An all-process intelligent management and control method for power grid basic operation

Abstract

The embodiment of the application provides a kind of whole process intelligent management and control method for power grid basic operation, belong to power grid operation process field.The method includes: based on the positioning algorithm of DOM structure feature and visual feature weighted fusion realizes cross-level page access and non-intrusive front-end data acquisition;Based on the result of cross-level page access and non-intrusive front-end data acquisition, through URL routing rules and page content feature analysis, the boundaries of different menus or function blocks in the original business system are identified and cut, and encapsulated as micro application;Based on demand, dynamically layout micro application;After dynamic layout, micro application obtains the multi-source data of power grid basic operation according to demand;Based on the multi-source data obtained, generate a work report for specific needs.The method can solve the technical problems of frequent multi-system switching, long business access path, complicated operation steps and low efficiency of manual operation in existing basic operation.

Description

Technical Field

[0001] This invention relates to the field of power grid operation processes, and more specifically to a full-process intelligent control method for power grid grassroots operations. Background Technology

[0002] The grassroots operations in the power industry involve multiple professional businesses such as marketing, data collection, and operation management. Each business corresponds to an independent digital system platform. Digital technology has been widely used in grassroots operations and has become an important means to support the development of grassroots businesses. The industry has put forward higher requirements for the efficiency of digital collaboration and the convenience of data application in grassroots operations.

[0003] In existing power grid grassroots operations, multiple specialized business systems operate in parallel, such as the Marketing 2.0 system and the Data Acquisition 2.0 system. These systems are independent of each other, creating technical and data barriers. Grassroots personnel must manually log into the corresponding systems according to different business needs, resulting in frequent system switching. Furthermore, each system has multiple menu levels and long access paths for business functions, requiring multiple manual clicks to complete a single operation. Simultaneously, when grassroots personnel collaborate across disciplines, data acquisition requires querying different systems separately. Business data with high real-time requirements lacks convenient access channels, relying on original vendors for system integration development, which is time-consuming and lacks sufficient support. Data processing and report compilation require manual querying and inputting of information across multiple systems, followed by manual calculation and backfilling of results, consuming significant manpower and time. Therefore, a comprehensive intelligent management and control method for power grid grassroots operations is needed to address the technical problems of frequent system switching, long access paths, cumbersome operation steps, and low efficiency of manual operations in existing grassroots operations. Summary of the Invention The purpose of this invention is to provide a full-process intelligent management and control method for power grid grassroots operations. This method can solve the technical problems of frequent switching between multiple systems, long business access paths, cumbersome operation steps, and low efficiency of manual operation in existing grassroots operations.

[0004] To achieve the above objectives, embodiments of the present invention provide a full-process intelligent management and control method for basic power grid operations, the method comprising: A positioning algorithm based on weighted fusion of DOM structure features and visual features enables cross-level page access and non-intrusive front-end data retrieval; Based on the results of cross-level page access and non-intrusive front-end data retrieval, the boundaries of different menus or functional blocks in the original business system are identified and segmented through URL routing rules and page content feature analysis, and then encapsulated as micro-applications. Dynamically deploy micro-applications based on requirements; After dynamic layout, the micro-applications acquire multi-source data on basic power grid operations according to requirements; Based on the acquired multi-source data, a work report tailored to specific needs is generated.

[0005] Optionally, a positioning algorithm based on a weighted fusion of DOM structural features and visual features enables cross-level page access and non-intrusive front-end data retrieval, including: Simulate real user sessions using headless browser technology and obtain legitimate access permissions through a dynamic token management mechanism; Based on the target page being accessed, the target page is parsed in parallel, and DOM tree structure features and visual rendering features are extracted simultaneously. Based on the extracted DOM tree structure features and visual rendering features, the matching degree between the target element and the pre-recorded element corresponding to the extracted DOM tree structure features and visual rendering features is calculated using an improved cosine similarity algorithm based on formula (1): Formula (1), in, Indicates the degree of matching. , These represent the weight coefficients corresponding to the DOM tree structure features and the visual rendering features, respectively. This represents the DOM tree structure feature vector of the current target element. This represents the feature vector of the baseline DOM tree structure of the pre-recorded elements. This represents the visual rendering feature vector of the current target element. Represents the baseline visual rendering feature vector of the pre-recorded element; If the matching degree is greater than a preset threshold, the captured element is determined to be the target element.

