Processing method and system for presenting double intelligent agents based on service disassembly and visualization

Abstract

The invention discloses a processing method and system for presenting double intelligent agents based on business disassembly and visualization. The processing method comprises the following steps: receiving a natural language request of a power system by a service disassembling agent, carrying out semantic analysis on the natural language request through a small language model, and executing semantic retrieval based on a power system knowledge graph to realize intent recognition and task disassembling of knowledge enhancement. Generating a corresponding task instruction chain by using a directed acyclic graph structure; the task instruction chain is sent to an instruction registration center, the instruction registration center dispatches a visual presentation agent, and an interactive instrument panel and a natural language abstract are generated through a large language model. According to the technical scheme, automatic understanding and efficient decomposition of power system services are achieved, the intelligence and automation level of data processing and visualization is improved, the task scheduling and management capacity of the system is enhanced, manual intervention is effectively reduced, and the operation and maintenance efficiency and the data display accuracy are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent power systems, and in particular to a processing method and system based on business disassembly and visual presentation of dual intelligent bodies. Background Art

[0002] In the current operation of the power system, with the continuous growth in the number of equipment and the volume of data, power companies are facing multiple challenges such as data heterogeneity, complex business operations, and inefficient manual processing. Traditional data analysis and business processing methods rely on a lot of manual intervention and are difficult to respond efficiently to diverse business needs. Especially in scenarios such as fault diagnosis, equipment status monitoring, and operation analysis, there are problems such as information dispersion, data silos, and fragmented business response processes, resulting in delayed operation and maintenance decisions and increased management costs. In recent years, although some systems have introduced rule-based or simple automated data processing platforms, when faced with complex and dynamic business needs and the diverse questions raised by users through natural language, existing technologies still have many limitations in automatic understanding, intelligent task decomposition, and efficient presentation of analysis results. It is difficult to meet the actual needs of intelligent, automated, and refined management of power systems.

[0003] With the rapid development of artificial intelligence and natural language processing technologies, technologies that use intelligent agents to automatically decompose complex business operations and visualize data are gradually emerging. New-generation intelligent methods such as multi-agent collaboration mechanisms, knowledge-enhanced retrieval driven by domain knowledge graphs, task modeling with directed acyclic graph structures, and process management using finite state machines provide a feasible path for semantic understanding, intelligent decomposition, and automatic execution of large-scale data in power systems. However, the industry currently lacks a solution that can deeply integrate business decomposition agents with visualization agents, combine knowledge graphs, task instruction chains, and lifecycle management, and achieve end-to-end intelligent processing solutions from user natural language requests to multi-dimensional data visualization. Therefore, there is an urgent need for an intelligent task decomposition and visualization method for power system business to improve the level of intelligent power system data processing. Summary of the Invention

[0004] The present invention provides a processing method and system based on dual intelligent bodies of business disassembly and visualization presentation, which are used to realize automatic understanding and efficient decomposition of power system business, improve the intelligence and automation level of data processing and visualization, enhance the system's task scheduling and management capabilities, effectively reduce manual intervention, and improve the accuracy of data display.

[0005] According to a first aspect of the present invention, a processing method based on business disassembly and visual presentation of dual-agents is provided, the processing method based on business disassembly and visual presentation of dual-agents comprising: The business disassembly agent receives natural language requests from the power system, performs semantic parsing on the natural language requests using a small language model, performs semantic retrieval based on the power system knowledge graph, implements knowledge-enhanced intent recognition and task disassembly, and generates the corresponding task instruction chain using a directed acyclic graph structure. The task instruction chain is sent to an instruction registration center, which dispatches a visualization presentation agent to generate an interactive dashboard and a natural language summary through a large language model to complete the visualization of power system data; The instruction registration center uses a finite state machine to perform lifecycle management on the entire processing flow to constrain the execution flow of the task instruction chain.

[0006] In one embodiment, it further includes: Performing semantic parsing on the natural language request using a small language model; Performing a search-enhanced semantic retrieval on the semantic parsing result based on the power system knowledge graph to obtain power system domain knowledge related to the request; The acquired power system domain knowledge is applied to understanding the natural language request, and a corresponding task instruction chain is generated.

[0007] In one embodiment, it further includes: Pre-training on general corpus to obtain the initial language model; Training the initial language model based on power system domain data to obtain a small language model optimized for the power system domain; The trained small language model is deployed into the business decomposition agent for performing intent recognition and task decomposition on natural language requests.

[0008] In one embodiment, it further includes: Each node in the task instruction chain corresponds to a power system data analysis subtask or visualization subtask; In the directed acyclic graph structure, the directed connections between the nodes represent the dependency relationships between the subtasks; The directed acyclic graph structure ensures that there is no circular dependency during the execution of the task instruction chain.

[0009] In one embodiment, it further includes: After receiving the task instruction chain generated by the business disassembly agent, each subtask instruction is registered as a waiting execution state; When all the pre-order dependency conditions of a subtask are met, the subtask is switched to the execution state and the visual presentation agent executes the subtask; After the subtask is executed, it is converted to the completed state. If an error occurs during the execution, it is converted to the error state to trigger the corresponding exception handling.

