System and method for power industry knowledge questions and answers and computing equipment

Abstract

The invention provides a system and a method for power industry knowledge questions and answers and a computing device, the system is based on a multi-agent collaboration architecture of a large language model, the system is composed of a user interface layer, a communication and collaboration layer, an agent layer and a knowledge base layer, the user interface layer is used for receiving natural language questions and displaying answers generated by the system; the communication and cooperation layer configuration provides information transmission and cooperation support between intelligent agents; the intelligent agent layer is composed of a plurality of intelligent agents which work cooperatively, each intelligent agent bears a specific function module in the system, and the intelligent agents deal with problems of users through reasonable division of labor and close cooperation. The knowledge base layer stores core knowledge of the industry. According to the technical scheme of the invention, the intelligent level of a knowledge question-answering system in the power industry can be effectively improved, the processing capability of complex problems is improved, dynamic updating of knowledge is realized, an intelligent agent cooperation mechanism is established, and the security and privacy protection of the system are enhanced.

Description

Technical Field

[0001] The present invention relates to the fields of electric power and artificial intelligence, and in particular to a system and method and computing equipment for answering questions about electric power industry knowledge. Background Art

[0002] With the rapid development of artificial intelligence technology, large language models have made significant progress in the field of natural language processing. Large language models have powerful language understanding and generation capabilities and can effectively handle various complex language tasks, such as text generation, question-answering systems, and dialogue systems. In recent years, large language models have gradually been applied to knowledge question-answering systems in professional fields to support intelligent services in industries such as medicine, law, and finance.

[0003] However, in the power industry, the existing knowledge question-answering system still has significant deficiencies. Traditional knowledge question-answering systems in the power industry usually rely on a single language model or expert system, which makes it difficult to obtain the latest knowledge in real time and quickly integrate it into the system. The single model drive lacks a collaborative mechanism between intelligent agents and cannot provide customized responses based on the specific needs and question types of users.

[0004] To this end, a technical solution is needed that can effectively improve the intelligence level of the knowledge question-and-answer system in the power industry, enhance the ability to handle complex problems, realize the dynamic updating of knowledge, establish an intelligent body collaboration mechanism, provide personalized services, and enhance the security and privacy protection of the system to meet the industry's needs for efficient, accurate and personalized services. Summary of the invention

[0005] The present invention aims to provide a system, method and computing equipment for knowledge question and answering in the electric power industry, which can effectively improve the intelligence level of the knowledge question and answer system in the electric power industry, improve the ability to handle complex problems, realize dynamic updating of knowledge, establish an intelligent body collaboration mechanism, provide personalized services, and enhance the security and privacy protection of the system to meet the industry's needs for efficient, accurate and personalized services.

[0006] According to one aspect of the present invention, a system for knowledge question answering in the electric power industry is provided. The system is based on a multi-agent collaborative architecture of a large language model. The system consists of a user interface layer, a communication and collaboration layer, an agent layer, and a knowledge base layer. The agent layer includes a language understanding agent, a knowledge agent, a strategy agent, and a reasoning and interpretation agent, wherein:

[0007] The user interface layer is configured as follows:

[0008] Used to receive natural language questions;

[0009] Display the answers generated by the system;

[0010] The communication and collaboration layer is configured as follows:

[0011] Provide information transmission and collaboration support between intelligent agents;

[0012] The language understanding agent is configured as follows:

[0013] Responsible for parsing the natural language question and identifying the core content of the natural language question;

[0014] The strategy agent is configured as follows:

[0015] Dynamically assign tasks based on the complexity and type of the problem;

[0016] Dispatching other agents to jointly complete the problem processing;

[0017] The knowledge agent is configured as follows:

[0018] Manage and maintain the system's knowledge base;

[0019] Responsible for obtaining the latest industry knowledge from various sources and collating, updating and integrating it;

[0020] The reasoning and explanation agent is configured as follows:

[0021] Perform deep reasoning on complex questions and generate logical and explanatory answers;

[0022] The knowledge base layer is configured as follows:

[0023] Core knowledge of the storage industry.

[0024] According to some embodiments, the language understanding agent is further configured to:

[0025] Use word segmentation, part-of-speech tagging, and dependency syntax analysis techniques to parse the sentence structure of the user input and extract key information related to the natural language question and the user's intent;

[0026] The parsed key information and the user intention are passed to the strategy agent.

[0027] According to some embodiments, the policy agent is further configured to:

[0028] According to the parsing results provided by the language understanding agent, determine the complexity and domain of the natural language problem, and decide to call one or more corresponding agents to handle the problem;

[0029] Set priorities for different subtasks and respond quickly to urgent and / or complex tasks;

[0030] Manage the collaboration between various intelligent agents and adjust the calling order and frequency of each intelligent agent according to the real-time situation.

[0031] According to some embodiments, the strategy agent is further configured to: the strategy agent aggregates the results of all subtasks to generate a final answer.

[0032] According to some embodiments, the knowledge agent is further configured to:

[0033] Obtain information from various power industry-related literature, databases, and real-time monitoring data;

[0034] Convert information from different sources into different formats, clean data, and integrate semantics;

[0035] The timeliness of existing knowledge in the knowledge base layer is regularly checked and updated according to newly acquired knowledge.

[0036] According to some embodiments, the reasoning and explanation agent is further configured to:

[0037] Use rule-based reasoning, case-based reasoning, and model-based reasoning methods to handle natural language problems that require deep analysis;

[0038] Provide detailed answer analysis to explain the key steps and logical relationships in the reasoning process.

[0039] According to some embodiments, the knowledge base layer includes technical standards, regulations, case studies, and real-time data.