[0006] Optionally, micro-applications can be dynamically deployed based on requirements, including: Obtain the application workbench canvas and divide the application workbench canvas into a standard grid; Convert each micro-application to the size occupied by a grid cell; The usage frequency of micro-applications can be obtained using formula (2): Formula (2) in, Indicates micro-applications Frequency of use This indicates the total number of clicks. Indicates micro-applications Number of clicks in the past 30 days; Based on the obtained usage frequency, the first-fit descent greedy algorithm is used to place micro-applications in descending order of usage frequency.

[0007] Optionally, micro-applications can be dynamically deployed based on requirements, including: Based on the layout of the placed micro-applications, when staff manually drag and drop a micro-application to a new location, it is detected whether that location overlaps with other micro-applications. In the case of overlap, calculate all possible adjustment schemes and select the scheme with the minimum total moving distance; Move overlapping micro-applications down or to the right to eliminate overlap and keep the overall layout compact; Place the micro-application in a new location without overlap.

[0008] Optionally, when dynamically laying out micro-applications, the objective function for layout optimization is:

[0009]

[0010] in, This represents the x-coordinate of the top-left corner of the i-th micro-application in the workbench canvas. This represents the top-left y-coordinate of the i-th micro-application in the workbench canvas. This indicates the number of grid cells occupied by the width of the i-th micro-application. This represents the number of grid cells occupied by the height of the i-th micro-application. This indicates the total width of the workbench canvas. This represents the total height of the workbench canvas, and n represents the total number of micro-applications.

[0011] Optionally, the dynamically deployed micro-application can acquire multi-source data on grid-level operations as needed, including: Offline and real-time data are obtained from the data platform respectively, and the historical accuracy of the two data sources is recorded. The credibility weights of the two data sources are calculated using formula (3): Formula (3) in, This indicates the credibility weight of offline data. Indicates the credibility weight of real-time data. This indicates the historical accuracy of offline data. Represents real-time data Historical accuracy This indicates an offline data source obtained from the data platform. This indicates a real-time data source obtained from the AI ​​browser front-end; Obtain the difference between the offline data and the real-time data, and determine whether the difference between the offline data and the real-time data is greater than a preset threshold; If the difference between the offline data and the real-time data is not greater than a preset threshold, the final data is obtained using formula (4): Formula (4) in, This indicates the final data.

[0012] Optionally, the dynamically deployed micro-application can acquire multi-source data on grid-level operations as needed, including: If the difference between the offline data and the real-time data is greater than a preset threshold, the three most recent historical records corresponding to the two data sources are obtained, and the changing trends of the offline data and the real-time data, as well as the historical changing trends corresponding to the historical records, are calculated. Based on the three most recent historical records, obtain the corresponding standard deviation threshold; Based on the obtained standard deviation threshold, calculate whether the offline data and real-time data exceed the standard deviation threshold respectively; If one of the two data sources exceeds the standard deviation threshold, the other data source is determined as the final data source. When neither of the two data sources exceeds the standard deviation threshold, the similarity between the changing trends of offline and real-time data and the historical changing trends corresponding to historical records is calculated, and the data source with greater similarity is selected as the final data.

[0013] Optionally, the similarity between the changing trends of offline data and real-time data and the historical changing trends corresponding to historical records is calculated, including: calculating the similarity between the two data sources and the historical changing trends using formula (5): Formula (5) in, This indicates the similarity between the changing trend of offline or real-time data and the corresponding historical changing trend. This indicates the slope of the trend in offline or real-time data. It represents the slope of the average trend of historical data.

[0014] Optionally, based on the acquired multi-source data, a work report tailored to specific needs can be generated, including: Through a visual interface, business rules can be configured as an enhanced rule tree. The rule tree has the final business metric as the root node, the original data items as leaf nodes, and the intermediate nodes as arithmetic operations, statistical aggregations, or business logic judgments. Based on the selected business scenario and time range, the system calls the acquired multi-source data, traverses the rule tree using a depth-first search algorithm, completes the data calculation, and outputs structured data results and business tags. The system receives the structured data output and business tags as context, accurately fills the data into the placeholders of the report template through template-data matrix mapping, then dynamically activates the conditional blocks according to the business tags, and calls the semantic rule library to generate data narrative text. Based on the completed data narrative text, different renderers are called to generate Word / PDF archive documents, PPT presentations, or HTML interactive dashboards. At the same time, the node results in the depth-first search calculation process are cached and reused directly when multiple rule trees share the same subtree, avoiding duplicate calculations.

