Step 3.1, select the state estimation scenario, substitute the real-time measurement data into the static graph model, estimate the state of the power grid based on the graph model integrated intelligent computing model and acceleration method, and obtain the current node voltage, phase angle and power distribution of each branch; Step 3.2, select the weak link location scenario, perform multi-time scale power prediction for all source and load nodes of the power grid based on the multi-space-time deduction model, substitute the single-point prediction results into the state estimation model and simulate random N-1 faults, repeatedly calculate and deduce the future state of the power flow distribution, and perform safety verification to locate the weak links with potential operation risks; Step 3.3, select the power supply path search scenario, perform topology analysis and power supply path analysis on the current state graph model, and search for backup power supply paths on the future state graph model, verify safety factors such as repeated power outages of equipment, over-limit of dispatching carrying capacity, conflict of power supply tasks, excessive number of households during a single power outage, and repeated power outages of users, and select feasible backup power supply paths; Step 3.4: Select a random fault handling scenario, sort the feasible backup power supply paths based on historical fault handling plans, consider the load limits of each component and the located operational weak points, and intelligently generate a random fault handling plan.

[0027] Through the above technical solutions, container orchestration technology is used to realize large-scale and efficient service-oriented intelligent computing and deduction models for power grids, support multi-business and multi-scenario applications, and provide basic common power grid computing and deduction service capabilities. By uniformly modeling the accessed multi-source heterogeneous data in the graph database, supporting parallel computing of custom algorithms, the overall data storage efficiency and computing performance of the system are improved. By introducing computing components such as the automatic machine learning framework and the model parameter search algorithm library, an automatic model training, iteration, and update architecture is constructed to provide common computing services for large-scale model optimization and iteration.

[0028] In this way, by designing a highly compatible and scalable enterprise-level grid intelligent computing and deduction common service architecture, we can further use new digital technologies such as artificial intelligence and graph computing to carry out research and development work in the areas of rapid calculation of grid status and accurate deduction of operation trends. We can provide reference data on grid operation status, assist in locating weak links in operation, support dispatchers in making optimized dispatching decisions, improve the intelligent level of multi-business scenario applications such as grid operation and optimized operation, and improve the digital support system for grids.

[0029] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may adopt 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 codes.

[0030] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes 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 generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. 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.

[0031] 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.

[0032] 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.

[0033] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0034] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0035] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0036] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0037] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.

Claims

1. A method for enterprise-level power grid intelligent computing and deducing common services, characterized in that: The method comprises: Access and integrate multi-source heterogeneous data from typical scenarios; Provide common services for enterprise-level power grid intelligent computing and simulation; Carry out pilot application verification of various typical scenarios, including regional power grid status estimation and weak link analysis.

2. The method for enterprise-level power grid intelligent computing and deducing common services according to claim 1, characterized in that: Access and integrate multi-source heterogeneous data in typical scenarios, including: Acquire data, build a complete power grid diagram data model, and implement unified storage in the Neo4j graph database; at the same time, dynamically update data through the data incremental update mechanism; Based on the large amount of historical load data of each node, an initialization data pool is constructed, and the historical data in the data pool is used as samples for training. The real-time data of the remaining nodes are calculated by combining the training model with the current known measured cross-section real-time data; the historical load data is obtained based on the full amount of smart meter data to obtain the load density curve; the load density curve is sampled by Monte Carlo to obtain the active and reactive power of each node, and the voltage amplitude, phase angle and state quantity of each node are obtained through power flow calculation; The corresponding part of the known quantity measurement is extracted from the calculation results as the training input, and the voltage amplitude and phase angle of each node obtained by calculation are used as the training output to train the deep neural network model. After the training is completed, the model is released.

3. The method for enterprise-level power grid intelligent computing and deducing common services according to claim 2, characterized in that: Data acquisition includes: acquiring grid static ledger data from the resource center and asset center of the grid resource business platform, acquiring equipment telemetry and telesignaling data from the real-time measurement center, and acquiring historical measurement data from the data platform.

4. The method for enterprise-level power grid intelligent computing and deducing common services according to claim 2, characterized in that: The state quantities include line transmission power and current.

5. The method for enterprise-level power grid intelligent computing and deduction of common services according to claim 1, characterized in that: Common services for enterprise-level power grid intelligent computing simulation include: Based on the highly available Neo4j graph database, combined with the power system standard CIM / E data model and multi-source heterogeneous business data, a unified deduced static topology model is designed; at the same time, a new label relationship representing dynamic topological connections that is different from the static physical connection labels is created; the graph data science library GDS graph algorithm is used to selectively load the content in the graph database into memory; Based on the Autokeras automatic machine learning framework and the Spark batch feature calculation engine, combined with the model parameter search algorithm library including PBT and grid search, an automatic iteration architecture of AutoML and neural architecture search NAS is built. By creating feature engineering tasks, model generation and parameter search tasks, a pipeline is formed from data feature calculation and model generation space limitation to automatic model parameter search. Based on the microservice architecture specifications, Docker lightweight container virtual technology is used to independently encapsulate graph database data services, graph computing services, intelligent inference algorithm services, and inference business application services. Container orchestration tools are used to orchestrate containers. At the same time, clustered deployment and resource scheduling functions are provided for applications based on container orchestration services.

6. The method for enterprise-level power grid intelligent computing and deduction of common services according to claim 5, characterized in that: The container orchestration tools include Kubernetes, Docker Swarm and Marathon.

7. The method for enterprise-level power grid intelligent computing and deduction of common services according to claim 1, characterized in that: Conduct pilot application verification of multiple typical scenarios, including: Select a state estimation scenario, substitute the real-time measurement data into the static graph model, estimate the state of the power grid based on the graph model integrated intelligent computing model and acceleration method, and obtain the current node voltage, phase angle and power distribution of each branch; Select a weak link location scenario, perform multi-time scale power forecasts on all source and load nodes of the power grid based on a multi-space-time deduction model, substitute the single-point prediction results into the state estimation model and simulate random N-1 faults, perform repeated calculations and deductions on the future state of the power flow distribution, and perform safety checks to locate weak links with potential operational risks; Select the power supply path search scenario, perform topology analysis and power supply path analysis on the current state graph model, search for backup power supply paths on the future state graph model, verify safety factors, and select feasible backup power supply paths; Select random fault handling scenarios, sort feasible backup power supply paths based on historical fault handling plans, and intelligently generate random fault handling plans based on the load limits of each component and the located operational weak points.

8. The method for enterprise-level power grid intelligent computing and deduction of common services according to claim 7, characterized in that: The safety factors include repeated power outages of equipment, exceeding the dispatch capacity, conflicts in power supply tasks, exceeding the number of households during a single power outage, and repeated power outages of users.

Images