[0040] According to another aspect of the present invention, a method for electric power industry knowledge question and answer is provided, which is used in an electric power industry knowledge question and answer system, and the method comprises:

[0041] Perform the following operations through the user interface layer:

[0042] Used to receive natural language questions;

[0043] Display the answers generated by the system;

[0044] The following operations are performed through the communication and collaboration layer:

[0045] Provide information transmission and collaboration support between intelligent agents;

[0046] The language understanding agent performs the following operations:

[0047] Responsible for parsing the natural language question and identifying the core content of the natural language question;

[0048] The following operations are performed by the policy agent:

[0049] Dynamically assign tasks based on the complexity and type of the problem;

[0050] Dispatching other agents to jointly complete the problem processing;

[0051] The following operations are performed by the knowledge agent:

[0052] Manage and maintain the system's knowledge base;

[0053] Responsible for obtaining the latest industry knowledge from various sources and collating, updating and integrating it;

[0054] The reasoning and interpretation agent performs the following operations:

[0055] Perform deep reasoning on complex questions and generate logical and explanatory answers;

[0056] The following operations are performed through the knowledge base layer:

[0057] Core knowledge of the storage industry.

[0058] According to another aspect of the present invention, a computer program product is provided, comprising a computer program, wherein when the computer program is executed by a processor, the method as described above is implemented.

[0059] According to another aspect of the present invention, there is provided a computing device, comprising:

[0060] Processor; and

[0061] A memory stores a computer program, and when the computer program is executed by the processor, the method described in any one of the above items is implemented.

[0062] According to an embodiment of the present invention, in a multi-agent collaborative architecture based on a large language model, the user interface layer is responsible for receiving natural language questions and displaying answers generated by the system; the communication and collaboration layer is responsible for providing information transmission and collaboration support between agents; the language understanding agent is responsible for parsing the natural language questions and identifying the core content of the natural language questions; the knowledge agent is responsible for managing and maintaining the system's knowledge base, and is responsible for obtaining the latest industry knowledge from various sources, and organizing, updating and integrating it; the strategy agent is responsible for dynamically allocating tasks according to the complexity and type of the problem, and scheduling other agents to jointly complete the problem processing; the reasoning and interpretation agent is responsible for deep reasoning for complex problems and generating logical and explanatory answers; the knowledge base layer is responsible for storing the core knowledge of the industry. The present invention can effectively improve the intelligence level of the knowledge question-and-answer system in the power industry, improve the processing capacity of complex problems, realize the dynamic update of knowledge, establish an agent collaboration mechanism, provide personalized services, and enhance the security and privacy protection of the system, so as to meet the industry's needs for efficient, accurate and personalized services.

[0063] According to some embodiments, the collaborative work of multiple intelligent agents can effectively deal with complex problems involving multiple fields and multiple levels in the power industry. Each intelligent agent can work together according to its own specialized functions to solve complex query problems, which significantly improves the accuracy and efficiency of the system in complex environments.

[0064] According to some embodiments, the knowledge agent can automatically collect the latest laws, documents, technical reports, etc. of the power industry and quickly integrate this information into the system to ensure that the question-answering system is always based on the latest and most comprehensive industry knowledge base. This dynamic update mechanism avoids the problem of inaccurate or insufficient timeliness of answers due to knowledge lag in the existing system.

[0065] According to some embodiments, the multi-agent collaboration mechanism enables the system to flexibly respond to diverse problems. The strategic agent dynamically schedules agents such as language understanding, knowledge acquisition, reasoning, and interpretation to effectively handle various complex scenarios, thereby greatly improving the response speed, accuracy, and scalability of the question-answering system.

[0066] According to some embodiments, the strategy agent and explanation agent of the present invention can customize personalized answers according to user needs. This allows the system to provide targeted answers based on the professional background, complexity of the questions, and the level of detail required for explanation of different users, meet the user's needs for in-depth analysis and result explanation, and improve the user experience.

[0067] According to some embodiments, through encrypted communication protocols between intelligent entities and strict protection of sensitive data, the present invention ensures that the power industry knowledge question and answer system can effectively prevent the leakage or illegal access of sensitive information while providing intelligent services, thereby improving the security and reliability of the system.

[0068] According to some embodiments, based on the multi-agent architecture, the system can flexibly add new agents to adapt to the ever-changing needs of the power industry. By upgrading or adding specific functional agents, the system can quickly respond to new business scenarios or technical requirements, greatly enhancing the scalability and maintainability of the system.

[0069] It is to be understood that the foregoing general description and the following detailed description are exemplary only and are not restrictive of the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for describing the embodiments are briefly introduced below.

[0071] Figure 1 A schematic diagram of a system for answering industry knowledge questions according to an example embodiment is shown.

[0072] Figure 2a A flow chart of an industry knowledge question answering method according to an example embodiment is shown.

[0073] Figure 2b A schematic diagram showing the process of a multi-agent question-answering system according to an example embodiment.

[0074] Figure 3a A schematic diagram showing a language understanding agent structure according to an example embodiment.

[0075] Figure 3b A schematic diagram illustrating a policy agent architecture according to an example embodiment.

[0076] Figure 3c A schematic diagram illustrating a knowledge agent structure according to an example embodiment.

[0077] Figure 3d A schematic diagram showing a reasoning and interpretation agent architecture according to an example embodiment.

[0078] Figure 4 A block diagram of a computing device is shown according to an exemplary embodiment. DETAILED DESCRIPTION

[0079] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in many forms and should not be construed as limited to the embodiments set forth herein; rather, these embodiments are provided so that the present invention will be comprehensive and complete and fully convey the concepts of the example embodiments to those skilled in the art. The same reference numerals in the figures represent the same or similar parts, and thus their repeated description will be omitted.

[0080] In addition, the described features, structures or characteristics may be combined in one or more embodiments in any suitable manner. In the following description, many specific details are provided to provide a full understanding of the embodiments of the present invention. However, those skilled in the art will appreciate that the technical solution of the present invention can be practiced without one or more of the specific details, or other methods, components, devices, steps, etc. may be adopted. In other cases, known methods, devices, implementations or operations are not shown or described in detail to avoid blurring various aspects of the present invention.