[0015] Through the above technical solution, this invention provides a full-process intelligent management and control method for power grid grassroots operations. It achieves cross-level page access and non-intrusive front-end data retrieval through a positioning algorithm based on a weighted fusion of DOM structural features and visual features. Based on the results of cross-level page access and non-intrusive front-end data retrieval, URL routing rules and page content feature analysis are used to identify and segment the boundaries of different menus or functional blocks in the original business system, allowing them to be encapsulated as micro-applications. After obtaining the micro-applications, they can be dynamically laid out according to requirements. The dynamically laid-out micro-applications can obtain multi-source data from power grid grassroots operations according to requirements, and based on this obtained multi-source data, work reports tailored to specific needs can be generated. This method can solve the technical problems of frequent multi-system switching, long business access paths, cumbersome operation steps, and low efficiency of manual operation in existing grassroots operations.

[0016] Other features and advantages of the embodiments of the present invention will be described in detail in the following detailed description section. Attached Figure Description

[0017] The accompanying drawings are provided to further illustrate embodiments of the present invention and form part of the specification. They are used together with the following detailed description to explain the embodiments of the present invention, but do not constitute a limitation thereof. In the drawings: Figure 1 This is a flowchart of a full-process intelligent management and control method for basic power grid operations according to an embodiment of the present invention. Detailed Implementation

[0018] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are for illustration and explanation only and are not intended to limit the scope of the present invention.

[0019] In the embodiments of this application, certain software, components, models and other existing solutions in the industry may be mentioned. These should be regarded as exemplary and are only intended to illustrate the feasibility of implementing the technical solution of this application. However, they do not mean that the applicant has used or necessarily used the solution.

[0020] Figure 1 This is a flowchart of a full-process intelligent management and control method for basic power grid operations according to an embodiment of the present invention. In this invention, the management and control process may include: In step S1, a positioning algorithm based on the weighted fusion of DOM structural features and visual features enables cross-level page access and non-intrusive front-end data retrieval.

[0021] In step S2, based on the results of cross-level page access and non-intrusive front-end data retrieval, the boundaries of different menus or functional blocks in the original business system are identified and cut off through URL routing rules and page content feature analysis, and then encapsulated into micro-applications.

[0022] In step S3, the micro-applications are dynamically laid out based on requirements.

[0023] In step S4, the dynamically deployed micro-application acquires multi-source data on power grid infrastructure operations as needed.

[0024] In step S5, a work report tailored to specific needs is generated based on the acquired multi-source data.

[0025] In this invention, when generating work reports, a positioning algorithm based on a weighted fusion of DOM structural features and visual features can be used to achieve cross-level page access and non-intrusive front-end data retrieval. Based on the results of cross-level page access and non-intrusive front-end data retrieval, URL routing rules and page content feature analysis are used to identify and segment the boundaries of different menus or functional blocks in the original business system, thus enabling encapsulation into micro-applications. After obtaining the micro-applications, they can be dynamically laid out based on requirements. The dynamically laid-out micro-applications can then acquire multi-source data from grassroots power grid operations according to requirements. Based on this acquired multi-source data, work reports tailored to specific needs can be generated. This method can solve the technical problems of frequent multi-system switching, long business access paths, cumbersome operation steps, and low efficiency of manual operation in existing grassroots operations.

[0026] In one embodiment of the present invention, the process of obtaining the target element may include: In step S6, a real user session is simulated using headless browser technology, and legitimate access permissions are obtained through a dynamic token management mechanism.

[0027] In step S7, the target page is parsed in parallel based on the accessed target page, and DOM tree structure features and visual rendering features are extracted at the same time.