[0010] In one embodiment, it further includes: Parsing visualization-related instructions in the task instruction chain into interface components of an interactive dashboard to display power system data analysis results; generating a natural language summary report describing the analysis results based on the large language model; The interactive dashboard can support users to interactively query and dynamically visualize power system data.

[0011] According to a second aspect of the present invention, a dual-agent processing system based on business decomposition and visual presentation is provided, comprising: A disassembly module is configured to receive natural language requests from the power system through a business disassembly agent, perform semantic parsing on the natural language requests using a small language model, perform semantic retrieval based on the power system knowledge graph, implement knowledge-enhanced intent recognition and task disassembly, and generate a corresponding task instruction chain using a directed acyclic graph structure; A visualization module is used to send the task instruction chain to an instruction registration center, which schedules a visualization presentation agent to generate an interactive dashboard and natural language summary through a large language model to complete the visualization of power system data; The management module is used to perform life cycle management on the entire processing flow through the instruction registration center using a finite state machine to constrain the execution flow of the task instruction chain.

[0012] In one embodiment, the disassembly module, the visualization module and the management module are controlled to execute any one of the above-mentioned dual-agent processing methods based on business disassembly and visualization presentation.

[0013] According to a third aspect of the present invention, there is provided an electronic device, the electronic device comprising: a communication interface, a processor, and a memory; Wherein, the memory is used to store program instructions, and when the program instructions are executed by the processor that is communicatively connected to the memory through the communication interface, any of the above-mentioned processing methods based on business disassembly and visual presentation of dual intelligent agents is implemented.

[0014] According to a fourth aspect of the present invention, a computer-readable storage medium is provided, on which computer program instructions are stored. When the computer program instructions are executed by a computer (for example, a processor in a computer), any of the above-mentioned processing methods based on business decomposition and visual presentation of dual intelligent agents is implemented.

[0015] In summary, the present invention provides a processing method and system based on a dual intelligent agent of business decomposition and visualization presentation, the method comprising: receiving a natural language request of the power system by a business decomposition intelligent agent, semantically parsing the natural language request through a small language model, performing semantic retrieval based on the power system knowledge graph, realizing knowledge-enhanced intention recognition and task decomposition, and generating a corresponding task instruction chain with a directed acyclic graph structure; sending the task instruction chain to an instruction registration center, which schedules a visualization presentation intelligent agent, generates an interactive dashboard and a natural language summary through a large language model to complete the visualization of power system data; and the instruction registration center uses a finite state machine to perform lifecycle management on the entire processing flow to constrain the execution process of the task instruction chain. The technical solution of the present application realizes intelligent understanding, knowledge enhancement and automated task decomposition of natural language requests of the power system through the collaboration of a business decomposition intelligent agent and a visualization presentation intelligent agent, effectively improving the response speed of business needs and the level of intelligent processing. The directed acyclic graph structure is used for task modeling, combined with the unified scheduling of the instruction registration center and the lifecycle management of the finite state machine, to ensure the efficient execution and process controllability of each subtask. The overall solution can significantly improve the automation, accuracy and operational experience of power system data analysis and visualization, and significantly reduce manual intervention and management costs.

[0016] Other features and advantages of the present invention will be described in the following description, and in part will become apparent from the description, or will be understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description and the accompanying drawings.

[0017] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the specific embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0019] Figure 1 A flowchart of a dual-agent processing method based on business decomposition and visual presentation provided by an embodiment of the present invention; Figure 2 A flowchart of another dual-agent processing method based on business decomposition and visual presentation provided by an embodiment of the present invention; Figure 3A flowchart of another dual-agent processing method based on business decomposition and visual presentation provided by an embodiment of the present invention; Figure 4 A flowchart of another dual-agent processing method based on business decomposition and visual presentation provided by an embodiment of the present invention; Figure 5 A flowchart of another dual-agent processing method based on business decomposition and visual presentation provided by an embodiment of the present invention; Figure 6 A flowchart of another dual-agent processing method based on business decomposition and visual presentation provided by an embodiment of the present invention; Figure 7 A structural diagram of a dual-agent processing system based on business decomposition and visual presentation provided by an embodiment of the present invention; Figure 8 A structural diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0020] The features and exemplary embodiments of various aspects of the present application will be described in detail below. In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain the present application, rather than to limit the present application. For those skilled in the art, the present application can be implemented without the need for some of these specific details. The following description of the embodiments is merely to provide a better understanding of the present application by illustrating the examples of the present application.

[0021] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. In the absence of further limitations, the elements defined by the phrase "comprising..." do not exclude the presence of other identical elements in the process, method, article, or device comprising the elements.

[0022] like Figure 1 As shown, the present invention provides a processing method based on business disassembly and visual presentation of dual-agents, the processing method based on business disassembly and visual presentation of dual-agents includes: In step S11, the business disassembly agent receives a natural language request from the power system, performs semantic parsing on the natural language request using a small language model, performs semantic retrieval based on the power system knowledge graph, implements knowledge-enhanced intent recognition and task disassembly, and generates a corresponding task instruction chain using a directed acyclic graph structure. In step S12, the task instruction chain is sent to an instruction registration center, which dispatches a visualization presentation agent to generate an interactive dashboard and a natural language summary through a large language model to complete the visualization of the power system data. In step S13, the instruction registration center uses a finite state machine to perform lifecycle management on the entire processing flow to constrain the execution flow of the task instruction chain.