[0081] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities may be implemented in software form, or in one or more hardware modules or integrated circuits, or in different networks and / or processor devices and / or microcontroller devices.

[0082] The flowcharts shown in the accompanying drawings are only exemplary and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps can be decomposed, and some operations / steps can be combined or partially combined, so the actual execution order may change according to actual conditions.

[0083] It should be understood that although the terms first, second, third, etc. may be used herein to describe various components, these components should not be limited by these terms. These terms are used to distinguish one component from another component. Therefore, the first component discussed below can be referred to as the second component without departing from the teachings of the present inventive concept. As used herein, the term "and / or" includes any one of the associated listed items and all combinations of one or more.

[0084] The user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in the present invention are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of relevant countries and regions, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0085] Those skilled in the art will appreciate that the drawings are merely schematic diagrams of example embodiments, and the modules or processes in the drawings are not necessarily necessary for implementing the present invention, and therefore cannot be used to limit the protection scope of the present invention.

[0086] In the power industry, the existing knowledge question-and-answer system still has significant deficiencies. Traditional knowledge question-and-answer systems in the power industry usually rely on a single language model or expert system. A single model cannot simultaneously possess comprehensive industry knowledge, contextual understanding, logical reasoning, and real-time dynamic response capabilities. It is difficult for existing systems to obtain the latest knowledge in real time. This lag affects the accuracy and timeliness of the system's answers and cannot meet the power industry's high requirements for information accuracy and timeliness. In the power industry, problems often involve knowledge in multiple fields and disciplines, such as power system design, operation and maintenance, and fault diagnosis. These problems require the system to have stronger collaboration and reasoning capabilities, but the current single model's way of dealing with problems is usually linear and static, and it is difficult to dynamically respond to complex user needs and problem scenarios. The answers of the current question-and-answer system cannot meet user expectations in terms of depth, breadth, and explanatoryness, especially when complex power technology problems need to be analyzed and explained in detail, the performance of the existing system is not ideal.

[0087] The current knowledge question-and-answer system in the power industry faces many challenges, such as how to improve the system's ability to handle complex problems, especially those involving multi-domain knowledge and deep reasoning; how to dynamically update knowledge to ensure that the system always answers based on the latest industry information; how to improve the system's responsiveness in diverse problem scenarios through the collaborative mechanism of intelligent agents; how to provide users with more personalized, explanatory and real-time intelligent services, etc. The existence of these challenges has greatly limited the breadth and depth of application of the existing knowledge question-and-answer system in the power industry. Therefore, there is an urgent need for an innovative technical means that can effectively improve the intelligence level of the knowledge question-and-answer system in the power industry and meet the industry's demand for efficient, accurate and personalized services.

[0088] To this end, the present invention proposes a system and method for knowledge question and answering in the electric power industry, which can effectively improve the intelligence level of the knowledge question and answer system in the electric power industry, improve the ability to handle complex problems, realize dynamic updating of knowledge, establish an intelligent body collaboration mechanism, provide personalized services, and enhance the security and privacy protection of the system to meet the industry's needs for efficient, accurate and personalized services.

[0089] Exemplary embodiments of the present invention are described below with reference to the accompanying drawings.

[0090] Figure 1 A schematic diagram of a system for electric power industry knowledge question answering according to an example embodiment is shown.

[0091] The knowledge question-answering system for the power industry adopts a multi-agent collaborative architecture based on a large language model, aiming to improve the ability to handle complex problems, dynamically update the knowledge base, achieve efficient agent collaboration, and provide users with personalized explanatory answers. The system consists of multiple functional agents, each of which is responsible for a specific task, and through collaboration, the entire process from user input of questions to generating the final answer is realized.

[0092] See also Figure 1 , Figure 1 The system shown is a multi-agent collaborative architecture based on a large language model. The system consists of a user interface layer, a communication and collaboration layer, an agent layer, and a knowledge base layer. The agent layer includes a language understanding agent, a knowledge agent, a strategy agent, and a reasoning and interpretation agent.

[0093] According to some embodiments, the user interface layer is configured to receive natural language questions and display answers generated by the system. The communication and collaboration layer is configured to provide information transfer and collaboration support between agents. The knowledge base layer is configured to store core knowledge of the industry.

[0094] The user interface layer is used to receive natural language questions input by users and display the answers generated by the system. The user interface guides users to ask questions through intuitive interactions and provides users with clear explanations and feedback during the answering process.

[0095] The knowledge base layer is used to store the core knowledge of the power industry, including technical standards, regulations, case analysis, real-time data, etc. The knowledge base is managed by knowledge agents and updated regularly to ensure the accuracy and timeliness of information.

[0096] The communication and collaboration layer provides support for information transmission and collaboration between intelligent agents. By designing a reasonable collaboration mechanism, it ensures that each intelligent agent can efficiently exchange information and work together, thereby improving the overall performance of the system.

[0097] According to some embodiments, the agent layer includes language understanding agent, knowledge agent, strategy agent, reasoning and interpretation agent. The agent layer is the core part of the system and is composed of multiple agents working together. Each agent undertakes a specific functional module in the system, and they handle user problems through reasonable division of labor and close collaboration.

[0098] The language understanding agent is responsible for parsing the natural language questions input by the user and extracting the key information and intent. This agent is the front-end entrance of the system, and its accuracy directly affects the task execution of subsequent agents. The structure of the language understanding agent is shown in Figure 3a .

[0099] The language understanding agent is configured to: parse the natural language question and identify the core content of the natural language question. The language understanding agent is also configured to: use word segmentation, part-of-speech tagging and dependency syntax analysis technology to parse the sentence structure of the user input, extract key information and user intent related to the natural language question; and pass the parsed key information and user intent to the strategy agent. The language understanding agent is responsible for parsing the natural language input by the user, identifying the core content and intent of the question, and providing accurate semantic information for subsequent processing steps.