[0028] In step S8, based on the extracted DOM tree structure features and visual rendering features, the matching degree between the target element and the pre-recorded element corresponding to the extracted DOM tree structure features and visual rendering features is calculated using the improved cosine similarity algorithm based on formula (1): Formula (1), in, Indicates the degree of matching. , These represent the weight coefficients corresponding to the DOM tree structure features and the visual rendering features, respectively. This represents the DOM tree structure feature vector of the current target element. This represents the feature vector of the baseline DOM tree structure of the pre-recorded elements. This represents the visual rendering feature vector of the current target element. This represents the baseline visual rendering feature vector of the pre-recorded element.

[0029] In step S9, if the matching degree is greater than a preset threshold, the captured element is determined to be the target element.

[0030] In this invention, when acquiring target elements, a headless browser technique can be used to simulate a real user session, and then a dynamic token management mechanism can be used to obtain legitimate access permissions. Based on the accessed target page, the target page can be parsed in parallel, and DOM tree structure features and visual rendering features can be extracted. Based on the extracted DOM tree structure features and visual rendering features, the matching degree between the target element and the pre-recorded element corresponding to the extracted DOM tree structure features and visual rendering features can be calculated using an improved cosine similarity algorithm based on formula (1). If the matching degree is greater than a preset threshold, the captured element can be determined to be the target element.

[0031] In one embodiment of the present invention, the process of dynamically laying out micro-applications may include: In step S10, the application workbench canvas is obtained and divided into a standard grid.

[0032] In step S11, each micro-application is converted into a grid cell size.

[0033] In step S12, the usage frequency of the micro-application is obtained using formula (2): Formula (2) in, Indicates micro-applications Frequency of use This indicates the total number of clicks. Indicates micro-applications Number of clicks in the past 30 days.

[0034] In step S13, based on the obtained usage frequency, the first-fit descent greedy algorithm is used to place the micro-applications in descending order of usage frequency.

[0035] In this invention, during dynamic layout, the application workbench canvas can be obtained and divided into a standard grid. Then, each micro-application can be converted into a grid cell size. The usage frequency of the micro-applications can be obtained using formula (2). Based on the obtained usage frequency, a first-fit descent greedy algorithm can be used to place the micro-applications in descending order of usage frequency.

[0036] In one embodiment of the present invention, the process of performing dynamic layout may further include: In step S14, based on the placed micro-application layout, when the staff manually drags the micro-application to the new location, it is detected whether the location overlaps with other micro-applications.

[0037] In step S15, in the case of overlap, all possible adjustment schemes are calculated, and the scheme with the smallest total moving distance is selected.

[0038] In step S16, overlapping micro-applications are shifted down or to the right to eliminate overlap and maintain a compact overall layout.

[0039] In step S17, the micro-application is placed in the new location without overlap.

[0040] In this invention, during dynamic layout, the system can detect whether a newly placed micro-application overlaps with other micro-applications when an operator manually drags and drops a micro-application to a new location. If overlap occurs, all possible adjustment schemes can be calculated, and the scheme with the minimum total movement distance can be selected. Overlapping micro-applications can be shifted downwards or to the right to eliminate overlap and maintain a compact overall layout. If no overlap occurs, the micro-application can be placed in the new location.

[0041] In one embodiment of the present invention, when dynamically laying out micro-applications, the objective function for layout optimization is:

[0042]

[0043] in, This represents the x-coordinate of the top-left corner of the i-th micro-application in the workbench canvas. This represents the top-left y-coordinate of the i-th micro-application in the workbench canvas. This indicates the number of grid cells occupied by the width of the i-th micro-application. This represents the number of grid cells occupied by the height of the i-th micro-application. This indicates the total width of the workbench canvas. This represents the total height of the workbench canvas, and n represents the total number of micro-applications.

[0044] In one embodiment of the present invention, the process of acquiring multi-source data may include: In step S18, offline data and real-time data are obtained from the data platform respectively, and the historical accuracy of the two data sources is recorded.

[0045] In step S19, the credibility weights of the two data sources are calculated using formula (3): Formula (3) in, This indicates the credibility weight of offline data. Indicates the credibility weight of real-time data. This indicates the historical accuracy of offline data. Represents real-time data Historical accuracy, This indicates an offline data source obtained from the data platform. This indicates a real-time data source obtained from the AI ​​browser front end.