[0023] In one embodiment, multi-agent collaboration and knowledge-enhanced analysis enable intelligent understanding of natural language requests and interactive presentation of results. Based on a dual-agent architecture consisting of a business decomposition agent and a visualization agent, the processing flow is coordinated by a command registration center. The business decomposition agent receives natural language requests from the power system domain, semantically parses the requests using a small language model, and performs semantic retrieval based on the power system knowledge graph. This enables knowledge-enhanced user intent recognition and task decomposition, generating a corresponding task instruction chain using a directed acyclic graph (DAG) structure. The task instruction chain is then sent to the command registration center, which schedules the visualization agent to execute subsequent steps. The visualization agent processes the received task instruction chain using a large language model, generating interactive dashboards (such as dynamic charts and indicator dashboards) related to the power system data, along with corresponding natural language summaries. This ultimately enables visual presentation and intelligent interpretation of the power system data requested by the user. During the entire processing process, the finite state machine (FSM) built into the instruction registration center performs lifecycle management on the execution of the task instruction chain, ensuring that each stage of the task chain is controlled according to the predetermined state transition sequence, constraining and supervising the execution process of the task chain, and ensuring the stability and reliability of the system processing process.

[0024] Through the aforementioned architecture, the business decomposition agent, acting as a front-end analysis module, is responsible for translating user natural language requests into executable task chains. Leveraging a small language model for semantic understanding, it parses the key intent and required business operations from user requests. Furthermore, leveraging the semantic search capabilities of the power system knowledge graph, it queries and compares the professional terms, equipment parameters, or historical data involved in the request, refining and supplementing task information through knowledge augmentation. Based on the semantic analysis results and information returned by the knowledge graph, the business decomposition agent automatically decomposes complex requests into a series of causally dependent subtask nodes and organizes these task nodes into a directed acyclic graph (DAG) structure to form an instruction chain. Through DAG modeling, the nodes and their dependencies in the task chain are explicitly represented, avoiding circular dependencies and ensuring the logical consistency and completeness of the task execution process. The DAG structure also lays the foundation for parallel processing and dynamic adjustment of tasks. For example, independent subtasks can be executed in parallel, improving overall execution efficiency. The task instruction chain generated by the business decomposition agent includes, but is not limited to, various subtasks, such as data acquisition, analytical calculations, and result generation. Each task node is provided with contextual parameters and execution goals.

[0025] The agent plays a central role in task chain dispatch and process control. After the DAG task instruction chain generated by the business decomposition agent is submitted to the agent, the agent automatically calls the corresponding visualization agent instance according to the task chain description to enter the execution phase. The agent strictly manages the lifecycle of the entire execution process through its internal finite state machine (FSM). For example, when the task chain is in the pending state, the agent triggers the visualization agent to begin execution. During the execution of subtasks, the FSM tracks the state changes of task nodes to ensure that the next phase is entered only after all dependent tasks are completed. If a task fails or the output does not meet expectations, the FSM can transfer the process to an exception handling or retry state according to predefined strategies to ensure the orderly progress of the processing flow. With the help of the FSM's state transition control, the instruction executor effectively constrains the execution path and rhythm of the task instruction chain, preventing abnormal situations such as out-of-order execution of tasks or infinite waiting, and achieving refined control of complex processes. The command executor also maintains a command registry to record the descriptions of various task commands and their corresponding execution modules. This facilitates the dynamic search and invocation of appropriate agents or tools to complete specific tasks, thereby achieving decoupled collaboration among the system's internal components. The business decomposition agent and the visualization agent are not directly coupled and interact with each other. Instead, they are scheduled by the command executor to connect the task chain from planning to execution, ensuring clear relationships and a clear division of labor among internal components.

[0026] The visualization agent, serving as a backend execution module, focuses on presenting and summarizing power data based on the task chain. When the command registration center dispatches this agent, it passes along the currently executable tasks and the required data context, authorized by the FSM. After receiving these task instructions, the large language model built into the visualization agent selects the appropriate presentation template and generation strategy based on the instruction type. For data tasks requiring graphical presentation, the agent automatically draws corresponding interactive dashboard components, such as trend line charts, pie charts, and geographic distribution maps. Furthermore, leveraging the powerful natural language generation capabilities of the large language model, the visualization agent also generates textual descriptions of the visualization results and business insight summaries. For example, in a power load analysis scenario, the visualization agent can automatically generate a dynamic line chart of the load curve based on the task chain instructions, accompanied by natural language descriptions of peak load periods and the causes of abnormal fluctuations. This allows users to intuitively understand the data through visualization while also gaining professional interpretation through textual descriptions. Throughout the visualization generation process, the agent can leverage a predefined template library and domain knowledge to adjust the dashboard layout and style, implementing an adaptive dashboard template generation strategy: automatically selecting the optimal chart type and layout method based on different data types and user needs, improving the clarity and pertinence of the presentation. During this process, the large-scale language model not only ensures the fluency and professionalism of the natural language summary, but also verifies the generated content in conjunction with information returned by the knowledge graph, avoiding factual errors and ensuring the accuracy and credibility of the output results. The visualization presentation agent feeds the generated dashboard and summary back to the instruction registration center, which aggregates the results and returns them to the user, enabling automatic response to the entire process from natural language queries to visual analysis conclusions.