[0100] The strategic agent is the decision-making core of the system, responsible for dynamically scheduling other agents to collaborate based on the complexity and type of the user's questions. The agent's adaptive scheduling mechanism ensures that the system can perform tasks with the best configuration in different scenarios. See the strategic agent architecture for details. Figure 3b .

[0101] The strategy agent is configured to: dynamically assign tasks according to the complexity and type of the problem, and dispatch other agents to jointly complete the problem processing. The strategy agent is also configured to: determine the complexity and field of the natural language problem based on the parsing results provided by the language understanding agent, and decide to call one and / or more corresponding agents to handle the problem; set priorities for different subtasks, and respond quickly to urgent tasks and / or complex tasks; manage the collaboration between various agents, and adjust the calling order and frequency of each agent according to the real-time situation. The strategy agent summarizes the results of all subtasks and generates the final answer. The strategy agent dynamically assigns tasks according to the complexity and type of the problem, and dispatches other agents to jointly complete the problem processing. The agent has an adaptive scheduling mechanism that can optimize the calling order and strategy of the agent in real time.

[0102] The knowledge agent is the core module in the system responsible for managing and maintaining the knowledge base. Its main responsibilities include obtaining information from various literature, databases, and real-time monitoring data related to the power industry, and integrating and updating this information into the knowledge base. The knowledge agent ensures that the answers of the question-answering system are based on the latest and most authoritative industry knowledge. The structure diagram of the agent is shown in Figure 3c .

[0103] The knowledge agent is configured to: manage and maintain the system's knowledge base, and is responsible for obtaining the latest industry knowledge from various sources, and organizing, updating and integrating it. The knowledge agent is also configured to: obtain information from various power industry-related literature, databases, and real-time monitoring data; perform format conversion, data cleaning and semantic integration of information from different sources; regularly check the timeliness of existing knowledge in the knowledge base layer, and update it based on newly acquired knowledge. The knowledge agent manages and maintains the system's knowledge base, and is responsible for obtaining the latest industry knowledge from various sources, and organizing, updating and integrating it to ensure that the system's answers are based on the latest authoritative information.

[0104] The Reasoning and Explanation Agent focuses on handling complex power industry problems, especially scenarios that require logical reasoning. The agent not only generates answers, but is also responsible for providing clear explanatory content to help users understand the background and reasons for complex technical problems. Figure 3d .

[0105] The reasoning and explanation agent is configured to: perform deep reasoning on complex problems and generate logical and explanatory answers. The reasoning and explanation agent is also configured to: use rule reasoning, case reasoning and model-based reasoning methods to handle natural language problems that require deep analysis; provide detailed answer analysis to explain the key steps and logical relationships in the reasoning process. The reasoning and explanation agent performs deep reasoning on complex problems and generates logical and explanatory answers to help users understand complex power technology problems.

[0106] Multiple agents work together, each taking on specific tasks. Each agent is designed to meet the special needs of the power industry, ensuring that the system can answer complex professional questions efficiently and accurately. Through a high degree of collaboration, the agents work together to complete the analysis, knowledge retrieval, reasoning and answer generation of complex problems. The collaborative process of the entire system starts with the user inputting the question, and finally returns an explanatory answer after multiple steps such as language understanding, task allocation, knowledge retrieval, reasoning generation, etc.

[0107] An example of an agent collaboration process is described below.

[0108] The user inputs a question through the interface layer, and the language understanding agent immediately parses the question and extracts the core semantic information and user intention. This is the starting point of the agent collaboration and directly affects the subsequent work of the agent.

[0109] The parsing results are passed to the policy agent, which analyzes the complexity of the problem and decides which agents need to be called to participate in the processing. For simple problems, the knowledge agent may only need to retrieve the answer directly from the knowledge base; for complex problems, it may be necessary to call both reasoning and explanation agents for in-depth analysis.

[0110] The strategy agent assigns simple problem-related tasks to the knowledge agent, which retrieves information related to the problem from the knowledge base and integrates preliminary answers or supporting materials and feeds them back to the strategy agent. For complex problems, the strategy agent passes preliminary information to the reasoning and explanation agent, which then conducts further logical reasoning and generates detailed explanatory answers. This process usually involves a deep learning model or a rule-based reasoning system, which selects the appropriate reasoning path based on the type and complexity of the problem.

[0111] The policy agent aggregates all subtask results and generates a final answer. The answer includes not only direct answers, but also explanations, references, or further action suggestions. This information is returned to the user through the user interface, and user feedback is collected in the process to optimize subsequent agent collaboration.

[0112] The collaborative process of agents is achieved through semantic-based task allocation, task relay between agents, dynamic decision-making of strategic agents, multi-agent parallel processing or prompts to ensure that each agent can complete the task efficiently at the right time.

[0113] Semantic-based task assignment: The policy agent uses prompts based on semantic analysis to determine the complexity and domain of the problem. For example, the policy agent may use the following prompt for internal processing: "Based on the semantic structure of the question, determine whether the question requires deep reasoning. If necessary, call the reasoning and explanation agent and prepare preliminary knowledge retrieval results."

[0114] Task relay between agents: Agents relay tasks through clear interfaces and prompts. For example, after completing the retrieval, the knowledge agent passes the results to the reasoning and explanation agent using the following prompt: "Based on the retrieved documents and data, further infer the root cause of the problem and generate an explanatory answer."

[0115] Dynamic decision-making of the strategic agent: The strategic agent makes decisions based on real-time feedback at each collaborative stage. For example: "If the information returned by the knowledge agent is not enough to directly answer the question, the reasoning and explanation agent is activated for further analysis."

[0116] Multi-agent parallel processing: For complex problems, the strategy agent may call multiple agents in parallel. For example, in a complex power fault diagnosis scenario, the strategy agent may simultaneously trigger two operations: "the knowledge agent retrieves the historical fault data of the power system" and "the reasoning and interpretation agent infers the possible fault cause based on the current fault characteristics."