[0046] In step S20, the difference between offline data and real-time data is obtained, and it is determined whether the difference between offline data and real-time data is greater than a preset threshold.

[0047] In step S21, if the difference between offline data and real-time data is not greater than a preset threshold, the final data is obtained using formula (4): Formula (4) in, This indicates the final data.

[0048] In this invention, when acquiring multi-source data, offline data and real-time data can be obtained from the data platform respectively, and the historical accuracy rates of the two data sources are recorded. Then, the credibility weights of the two data sources can be calculated using formula (3). The difference between the offline data and the real-time data can be obtained, and then it can be determined whether the difference between the offline data and the real-time data is greater than a preset threshold. If the difference between the offline data and the real-time data is not greater than the preset threshold, the final data can be obtained using formula (4).

[0049] In one embodiment of the present invention, the process of acquiring multi-source data may further include: In step S22, if the difference between offline data and real-time data is greater than a preset threshold, the three most recent historical records corresponding to the two data sources are obtained, and the changing trends of offline data and real-time data, as well as the historical changing trends corresponding to the historical records, are calculated.

[0050] In step S23, the corresponding standard deviation threshold is obtained based on the three most recent historical records.

[0051] In step S24, based on the obtained standard deviation threshold, it is calculated whether the offline data and real-time data exceed the standard deviation threshold respectively.

[0052] In step S25, if one of the two data sources exceeds the standard deviation threshold, the other data source is determined as the final data.

[0053] In step S26, when neither of the two data sources exceeds the standard deviation threshold, the similarity between the changing trends of offline data and real-time data and the historical changing trends corresponding to historical records is calculated, and the data source with greater similarity is selected as the final data.

[0054] In this invention, when acquiring multi-source data, if the difference between offline data and real-time data exceeds a preset threshold, the three most recent historical records corresponding to the two data sources can be obtained, and the changing trends of the offline and real-time data, as well as the historical changing trends corresponding to the historical records, can be calculated. Based on the three most recent historical records, the corresponding standard deviation threshold can be obtained. Based on the obtained standard deviation threshold, it can be calculated whether the offline and real-time data exceed the standard deviation threshold. If one of the two data sources exceeds the standard deviation threshold, the other data source can be determined as the final data. If neither data source exceeds the standard deviation threshold, the similarity between the changing trends of the offline and real-time data and the historical changing trends corresponding to the historical records can be calculated, and the data source with greater similarity can be selected as the final data.

[0055] In one embodiment of the present invention, the similarity between the changing trends of offline data and real-time data and the historical changing trends corresponding to historical records can be calculated using formula (5): Formula (5) in, This indicates the similarity between the changing trend of offline or real-time data and the corresponding historical changing trend. This indicates the slope of the trend in offline or real-time data. It represents the slope of the average trend of historical data.

[0056] In one embodiment of the present invention, the process of generating a work report may include: In step S27, the business rules are configured as an enhanced rule tree through a visual interface. The rule tree has the final business indicator as the root node, the original data items as leaf nodes, and the intermediate nodes as arithmetic operations, statistical aggregations, or business logic judgments.

[0057] In step S28, based on the selected business scenario and time range, the acquired multi-source data is invoked, and the rule tree is traversed through a depth-first search algorithm to complete the data operation, outputting structured data results and business tags.

[0058] In step S29, the structured data results and business tags are received as context. The data is accurately filled into the placeholders of the report template through template-data matrix mapping. Then, the condition blocks are dynamically activated according to the business tags, and the semantic rule library is called to generate data narrative text.

[0059] In step S30, based on the completed data narrative text, different renderers are called to generate Word / PDF archive documents, PPT presentations, or HTML interactive dashboards. At the same time, the node results in the depth-first search calculation process are cached and reused directly when multiple rule trees share the same subtree, avoiding redundant calculations.