[0027] Traditional rule-based configuration methods often require manual pre-definition of rules or templates for power business queries and presentations, lacking flexibility for new query requirements and making them difficult to scale in a timely manner. Furthermore, rule-based systems can typically only process a limited number of known question patterns and are unable to understand complex and varied natural language expressions. This embodiment, however, uses semantic parsing and knowledge graph enhancement driven by a small language model to enable the system to flexibly understand various natural language requests and automatically plan task chains, eliminating reliance on manual rules and achieving deep semantic understanding of user intent and dynamic task generation. Furthermore, single-model approaches (for example, relying solely on a large language model to directly generate answers) suffer from knowledge gaps and opaque reasoning chains when addressing specialized power issues. While large, general-purpose models possess powerful language generation capabilities, they lack specialized knowledge and real-time data access in the power sector, making them prone to inaccurate or unreliable responses. Furthermore, requiring a single model to independently complete all steps, from intent recognition to data extraction to result generation, lacks modular process control and makes timely error correction and optimization difficult. In contrast, this embodiment utilizes a dual-agent collaborative mechanism, with two dedicated agents handling task planning and result presentation, respectively, forming a closed-loop feedback loop coordinated by a command registration center. This modular decoupled architecture not only fully leverages the advantages of models of different scales—small models quickly and accurately understand domain semantics, while large models generate rich expressions and visualizations—but also enables monitoring and scheduling of each process step through a finite state machine, ensuring controllable and robust system behavior. Therefore, compared to the rigid rule-based systems or black-box single-model systems of existing technologies, this invention can complete complex power data query tasks with higher intelligence density and reliability. It possesses the adaptive semantic understanding and knowledge reasoning capabilities lacking in rule-based solutions, while avoiding the content distortion and process loss that can occur with single-model solutions. This significantly improves the automation level and user experience of power data services. Furthermore, a new dual-agent collaborative mechanism is proposed, in which the business decomposition agent and the visualization agent each perform their respective functions while cooperating with each other. Information sharing and process integration are achieved through a command registration center and FSM. For example, when generating dashboards and reports, the visualization agent can provide feedback on preliminary data analysis to the business decomposition agent, which serves as the basis for dynamic adjustment of the task chain. This implements a feedback synchronization mechanism between agents, thereby continuously optimizing task execution strategies. A directed acyclic graph (DAG) structure is introduced to model task chains, enabling nonlinear organization and dependency management of subtasks. Compared to linear sequential execution, DAG task chains support concurrent execution and flexible scheduling of subtasks. When power data queries involve multiple independent computations and analyses, the system can simultaneously execute different subtasks, improving efficiency. Even when encountering new information or environmental changes during task execution, the DAG task chain can be dynamically adjusted based on FSM strategies (for example, modifying task priorities or adding new subtask nodes), maintaining optimal responsiveness to user needs.A finite state machine (FSM) controls the entire process within the command registration center, formalizing the complex and ever-changing task execution process into a number of controlled states, effectively improving the security and predictability of system operation. FSM control not only allows for the definition of advanced behaviors such as looping and retrying after timeout, enabling self-healing and fault tolerance for abnormal situations, but also enables the system to repeatedly execute sub-processes according to predetermined strategies, meeting the requirements of scenarios requiring periodic operation, such as power monitoring. Through knowledge graph-enhanced analysis, this invention incorporates specialized knowledge from the power sector into natural language processing and task planning, significantly improving the system's accuracy in understanding and handling specialized problems. The knowledge graph provides domain entity relationships and rule constraints, enabling the business decomposition agent to reference background knowledge, terminology explanations, and historical correlation data in the graph when interpreting user intent, thereby generating more reasonable and complete task chains. During the result generation phase, the knowledge graph also assists large-scale models in validating key data and conclusions, ensuring the professional accuracy of dashboards and summaries. The dynamic adjustment mechanism of the task chain enables the system to adapt to new situations in a timely manner and perform secondary planning. The feedback synchronization between intelligent agents ensures the consistency of analysis results and presentation content and closed-loop optimization. The adaptive dashboard template generation strategy automatically selects the visualization method based on user preferences and data characteristics, improving the effectiveness and aesthetics of information presentation.

[0028] The technical solution in this embodiment achieves intelligent understanding, knowledge enhancement, and automated task decomposition of natural language requests for the power system through the collaboration of a business decomposition agent and a visualization presentation agent, effectively improving the response speed and intelligent processing level of business needs. The use of a directed acyclic graph structure for task modeling, combined with the unified scheduling of the instruction registration center and the lifecycle management of the finite state machine, ensures the efficient execution and process controllability of each subtask. The overall solution can greatly improve the automation, accuracy, and operational experience of power system data analysis and visualization, and significantly reduce manual intervention and management costs.

[0029] In one embodiment, Figure 2 As shown, the following steps S21-S23 are also included: In step S21, a small language model is used to perform semantic analysis on the natural language request; In step S22, a search-enhanced semantic search is performed on the semantic parsing result based on the power system knowledge graph to obtain power system domain knowledge related to the request; In step S23, the acquired power system domain knowledge is applied to the understanding of the natural language request, and a corresponding task instruction chain is generated.