[0117] This prompt-based agent collaboration mechanism ensures that the system can flexibly respond to a variety of questions and provide accurate, detailed, and explanatory answers.

[0118] In view of the limitations of the existing knowledge question-answering system in the power industry in dealing with complex problems, dynamically updating knowledge, intelligent agent collaboration, and personalized responses, the present invention aims to solve the following key technical problems:

[0119] Improve the ability to handle complex problems. Existing single language models are difficult to deal with complex problems in the power industry, especially queries involving multi-domain knowledge. To this end, the present invention proposes a system architecture based on a multi-agent collaboration mechanism, which combines agents with different functions to collaboratively handle complex power industry problems. This multi-agent collaboration model can effectively integrate various types of professional knowledge and reasoning capabilities, thereby greatly improving the system's ability to handle complex problems.

[0120] Dynamic updating and continuous learning of knowledge are achieved. The power industry has frequent technological updates, and existing systems are difficult to capture the latest information in real time, resulting in a decrease in the accuracy and timeliness of questions and answers. The present invention designs an intelligent agent network in which a specialized knowledge agent is responsible for dynamically acquiring, analyzing, and integrating the latest industry knowledge, and automatically updating this knowledge to the system. This mechanism ensures that the question-and-answer system can always answer based on the latest industry information, improving the accuracy and timeliness of the system.

[0121] Establishment of the agent collaboration mechanism. Most existing question-answering systems rely on the linear processing method of a single model and lack the ability to collaborate between agents. The present invention forms a multi-agent collaboration network by introducing multiple agents with different functions, such as knowledge agents, language understanding agents, strategy agents, and reasoning agents. This network can dynamically allocate and coordinate the tasks of agents through adaptive communication protocols, thereby improving the system's response speed and accuracy to complex queries.

[0122] Provide personalized and highly explanatory intelligent services. Existing systems are deficient in personalized services and result interpretation, and it is difficult to meet users' needs for in-depth analysis of complex problems. The present invention combines strategic agents and explanatory agents to automatically adjust the depth and breadth of answers according to user needs and generate detailed explanatory answers. This mechanism not only improves user satisfaction, but also enhances the applicability of the system to complex power problems.

[0123] Enhance the security and privacy protection of the system. The power industry involves a large amount of sensitive information, and the existing system's measures for information security and privacy protection are still incomplete. This invention ensures that the system can effectively protect the sensitive information of users and the industry while providing intelligent services through a secure communication protocol between intelligent agents and an encrypted transmission and storage mechanism for sensitive data.

[0124] As described above, the present invention solves the deficiencies of the existing technology in the knowledge question-answering system of the power industry by designing a large-scale language model question-answering system based on multi-agent collaboration, specifically including improving the processing ability of complex problems, realizing the dynamic update of knowledge, establishing an intelligent agent collaboration mechanism, providing personalized services, and enhancing the security and privacy protection of the system. The system will significantly improve the intelligence level and user experience of knowledge question-answering in the power industry, and meet the industry's demand for efficient, accurate, real-time, and personalized services.

[0125] Figure 2a A flow chart illustrating a method for answering questions about the electric power industry according to an example embodiment is shown.

[0126] Figure 2b A schematic diagram showing the process of a multi-agent question-answering system according to an example embodiment.

[0127] See also Figure 2a , Figure 2a A method for answering questions about the power industry is shown, which is used in the knowledge question-answering system for the power industry. The system workflow starts with the user raising a question, and after collaborative processing by multiple agents, the answer is finally generated and displayed to the user. The process of the multi-agent question-answering system can be found in Figure 2b .

[0128] According to some embodiments, the method performs the following operations through the user interface layer: for receiving natural language questions; displaying answers generated by the system. The method performs the following operations through the communication and collaboration layer: providing information transmission and collaboration support between agents. The method performs the following operations through the language understanding agent: responsible for parsing the natural language questions and identifying the core content of the natural language questions. The method performs the following operations through the strategy agent: dynamically assigning tasks according to the complexity and type of the questions; scheduling other agents to jointly complete the processing of the questions. The method performs the following operations through the knowledge agent: managing and maintaining the knowledge base of the system; responsible for obtaining the latest industry knowledge from various sources, and organizing, updating and integrating it. The method performs the following operations through the reasoning and interpretation agent: performing deep reasoning on complex questions and generating logical and explanatory answers. The method performs the following operations through the knowledge base layer: storing the core knowledge of the industry.

[0129] At S201 , the user inputs a question.

[0130] According to some embodiments, a user inputs a natural language question through an interface layer, and the system begins parsing the input.

[0131] At S203, the user interface layer passes the input question to the language understanding agent.

[0132] According to some embodiments, after receiving the natural language question from the user interface layer, the language understanding agent starts to parse the natural language question and identify the core content of the natural language question.

[0133] In S205, the language understanding agent extracts key information and intent, and passes the key information and intent to the strategy agent.

[0134] According to some embodiments, the language understanding agent performs natural language processing, semantic analysis, and intent recognition, and transmits the extracted key information and intent to the strategy agent.

[0135] In S207, the strategy agent dispatches the knowledge agent and the reasoning and explanation agent to handle the problem.

[0136] According to some embodiments, the strategy agent determines the complexity of the problem based on the parsing results, and dynamically schedules the knowledge agent and the reasoning and explanation agent to handle the problem.

[0137] In S209, the knowledge agent retrieves and integrates information in the knowledge base, returns a preliminary answer, and the reasoning and explanation agent generates a detailed explanation.

[0138] According to some embodiments, the knowledge agent retrieves relevant information from the knowledge base, integrates it, and passes preliminary answers or relevant information back to the strategy agent. For complex problems, the reasoning and explanation agent performs logical reasoning based on the information provided by the knowledge agent and generates detailed explanatory answers.