[0060] In this invention, when generating a work report, business rules (such as "line loss rate = (power supply - electricity sales) / power supply × 100%)" can be configured as an enhanced rule tree through a visual interface. The rule tree uses the final business indicator as the root node, the original data items as leaf nodes, and intermediate nodes for arithmetic operations, statistical aggregation, or business logic judgments. Based on the selected business scenario and time range, the acquired multi-source data is called, and the rule tree is traversed using a depth-first search algorithm to complete the data calculation and output structured data results (such as {"line loss rate": 5.2%, "number of abnormal transformer areas": 3}) and business tags (such as {"line loss abnormal": true}). The output structured data results and business tags can be used as context, and then the data can be accurately filled into the placeholders of the report template through template-data matrix mapping. Then, conditional blocks can be dynamically activated according to business tags (such as expanding the "deep analysis of abnormal transformer areas" section only when "line loss abnormal" is true), and the semantic rule library can be called to generate data narrative text (such as "electricity sales this month show an increasing trend"). Based on the completed data narrative text, different renderers can be called to generate Word / PDF archive documents, PPT presentations, or HTML interactive dashboards. At the same time, the node results in the depth-first search calculation process can be cached and reused directly when multiple rule trees share the same subtree, avoiding duplicate calculations.

[0061] Construction and Representation of Enhanced Rule Trees: Business rules are formalized into structured, enhanced rule trees to carry logic ranging from simple arithmetic to complex business judgments.

[0062] 1. Tree structure modeling: The final business indicator is the root node, the original data is the leaf node, and the intermediate nodes can represent arithmetic operations (addition, subtraction, multiplication, division), statistical aggregation (summation, average, maximum value) and business logic judgment (condition judgment, data cleaning).

[0063] 2. Node type expansion: In addition to basic data nodes and computation nodes, condition judgment nodes (e.g., judging "line loss rate > 7%)" and data cleaning nodes (e.g., handling null values ​​or outliers) are introduced, so that the rule tree can directly encapsulate complete business rules.

[0064] 3. Dynamic parameter support: Key parameters such as thresholds and periods in the rule tree are designed as injectable variables, enabling the same tree to be reused for calculations in different time periods or business scenarios through parameter binding.

[0065] Intelligent computing and caching based on depth-first search: The rule tree is executed using a depth-first search (DFS) algorithm, and performance is significantly optimized through a caching mechanism.

[0066] 1. Recursive calculation: Depth-first traversal starting from the root node. For data nodes, the data access module is called to obtain the value D; for operation nodes, the values ​​of all child nodes are recursively calculated before the operator Op of this node is executed.

[0067] 2. Intelligent Caching and Versioned Recalculation: To balance computational efficiency and data updates, a version-aware caching mechanism is designed. The cache key for each subtree's computation result is generated jointly by its structure hash, the input data version hash, and dynamic parameter values. CacheKey=Hash(SubTree_Signature)⊕Hash(Data_Version)⊕Hash(Param_Values) When the source data is updated, the system can accurately identify and invalidate only cache entries that depend on the data by changing the data version hash, thereby compressing the scope of recalculation from the global to the smallest affected subtree, greatly improving the efficiency of repeated operations.

[0068] Execution process optimization: pruning, short-circuit evaluation, and parallelization Multiple optimization strategies are integrated into the execution of DFS (Depth-First Search) to improve computation speed and resource utilization.

[0069] 1. Invalid computation pruning: During traversal, if the value of a child node is detected to be empty (NULL) or obviously invalid, the subsequent calculation of that branch can be terminated immediately, and the predefined default value or empty result can be returned.

[0070] 2. Short-circuit evaluation: For AND and OR logical judgment nodes, short-circuit evaluation optimization is used. AND operation: If any child node evaluates to FALSE, the entire expression is immediately determined to be false, and the remaining child nodes are not evaluated. Formal representation: AND(v_1, v_2, ...) = FALSE. v_i = FALSE. The OR operation: If any child node evaluates to TRUE, the entire expression is immediately considered true, formally represented as: OR(v_1, v_2, ...) = TRUE. v_i = TRUE.

[0071] 3. Parallel scheduling of subtasks: The system automatically analyzes the rule tree, identifies independent subtrees that have no data dependency on each other, and schedules them to different computing units for parallel execution, shortening the overall job time.

[0072] Context-aware intelligent report orchestration and generation The output structured data set, Data_Set, together with business tags (such as exception flags), can constitute the "context" for report generation. The intelligent template orchestration engine can leverage this context to drive the dynamic assembly of reports.

[0073] 1. Template-Data Matrix Mapping: The report template predefines a mapping matrix M between data placeholders and data items. The generation process can be abstracted as: Report_Matrix = M · Data_Vector, where Data_Vector is a vector composed of data sets, achieving precise data filling.