[0030] In one embodiment, the intelligent understanding and automatic business decomposition of natural language requests in the power system domain are divided into three steps: semantic analysis, knowledge enhancement, and task generation. A small language model that has been optimized for the domain is used to perform semantic analysis on the user's natural language request. This small language model has been pre-trained and fine-tuned on a large amount of power system business corpus, and can efficiently identify common terms, equipment names, business scenarios, and complex expressions in the power industry, thereby accurately converting the user's natural language input into structured semantic expressions. For example, when faced with a request such as "analyze and visualize the load changes of transformers in substation B in summer", the model can not only accurately identify key entities and intents such as "transformer", "load changes", and "summer", but also distinguish multiple business requirements such as analysis and visualization contained in the request.

[0031] On the basis of semantic parsing, the power system knowledge graph is further used to perform semantic retrieval enhancement on the parsing results. The power system knowledge graph brings together structured domain knowledge such as equipment information, operating rules, historical cases, data sensors and their associations. By combining the semantic parsing results with the knowledge graph, domain knowledge related to the current request can be automatically retrieved, such as the historical operating data of the target equipment, typical failure modes, standard analysis methods, etc. Not only does it complete the implicit information that may exist in the user's request, but it can also enhance knowledge for unclear or incomplete requirements, so that the intelligent agent has a deeper understanding of the business environment and context. It can provide personalized and professional task decision-making basis for specific business scenarios, significantly improving the accuracy and practicality of task decomposition.

[0032] The professional domain knowledge that has undergone knowledge enhancement is applied to the understanding of natural language requests and task decomposition, automatically generating a task instruction chain containing multiple business subtasks. The task instruction chain connects various tasks such as data acquisition, analysis, and visualization in a structured manner, providing a foundation for subsequent automated scheduling and execution. This technical solution not only realizes the intelligent transformation of natural language requests into executable task chains, but also breaks through the limitations of traditional rule-based or template-based methods in complex industry knowledge processing through knowledge graph completion and enhancement, significantly improving the automation, accuracy, and responsiveness of intelligent operation and maintenance and data analysis of power systems.

[0033] In one embodiment, Figure 3 As shown, the following steps S31-S33 are also included: In step S31, pre-training is performed on the general corpus to obtain an initial language model; In step S32, the initial language model is trained based on the power system domain data to obtain a small language model optimized for the power system domain; In step S33, the small language model obtained through training is deployed into the business decomposition agent for performing intent recognition and task decomposition on natural language requests.

[0034] In one embodiment, the complete construction and application path of a small language model used for natural language request processing in the business context of the power system can be divided into three major steps: model pre-training, domain adaptation, and agent deployment. By pre-training on large-scale general corpus, an initial language model with broad language understanding capabilities can be obtained. The initial language model has good performance in basic grammar, general vocabulary, and general expression understanding, laying a solid foundation for subsequent domain-specific tasks. General pre-training provides an efficient parameter initialization and knowledge transfer foundation for the subsequent transfer learning and targeted optimization of the model, so that the model not only has the ability to process daily language requests, but also can adapt to complex and changing context structures.

[0035] Based on the general pre-trained model, a large amount of proprietary data in the power system field is further introduced for refined training. This stage not only includes core terms and expressions in the power industry, such as equipment names, operating status, business rules, and fault cases, but also covers scenario-based data such as real operation and maintenance logs and user business query history. Through this field fine-tuning, the model can fully learn the knowledge structure and common expression patterns of the power industry, significantly improving its understanding and adaptability to professional content such as power system business requirements, equipment associations, and analysis processes. After training, the small language model obtained can accurately identify the core intent and entities in the power system scenario while maintaining low computing resource consumption and high inference efficiency, providing reliable support for intelligent task decomposition.

[0036] The trained small language model, specifically for the power sector, is deployed in the business decomposition agent, becoming a foundational component for the system's natural language intent recognition and automated task decomposition. When users submit complex business requests such as analysis, query, or visualization via natural language, the agent quickly and accurately performs semantic parsing, mapping the request into structured business intent and task decomposition results. This not only significantly improves the speed and accuracy of user response in intelligent power system operation and maintenance, but also lays a solid technical foundation for subsequent knowledge graph retrieval, task instruction chain generation, and full-process automation.

[0037] In one embodiment, Figure 4 As shown, the following steps S41-S43 are also included: In step S41, each node in the task instruction chain corresponds to a power system data analysis subtask or a visualization subtask; In step S42, in the directed acyclic graph structure, the directed connections between the nodes represent the dependency relationships between the subtasks; In step S43, the directed acyclic graph structure is used to ensure that there is no circular dependency during the execution of the task instruction chain.

[0038] In one embodiment, the task instruction chain is organized in a directed acyclic graph structure to achieve detailed modeling and efficient management of complex business processes in the power system. Each node in the task instruction chain corresponds to an independent data analysis subtask or visualization subtask, which can decompose the user's overall business request into several independently executable and clearly describable functional units. For example, in the scenario of power system operation status analysis, the node may include data collection, anomaly detection, feature extraction, result aggregation, chart generation and other links, and each node carries different data processing or business logic tasks. Granular design helps the system to flexibly configure and reuse various subtask modules according to actual business needs, thereby improving the flexibility of task planning and the overall scalability of the system.