[0139] In S211, the policy agent summarizes the answers and returns them to the user interface layer.

[0140] According to some embodiments, the strategy agent summarizes the final answer generated in S209 and presents it to the user through the user interface layer.

[0141] At S213 , the user interface layer displays the answer to the user and receives feedback provided by the user.

[0142] According to some embodiments, the user may continue to ask in-depth questions or regenerate answers based on the answers received.

[0143] At S215 , the user interface transmits the collected feedback to the policy agent to further optimize the service.

[0144] According to some embodiments, the system collects user feedback and continues to further optimize the service based on user requirements and questions.

[0145] The architectural design ensures that the system can efficiently handle complex power industry issues and provide accurate, timely and personalized answers through the collaboration of intelligent agents and dynamic updating of knowledge.

[0146] This invention innovatively introduces a multi-agent collaboration mechanism to solve complex problems in the knowledge question-answering system of the power industry. Each agent performs its duties in a specific field, and through collaborative work, efficient and accurate question-answering processing is achieved, which significantly improves the flexibility and processing power of the system. By designing a strategic agent, it is possible to adaptively schedule various functional agents to collaborate according to the complexity of the problem and user needs. This dynamic scheduling mechanism ensures that the system can respond with the optimal configuration in various scenarios, improving the efficiency and accuracy of the system.

[0147] In the present invention, the knowledge agent is responsible for automatically acquiring and updating the latest knowledge of the power industry and quickly integrating this knowledge into the question-answering system. This mechanism ensures that the system can provide accurate answers based on the latest information, overcoming the shortcomings of the existing system's knowledge lag. In view of the sensitive information of the industry, the present invention introduces an encrypted communication protocol between agents and designs a strict access control mechanism to ensure data security and privacy protection. This multi-level security design effectively prevents data leakage and illegal access.

[0148] The existing single language model is difficult to deal with complex problems in the power industry, especially queries involving multi-domain knowledge. To this end, the present invention proposes a system architecture based on a multi-agent collaboration mechanism, which combines agents with different functions to collaboratively handle complex power industry problems. This multi-agent collaboration model can effectively integrate various types of professional knowledge and reasoning capabilities, thereby greatly improving the system's ability to handle complex problems.

[0149] Industry technology is frequently updated, and existing systems are difficult to capture the latest information in real time, resulting in a decrease in the accuracy and timeliness of questions and answers. The industry involves a large amount of sensitive information, and the existing systems have incomplete measures for information security and privacy protection. The present invention ensures that the system can effectively protect the sensitive information of users and industries while providing intelligent services through a secure communication protocol between agents and an encrypted transmission and storage mechanism for sensitive data. The present invention designs an agent network in which a specialized knowledge agent is responsible for dynamically acquiring, analyzing and integrating the latest industry knowledge and automatically updating this knowledge to the system. This mechanism ensures that the question-and-answer system can always answer based on the latest industry information, improving the accuracy and timeliness of the system. Most existing question-and-answer systems rely on the linear processing method of a single model and lack the ability to collaborate between agents. The present invention forms a multi-agent collaborative network by introducing multiple agents with different functions, such as knowledge agents, language understanding agents, strategy agents and reasoning agents. This network can dynamically allocate and coordinate the tasks of agents through an adaptive communication protocol, thereby improving the system's response speed and accuracy to complex queries.

[0150] Existing systems are deficient in personalized services and result interpretation, and it is difficult to meet users' needs for in-depth analysis of complex problems. The present invention combines strategic agents and explanatory agents to automatically adjust the depth and breadth of answers according to user needs and generate detailed explanatory answers. This mechanism not only improves user satisfaction, but also enhances the applicability of the system to complex power problems.

[0151] The present invention solves the deficiencies of existing technologies in the knowledge question-and-answer system of the power industry, including improving the processing capability of complex problems, realizing the dynamic update of knowledge, establishing an intelligent body collaboration mechanism, providing personalized services, and enhancing the security and privacy protection of the system. The system will significantly improve the intelligence level and user experience of the knowledge question-and-answer system in the power industry, and meet the industry's demand for efficient, accurate, real-time, and personalized services.

[0152] Figure 3a A schematic diagram showing a language understanding agent structure according to an example embodiment.

[0153] See also Figure 3a , Figure 3a Shows the main functions of a language understanding agent.

[0154] Natural language processing: Use technologies such as word segmentation, part-of-speech tagging, and dependency syntax analysis to parse the sentence structure of user input and extract the core content related to the question. Semantic analysis: Identify the key terms and concepts in the user's questions, especially the professional terms in the power industry, and understand their contextual meaning. Intent recognition: Based on semantic analysis, determine the user's real needs, such as whether they are asking about the cause of a power system failure or need some kind of operational guidance. Information extraction and transmission: Pass the parsed key information and user intent to the strategy agent to provide a basis for the subsequent task allocation of the agent.

[0155] Figure 3b A schematic diagram illustrating a policy agent architecture according to an example embodiment.

[0156] See also Figure 3b , Figure 3b Shows the main functions of the policy agent.

[0157] Task allocation: Based on the analysis results provided by the language understanding agent, determine the complexity and domain of the problem, and decide which agent to call to handle the problem. Priority management: Set priorities for different tasks to ensure that urgent or complex problems can be responded to quickly. Adaptive scheduling: Adjust the calling order and frequency of agents according to real-time conditions to optimize system resource utilization and response time. Collaboration and coordination: Manage the collaboration between agents to ensure smooth information flow and avoid duplication of work or resource conflicts.

[0158] Figure 3c A schematic diagram illustrating a knowledge agent structure according to an example embodiment.

[0159] See also Figure 3c , Figure 3cThe main functions of the knowledge agent are shown. In the power industry knowledge question-and-answer system of the present invention, the acquisition and update of knowledge is the key link for the system to maintain high accuracy and timeliness. The knowledge agent is responsible for automatically acquiring the latest power industry knowledge from multiple information sources and integrating this knowledge into the system's knowledge base to ensure that the system can always provide answers based on the latest information.