[0074] 2. Dynamic Activation of Conditional Blocks: Chapters, tables, or paragraphs in the report template can be defined as "conditional blocks," and their activation is determined by the business tags in the context. For example, the "In-depth Analysis of Abnormal Transformer Areas" chapter will only be instantiated and populated with data when Data_Set["Number of Transformer Areas with Abnormal Line Loss"]>0, achieving intelligent focus of the report content.

[0075] 3. Automatic Data Narrative Generation: The engine has a built-in semantic rule library that can automatically synthesize concise interpretable text based on the results of key indicators and context. For example, when the electricity sales growth rate is positive, the description "Electricity sales this month showed an upward trend" is automatically added, improving the readability and insight of the report.

[0076] 4. Multimodal adaptive output: Based on the same populated report data kernel, the system can call different renderers to generate Word / PDF documents suitable for archiving and reporting, PPT presentations for meeting presentations, or interactive HTML dashboards for real-time monitoring with one click, achieving separation of content and presentation format.

[0077] Through the above technical solution, this invention provides a full-process intelligent management and control method for power grid grassroots operations. It achieves cross-level page access and non-intrusive front-end data retrieval through a positioning algorithm based on a weighted fusion of DOM structural features and visual features. Based on the results of cross-level page access and non-intrusive front-end data retrieval, URL routing rules and page content feature analysis are used to identify and segment the boundaries of different menus or functional blocks in the original business system, allowing them to be encapsulated as micro-applications. After obtaining the micro-applications, they can be dynamically laid out according to requirements. The dynamically laid-out micro-applications can obtain multi-source data from power grid grassroots operations according to requirements, and based on this obtained multi-source data, work reports tailored to specific needs can be generated. This method can solve the technical problems of frequent multi-system switching, long business access paths, cumbersome operation steps, and low efficiency of manual operation in existing grassroots operations.

[0078] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied 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.

[0079] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0080] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0081] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0082] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0083] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0084] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, 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 technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0085] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0086] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A method for intelligent control of the entire process of power grid grassroots operations, characterized in that, The method includes: A positioning algorithm based on weighted fusion of DOM structure features and visual features enables cross-level page access and non-intrusive front-end data retrieval; Based on the results of cross-level page access and non-intrusive front-end data retrieval, the boundaries of different menus or functional blocks in the original business system are identified and segmented through URL routing rules and page content feature analysis, and then encapsulated as micro-applications. Dynamically deploy micro-applications based on requirements; After dynamic layout, the micro-applications acquire multi-source data on basic power grid operations according to requirements; Based on the acquired multi-source data, a work report tailored to specific needs is generated.

2. The method according to claim 1, characterized in that, A positioning algorithm based on a weighted fusion of DOM structure features and visual features enables cross-level page access and non-intrusive front-end data retrieval, including: Simulate real user sessions using headless browser technology and obtain legitimate access permissions through a dynamic token management mechanism; Based on the target page being accessed, the target page is parsed in parallel, and DOM tree structure features and visual rendering features are extracted simultaneously. Based on the extracted DOM tree structure features and visual rendering features, the matching degree between the target element and the pre-recorded element corresponding to the extracted DOM tree structure features and visual rendering features is calculated using an improved cosine similarity algorithm based on formula (1): Official (1), in, Indicates the degree of matching. , These represent the weight coefficients corresponding to the DOM tree structure features and the visual rendering features, respectively. This represents the DOM tree structure feature vector of the current target element. This represents the feature vector of the baseline DOM tree structure of the pre-recorded elements. This represents the visual rendering feature vector of the current target element. Represents the baseline visual rendering feature vector of the pre-recorded element; If the matching degree is greater than a preset threshold, the captured element is determined to be the target element.

3. The method according to claim 1, characterized in that, Dynamically deploying micro-applications based on requirements, including: Obtain the application workbench canvas and divide the application workbench canvas into a standard grid; Convert each micro-application to the size occupied by a grid cell; The usage frequency of micro-applications can be obtained using formula (2): Official (2) in, Indicates micro-applications Frequency of use This indicates the total number of clicks. Indicates micro-applications Number of clicks in the past 30 days; Based on the obtained usage frequency, the first-fit descent greedy algorithm is used to place micro-applications in descending order of usage frequency.