[0039] In this directed acyclic graph (DAG) structure, directed connections between nodes clearly indicate the forward-and-backward dependencies between subtasks. This means that the execution of a node requires the completion of its predecessor node. This not only ensures the correctness of data flow and business processes but also provides fundamental support for automated task scheduling. Based on the dependency structure within the DAG, the command registration center automatically identifies subtasks that can be executed in parallel or sequentially, dynamically adjusting scheduling strategies to maximize resource utilization and minimize service latency. Given the growing demand for parallel processing of data analysis and visualization tasks, the DAG's task chain organization significantly improves the system's processing capabilities and response efficiency for highly concurrent and multi-dimensional requests. The DAG structure inherently eliminates circular dependencies, ensuring smooth execution of the entire task command chain without deadlocks or infinite loops caused by interdependent tasks. This is particularly important for the multi-step, multi-branch processing processes common in power system operations. For example, in complex data cleaning and multi-layer analysis processes, it automatically identifies all nodes without pre-depen- dents and prioritizes their execution. Once a node completes, it gradually triggers downstream dependent tasks, ensuring efficient, controllable, and secure business processes. By organizing task chains in a directed acyclic graph, it enables structured management of complex tasks.

[0040] In one embodiment, Figure 5 As shown, the following steps S51-S53 are also included: In step S51, after receiving the task instruction chain generated by the business disassembly agent, each subtask instruction is registered as a waiting execution state; In step S52, when all the pre-order dependency conditions of a subtask are met, the subtask is switched to the execution state and the visual presentation agent executes the subtask; In step S53, after the subtask is executed, it is converted to a completed state. If an error occurs during the execution, it is converted to an error state to trigger corresponding exception handling.

[0041] In one embodiment, the refined scheduling and full-process lifecycle control mechanism within task instruction chain execution management dynamically manages the status of each subtask node through an instruction registration center to ensure the orderly progress of the overall task and manageable exceptions. When a task instruction chain generated by a business decomposition agent is submitted to the instruction registration center, each subtask node in the chain is automatically registered as "waiting for execution." This not only enables global visualization of the task execution process but also provides a data foundation for subsequent state switching, task scheduling, and dependency analysis. All subtasks are statically distributed and can only be activated when their predecessor dependencies are met, avoiding the problems of unordered task triggering and resource waste in traditional scheduling. During the actual scheduling phase, the instruction registration center continuously monitors the dependency conditions of each subtask. When all predecessor dependent tasks of a subtask are completed, the task node's status is automatically switched from "waiting for execution" to "executing" and assigned to a visualization agent or other compatible execution module for actual operation. For example, in a power system big data analysis scenario, a "visualization presentation" subtask can only be initiated after all its predecessor task nodes, such as data analysis and anomaly detection, have been completed. This "dependency-driven, step-by-step activation" scheduling strategy not only improves the accuracy and security of process execution, but also significantly improves the efficiency of multi-task parallel processing.

[0042] During the actual execution phase of the task node, its status will be updated based on the running results of the subtask. When the subtask is successfully completed, its status is switched to "Completed"; if an error or exception occurs during the execution process, the node status will be automatically changed to "Error", and the corresponding exception handling process will be immediately triggered, effectively ensuring the robustness and recoverability of each link in the task chain. For example, when an analysis task fails due to missing data, the system can retry, switch to a backup process, or notify the user to intervene according to the preset exception handling strategy to prevent the exception from spreading to subsequent tasks or causing the overall process to be blocked. The above management mechanism realizes closed-loop control of the entire process from task generation to execution, completion, and even exception handling through dynamic switching and fine-grained monitoring of subtask status.

[0043] In one embodiment, Figure 6 As shown, the following steps S61-S63 are also included: In step S61, the visualization-related instructions in the task instruction chain are parsed into interface components of an interactive dashboard to display power system data analysis results; In step S62, a natural language summary report describing the analysis result is generated based on the large language model; In step S63, the interactive dashboard can support users to interactively query and dynamically visualize power system data.

[0044] In one embodiment, the key technical processes for intelligent visualization and report generation of power system data include automatic parsing of visualization instructions within task chains, intelligent interpretation of visualization results, and multi-dimensional enhancements to user interaction capabilities. The system automatically identifies data visualization-related instructions within task chains and parses these instructions into interface components that form interactive dashboards. Specifically, different types of data analysis results are automatically mapped into appropriate chart components, such as trend line charts, distribution histograms, heat maps, and equipment status dashboards, enabling multi-dimensional visualization of complex analysis conclusions. Automatic component parsing and dashboard generation eliminate the limitations of traditional manual configuration of chart templates, enabling the system to flexibly and dynamically organize and display results for different analysis tasks and data types. Furthermore, based on the generation of visualization components, the system leverages the natural language processing capabilities of a large-scale language model to intelligently interpret the analysis results and generate summary reports. The large-scale language model comprehensively analyzes the multi-dimensional data and key indicators in the dashboard, automatically generating concise, accurate, and insightful natural language summaries. For example, in a load anomaly analysis scenario, the model not only provides quantitative conclusions such as "this month's transformer load peak is higher than the historical average for the same period," but also automatically generates business explanations and recommendations based on anomalies and related factors, helping users quickly understand the underlying trends and potential risks in the data. The integrated "data-visualization-text" output significantly reduces the user's professional threshold and provides valuable support for intelligent decision-making. The generated interactive dashboard not only supports static data browsing but also boasts powerful user interaction and dynamic display capabilities. Users can customize filter conditions, adjust data ranges, link different chart perspectives, and even drill down into anomalies and query related data. This highly interactive and personalized data exploration experience enables the system to meet the business needs of diverse scenarios such as power operations, management, and decision-making, significantly improving data utilization efficiency and the depth of information acquisition. Through automated visual analysis, intelligent summary generation, and a rich interactive experience, the system establishes a new paradigm for intelligent presentation of power system data analysis, fully demonstrating technological advancement and creativity.