[0160] Knowledge acquisition of knowledge agents: Automatically obtain the latest information from various data sources (such as industry standards, technical reports, regulatory documents, real-time monitoring data, etc.). The technologies used include web crawling, API interface calls, database queries, etc.

[0161] Knowledge agents obtain diverse data sources. Knowledge agents regularly extract information from official standards and regulatory documents of the power industry. These documents are usually publicly available through government websites, industry associations’ publishing platforms, etc. Knowledge agents can parse the structure and content of these documents and convert them into knowledge items that can be used by the system. Knowledge agents obtain the latest technological advances and research results from academic databases, conference proceedings, and reports from research institutions. This information can enrich the cutting-edge technologies and theories in the knowledge base and provide more in-depth answers to the system. During the operation of the power system, knowledge agents can obtain data (such as power load, equipment status, etc.) from the real-time monitoring system. These data can provide the latest actual situation support for answering questions, especially in fault diagnosis and prediction scenarios. Knowledge agents can also access expert systems or case libraries in the power industry, incorporate empirical knowledge and historical cases into the knowledge base, and provide empirical support for the system to handle complex problems.

[0162] Knowledge agents can automatically crawl information. Knowledge agents are configured with web crawlers that can automatically crawl public information (such as news, reports, announcements, etc.) on specific websites. The crawlers will run regularly to ensure that the system can capture the latest industry trends. Knowledge agents obtain information from external databases in real time by calling public or authorized API interfaces. This method is particularly suitable for obtaining structured data, such as equipment performance parameters, real-time electricity market prices, etc.

[0163] Knowledge agents can acquire knowledge based on natural language processing and information extraction. Knowledge agents use natural language processing technology to parse text content, extract key industry terms, definitions, technical parameters and other information, and store this information in a structured manner in the knowledge base. Through entity recognition and relationship extraction technology, knowledge agents can identify important entities (such as equipment names, technical terms) and their relationships from unstructured text, build a knowledge graph, and enhance the relevance and query capabilities of the knowledge base.

[0164] Knowledge integration of knowledge agents: convert the format of information from different sources, clean the data and integrate the semantics to ensure that the entries in the knowledge base are accurate and non-redundant. Knowledge update: regularly check the timeliness of existing knowledge and update it according to the newly acquired knowledge to maintain the timeliness and effectiveness of the knowledge base.

[0165] To ensure that the system can answer questions based on the latest information, the knowledge agent implements an efficient knowledge update and integration mechanism, including regular updates and reviews, knowledge integration and optimization, and real-time updates and notifications.

[0166] The knowledge agent sets regular automatic update tasks, periodically obtains the latest information from the configured data source, compares it with the entries in the existing knowledge base, and automatically updates outdated or redundant information. For key knowledge entries that are automatically updated, the system also sets up an expert review mechanism. Before the updated entries are officially included in the knowledge base, they must be reviewed by power industry experts to ensure the accuracy and authority of the information.

[0167] The knowledge agent performs semantic analysis and fusion of knowledge items from different sources to avoid duplicate storage and ensure the consistency of information. For example, after fusion of different version descriptions of the same technology, the system only retains the latest and most comprehensive entries. In order to improve the query efficiency of the knowledge base, the knowledge agent regularly optimizes the index of knowledge items, using technologies such as inverted index and multi-level cache to ensure that the system can quickly respond to user questions.

[0168] For areas with high real-time requirements, such as power equipment monitoring and load forecasting, the knowledge agent can obtain the latest data in real time and update it to the knowledge base immediately to ensure that the answers provided by the system are based on the latest actual situation. The system provides users with notification services for knowledge updates, especially when it involves important regulatory changes or key technology updates. Users will receive relevant notifications. The knowledge agent also maintains detailed change records, and users can view the evolution of historical knowledge.

[0169] Knowledge storage and management of knowledge agents: classify, index and store the knowledge base, optimize query efficiency, and support efficient knowledge retrieval.

[0170] Through the above knowledge acquisition and updating mechanism, the system can maintain the efficiency and accuracy of its knowledge base, ensure that users get answers based on the latest industry information when consulting, and meet the power industry's high requirements for information timeliness and authority.

[0171] Figure 3d A schematic diagram showing a reasoning and interpretation agent architecture according to an example embodiment.

[0172] See also Figure 3d , Figure 3dShow the main functions of reasoning and explanation agents.

[0173] Logical reasoning: Use rule-based reasoning, case-based reasoning, or model-based reasoning methods to handle problems that require in-depth analysis, such as fault diagnosis, system optimization, etc. Explanation generation: Provide users with detailed answer analysis, explain the key steps and logical relationships in the reasoning process, and enhance the credibility and understandability of the answer. Interactive feedback: When interacting with users, dynamically adjust the reasoning process and explanation content based on user feedback to ensure the pertinence and accuracy of the answer.

[0174] Figure 4 A block diagram of a computing device is shown according to an exemplary embodiment.

[0175] like Figure 4 As shown, computing device 30 includes processor 12 and memory 14. Computing device 30 may also include bus 22, network interface 16, and I / O interface 18. Processor 12, memory 14, network interface 16, and I / O interface 18 may communicate with each other via bus 22.

[0176] The processor 12 may include one or more general-purpose CPUs (Central Processing Units, processors), microprocessors, or application-specific integrated circuits, etc., for executing relevant program instructions. According to some embodiments, the computing device 30 may also include a high-performance graphics card (GPU) 20 for accelerating the processor 12.

[0177] The memory 14 may include a machine system readable medium in the form of a volatile memory, such as a random access memory (RAM), a read-only memory (ROM) and / or a cache memory. The memory 14 is used to store one or more programs including instructions and data. The processor 12 can read the instructions stored in the memory 14 to execute the above-mentioned method according to the embodiment of the present invention.