4. The method according to claim 3, characterized in that, Dynamically deploying micro-applications based on requirements, including: Based on the layout of the placed micro-applications, when staff manually drag and drop a micro-application to a new location, it is detected whether that location overlaps with other micro-applications. In the case of overlap, calculate all possible adjustment schemes and select the scheme with the minimum total moving distance; Move overlapping micro-applications down or to the right to eliminate overlap and keep the overall layout compact; Place the micro-application in a new location without overlap.

5. The method according to claim 4, characterized in that, When dynamically laying out micro-applications, the objective function for layout optimization is: in, This represents the x-coordinate of the top-left corner of the i-th micro-application in the workbench canvas. This represents the top-left y-coordinate of the i-th micro-application in the workbench canvas. This indicates the number of grid cells occupied by the width of the i-th micro-application. This represents the number of grid cells occupied by the height of the i-th micro-application. This indicates the total width of the workbench canvas. This represents the total height of the workbench canvas, and n represents the total number of micro-applications.

6. The method according to claim 1, characterized in that, After dynamic deployment, the micro-application acquires multi-source data on basic power grid operations as needed, including: Offline and real-time data are obtained from the data platform respectively, and the historical accuracy of the two data sources is recorded. The credibility weights of the two data sources are calculated using formula (3): Official (3) in, This indicates the credibility weight of offline data. Indicates the credibility weight of real-time data. This indicates the historical accuracy of offline data. Represents real-time data Historical accuracy, This indicates an offline data source obtained from the data platform. This indicates a real-time data source obtained from the AI ​​browser front-end; Obtain the difference between the offline data and the real-time data, and determine whether the difference between the offline data and the real-time data is greater than a preset threshold; If the difference between the offline data and the real-time data is not greater than a preset threshold, the final data is obtained using formula (4): Official (4) in, This indicates the final data.

7. The method according to claim 6, characterized in that, After dynamic deployment, the micro-application acquires multi-source data on basic power grid operations as needed, including: If the difference between the offline data and the real-time data is greater than a preset threshold, the three most recent historical records corresponding to the two data sources are obtained, and the changing trends of the offline data and the real-time data, as well as the historical changing trends corresponding to the historical records, are calculated. Based on the three most recent historical records, obtain the corresponding standard deviation threshold; Based on the obtained standard deviation threshold, calculate whether the offline data and real-time data exceed the standard deviation threshold respectively; If one of the two data sources exceeds the standard deviation threshold, the other data source is determined as the final data source. When neither of the two data sources exceeds the standard deviation threshold, the similarity between the changing trends of offline and real-time data and the historical changing trends corresponding to historical records is calculated, and the data source with greater similarity is selected as the final data.

8. The method according to claim 7, characterized in that, Calculating the similarity between the changing trends of offline and real-time data and the historical changing trends corresponding to historical records includes: calculating the similarity between the two data sources and the historical changing trends using formula (5): Official (5) in, This indicates the similarity between the changing trend of offline or real-time data and the corresponding historical changing trend. This indicates the slope of the trend in offline or real-time data. It represents the slope of the average trend of historical data.

9. The method according to claim 1, characterized in that, Based on the acquired multi-source data, a work report tailored to specific needs is generated, including: Through a visual interface, business rules can be configured as an enhanced rule tree. The rule tree has the final business metric as the root node, the original data items as leaf nodes, and the intermediate nodes as arithmetic operations, statistical aggregations, or business logic judgments. Based on the selected business scenario and time range, the system calls the acquired multi-source data, traverses the rule tree using a depth-first search algorithm, completes the data calculation, and outputs structured data results and business tags. The system receives the structured data output and business tags as context, accurately fills the data into the placeholders of the report template through template-data matrix mapping, then dynamically activates the conditional blocks according to the business tags, and calls the semantic rule library to generate data narrative text. Based on the completed data narrative text, different renderers are called to generate Word / PDF archive documents, PPT presentations, or HTML interactive dashboards. At the same time, the node results in the depth-first search calculation process are cached and reused directly when multiple rule trees share the same subtree, avoiding duplicate calculations.

Images