[0045] In one embodiment, Figure 7 This is a block diagram of a processing system based on business decomposition and visual presentation of dual agents according to an exemplary embodiment. Figure 7As shown, the dual-agent processing system based on business decomposition and visualization presentation includes a decomposition module 71, a visualization module 72 and a management module 73.

[0046] The disassembly module 71 is configured to receive a natural language request from the power system through a business disassembly agent, perform semantic parsing on the natural language request using a small language model, perform semantic retrieval based on the power system knowledge graph, implement knowledge-enhanced intent recognition and task disassembly, and generate a corresponding task instruction chain using a directed acyclic graph structure; The visualization module 72 is used to send the task instruction chain to the instruction registration center, which schedules the visualization presentation agent to generate an interactive dashboard and natural language summary through a large language model to complete the visualization presentation of power system data; The management module 73 is used to perform lifecycle management on the entire processing flow through the instruction registration center using a finite state machine to constrain the execution flow of the task instruction chain.

[0047] The disassembly module 71, visualization module 72 and management module 73 included in the processing system block diagram based on business disassembly and visualization presentation of dual intelligent bodies are controlled to execute the processing method based on business disassembly and visualization presentation of dual intelligent bodies described in any of the above embodiments.

[0048] like Figure 8 As shown, the present invention provides an electronic device 800, which includes: a communication interface, a processor 801, and a memory 802; In which, the memory 802 is used to store program instructions. When the program instructions are executed by the processor 801 that is communicatively connected to the memory 802 through the communication interface, the business disassembly agent receives the natural language request of the power system, performs semantic analysis on the natural language request through a small language model, performs semantic retrieval based on the power system knowledge graph, realizes knowledge-enhanced intention recognition and task decomposition, and generates a corresponding task instruction chain in a directed acyclic graph structure; sends the task instruction chain to the instruction registration center, and the instruction registration center schedules the visualization presentation agent to generate an interactive dashboard and natural language summary through a large language model to complete the visualization presentation of the power system data; and uses a finite state machine through the instruction registration center to perform life cycle management on the entire processing flow to constrain the execution process of the task instruction chain.

[0049] The present invention provides a computer-readable storage medium having computer program instructions stored thereon. When the computer program instructions are executed by a processor, a business disassembly agent receives a natural language request of an electric power system, performs semantic analysis on the natural language request through a small language model, performs semantic retrieval based on a knowledge graph of the electric power system, realizes knowledge-enhanced intention recognition and task disassembly, and generates a corresponding task instruction chain in a directed acyclic graph structure; sends the task instruction chain to an instruction registration center, and the instruction registration center schedules a visualization presentation agent, generates an interactive dashboard and a natural language summary through a large language model, so as to complete the visualization presentation of the electric power system data; and uses a finite state machine through the instruction registration center to perform lifecycle management on the entire processing flow to constrain the execution flow of the task instruction chain.

[0050] Should be understood that the specific features, operations and details described herein above regarding method of the present invention also may be similarly applied to system of the present invention, or, vice versa. In addition, each step of method of the present invention described above may be performed by the corresponding components or units of system of the present invention.

[0051] It should be understood that the various modules / units of the system of the present invention can be implemented in whole or in part by software, hardware, firmware, or a combination thereof. Each module / unit can be embedded in the processor of a computer device in the form of hardware or firmware or independent of the processor, or can be stored in the memory of a computer device in the form of software for the processor to call to perform the operations of each module / unit. Each module / unit can be implemented as an independent component or module, or two or more modules / units can be implemented as a single component or module.

[0052] In one embodiment, a computer device is provided, comprising a memory and a processor. The memory stores computer instructions executable by the processor, which, when executed by the processor, instruct the processor to perform the steps of the method according to an embodiment of the present invention. The computer device can be broadly defined as a server, a terminal, or any other electronic device with the necessary computing and / or processing capabilities. In one embodiment, the computer device can include a processor, memory, a network interface, a communication interface, etc. connected via a system bus. The processor of the computer device can be used to provide the necessary computing, processing, and / or control capabilities. The memory of the computer device can include a non-volatile storage medium and an internal memory. An operating system, a computer program, etc. can be stored in or on the non-volatile storage medium. The internal memory can provide an environment for the operation of the operating system and computer program in the non-volatile storage medium. The network interface and communication interface of the computer device can be used to connect to and communicate with external devices via a network. When the computer program is executed by the processor, the steps of the method according to the present invention are performed.

[0053] The present invention can be implemented as a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, causes the steps of the method of an embodiment of the present invention to be performed. In one embodiment, the computer program is distributed on a plurality of computer devices or processors coupled to a network so that the computer program is stored, accessed, and executed in a distributed manner by one or more computer devices or processors. A single method step / operation, or two or more method steps / operations, can be performed by a single computer device or processor or by two or more computer devices or processors. One or more method steps / operations can be performed by one or more computer devices or processors, and one or more other method steps / operations can be performed by one or more other computer devices or processors. One or more computer devices or processors can perform a single method step / operation, or perform two or more method steps / operations.

[0054] Those skilled in the art will appreciate that the method steps of the present invention can be performed by instructing related hardware, such as a computer device or processor, through a computer program. The computer program can be stored in a non-transitory computer-readable storage medium, and when the computer program is executed, the steps of the present invention are performed. Depending on the circumstances, any reference herein to memory, storage, database, or other media may include non-volatile and / or volatile memory. Examples of non-volatile memory include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (GGPROM), flash memory, magnetic tape, floppy disk, magneto-optical data storage device, optical data storage device, hard disk, solid-state disk, etc. Examples of volatile memory include random access memory (RAM), external cache memory, etc.

[0055] The various technical features described above can be combined arbitrarily. Although not all possible combinations of these technical features are described, any combination of these technical features should be considered to be covered by this specification as long as such combination does not conflict.

[0056] 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, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A dual-agent processing method based on business decomposition and visual presentation, characterized in that: include: The business disassembly agent receives natural language requests from the power system, performs semantic parsing on the natural language requests using a small language model, performs semantic retrieval based on the power system knowledge graph, implements knowledge-enhanced intent recognition and task disassembly, and generates the corresponding task instruction chain using a directed acyclic graph structure. The task instruction chain is sent to an instruction registration center, which dispatches a visualization presentation agent to generate an interactive dashboard and a natural language summary through a large language model to complete the visualization of power system data; The instruction registration center uses a finite state machine to perform lifecycle management on the entire processing flow to constrain the execution flow of the task instruction chain.

2. The dual-agent processing method based on business decomposition and visual presentation as claimed in claim 1, characterized in that: Also includes: Performing semantic parsing on the natural language request using a small language model; Performing a search-enhanced semantic retrieval on the semantic parsing result based on the power system knowledge graph to obtain power system domain knowledge related to the request; The acquired power system domain knowledge is applied to understanding the natural language request, and a corresponding task instruction chain is generated.

3. The dual-agent processing method based on business decomposition and visual presentation as claimed in claim 1, characterized in that: Also includes: Pre-training on general corpus to obtain the initial language model; Training the initial language model based on power system domain data to obtain a small language model optimized for the power system domain; The trained small language model is deployed into the business decomposition agent for performing intent recognition and task decomposition on natural language requests.

4. The dual-agent processing method based on business decomposition and visual presentation as claimed in claim 1, characterized in that: Also includes: Each node in the task instruction chain corresponds to a power system data analysis subtask or visualization subtask; In the directed acyclic graph structure, the directed connections between the nodes represent the dependency relationships between the subtasks; The directed acyclic graph structure ensures that there is no circular dependency during the execution of the task instruction chain.

5. The dual-agent processing method based on business decomposition and visual presentation as claimed in claim 1, characterized in that: Also includes: After receiving the task instruction chain generated by the business disassembly agent, each subtask instruction is registered as a waiting execution state; When all the pre-order dependency conditions of a subtask are met, the subtask is switched to the execution state and the visual presentation agent executes the subtask; After the subtask is executed, it is converted to the completed state. If an error occurs during the execution, it is converted to the error state to trigger the corresponding exception handling.

6. The dual-agent processing method based on business decomposition and visual presentation as claimed in claim 1, characterized in that: Also includes: Parsing visualization-related instructions in the task instruction chain into interface components of an interactive dashboard to display power system data analysis results; generating a natural language summary report describing the analysis results based on the large language model; The interactive dashboard can support users to interactively query and dynamically visualize power system data.

7. A dual-agent processing system based on business disassembly and visualization, characterized in that: include: A disassembly module is configured to receive natural language requests from the power system through a business disassembly agent, perform semantic parsing on the natural language requests using a small language model, perform semantic retrieval based on the power system knowledge graph, implement knowledge-enhanced intent recognition and task disassembly, and generate a corresponding task instruction chain using a directed acyclic graph structure; A visualization module is used to send the task instruction chain to an instruction registration center, which schedules a visualization presentation agent to generate an interactive dashboard and natural language summary through a large language model to complete the visualization of power system data; The management module is used to perform life cycle management on the entire processing flow through the instruction registration center using a finite state machine to constrain the execution flow of the task instruction chain.

8. The dual-agent processing system based on business decomposition and visualization as claimed in claim 7, characterized in that: The disassembly module, the disassembly module and the disassembly module are controlled to execute the processing method based on business disassembly and visual presentation dual agents as described in any one of claims 1 to 6.

9. An electronic device, characterized in that: include: Communication interface, processor, memory; Wherein, the memory is used to store program instructions, and when the program instructions are executed by the processor that is communicatively connected to the memory through the communication interface, the electronic device implements the processing method based on business disassembly and visual presentation of dual intelligent bodies as described in any one of claims 1 to 6.

10. A computer-readable storage medium having program instructions stored thereon, characterized in that: When the program instructions are executed by a computer, the computer implements the processing method based on business disassembly and visual presentation of dual intelligent agents as described in any one of claims 1 to 6.

Images