[0178] The computing device 30 may also communicate with one or more networks via the network interface 16. The network interface 16 may be a wireless network interface.

[0179] The bus 22 may include an address bus, a data bus, a control bus, etc. The bus 22 provides a path for exchanging information between components.

[0180] It should be noted that, in the specific implementation process, the computing device 30 may also include other components necessary for normal operation. In addition, those skilled in the art may understand that the above device may only include components necessary for implementing the embodiments of this specification, and need not include all components shown in the figure.

[0181] The present invention also provides a computer-readable storage medium having a computer program stored thereon, which implements the steps of the above method when executed by a processor. The computer-readable storage medium may include, but is not limited to, any type of disk, including a floppy disk, an optical disk, a DVD, a CD-ROM, a microdrive, and a magneto-optical disk, a ROM, a RAM, an EPROM, an EEPROM, a DRAM, a VRAM, a flash memory device, a magnetic card or an optical card, a nanosystem (including a molecular memory IC), a network storage device, a cloud storage device, or any type of medium or device suitable for storing instructions and / or data.

[0182] An embodiment of the present invention further provides a computer program product, which includes a computer program. The computer program is operable to enable a computer to execute part or all of the steps of any one of the methods described in the above method embodiments.

[0183] Those skilled in the art can clearly understand that the technical solution of the present invention can be implemented with the help of software and / or hardware. "Unit" and "module" in this specification refer to software and / or hardware that can independently complete or cooperate with other components to complete specific functions, where the hardware can be, for example, a field programmable gate array, an integrated circuit, etc.

[0184] It should be noted that, for the above-mentioned method embodiments, for the sake of simplicity, they are all described as a series of action combinations, but those skilled in the art should know that the present invention is not limited by the described action sequence, because according to the present invention, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present invention.

[0185] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0186] In the several embodiments provided by the present invention, it should be understood that the disclosed devices can be implemented in other ways. For example, the device embodiments described above are only schematic, such as the division of units, which is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some service interfaces, and the indirect coupling or communication connection of devices or units can be electrical or other forms.

[0187] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0188] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0189] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or the whole or part of the technical solution can be embodied in the form of a software product, which is stored in a memory and includes several instructions for a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of the methods of various embodiments of the present invention.

[0190] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0191] The exemplary embodiments of the present invention are specifically shown and described above. It should be understood that the present invention is not limited to the detailed structures, configurations or implementations described herein; on the contrary, the present invention is intended to cover various modifications and equivalent configurations included in the spirit and scope of the appended clauses.

Claims

1. A system for knowledge question answering in the power industry, the system is based on a multi-agent collaborative architecture of a large language model, the system consists of a user interface layer, a communication and collaboration layer, an agent layer and a knowledge base layer, the agent layer includes a language understanding agent, a knowledge agent, a strategy agent, and a reasoning and interpretation agent, wherein: The user interface layer is configured as follows: Used to receive natural language questions; Display the answers generated by the system; The communication and collaboration layer is configured as follows: Provide information transmission and collaboration support between intelligent agents; The language understanding agent is configured as follows: Responsible for parsing the natural language question and identifying the core content of the natural language question; the strategy agent is configured as follows: Dynamically assign tasks based on the complexity and type of the problem; Dispatching other agents to jointly complete the problem processing; The knowledge agent is configured as follows: Manage and maintain the system's knowledge base; Responsible for obtaining the latest industry knowledge from various sources and collating, updating and integrating it; The reasoning and explanation agent is configured as follows: Perform deep reasoning on complex questions and generate logical and explanatory answers; The knowledge base layer is configured as follows: Core knowledge of the storage industry.

2. The system according to claim 1, characterized in that The language understanding agent is also configured to: Use word segmentation, part-of-speech tagging, and dependency syntax analysis techniques to parse the sentence structure of the user input and extract key information related to the natural language question and the user's intent; The parsed key information and the user intention are passed to the strategy agent.

3. The system according to claim 1, characterized in that The strategy agent is also configured to: According to the parsing results provided by the language understanding agent, determine the complexity and domain of the natural language problem, and decide to call one or more corresponding agents to handle the problem; Set priorities for different subtasks and respond quickly to urgent and / or complex tasks; Manage the collaboration between various intelligent agents and adjust the calling order and frequency of each intelligent agent according to the real-time situation.

4. The system according to claim 1, characterized in that The strategic agent is also configured to aggregate the results of all subtasks to generate a final answer.

5. The system according to claim 1, characterized in that The knowledge agent is also configured as: Obtain information from various power industry-related literature, databases, and real-time monitoring data; Convert information from different sources into different formats, clean data, and integrate semantics; The timeliness of existing knowledge in the knowledge base layer is regularly checked and updated according to newly acquired knowledge.

6. The system according to claim 1, characterized in that The reasoning and explanation agent is also configured as: Use rule-based reasoning, case-based reasoning, and model-based reasoning methods to handle natural language problems that require deep analysis; Provide detailed answer analysis to explain the key steps and logical relationships in the reasoning process.

7. The system according to claim 1, characterized in that The knowledge base layer includes technical standards, regulations, case analysis and real-time data.

8. A method for electric power industry knowledge question and answer, used in an electric power industry knowledge question and answer system, the method comprising: Perform the following operations through the user interface layer: Used to receive natural language questions; Display the answers generated by the system; The following operations are performed through the communication and collaboration layer: Provide information transmission and collaboration support between intelligent agents; The language understanding agent performs the following operations: Responsible for parsing the natural language question and identifying the core content of the natural language question; performing the following operations through the strategy agent: Dynamically assign tasks based on the complexity and type of the problem; Dispatching other agents to jointly complete the problem processing; The following operations are performed by the knowledge agent: Manage and maintain the system's knowledge base; Responsible for obtaining the latest industry knowledge from various sources and collating, updating and integrating it; The reasoning and interpretation agent performs the following operations: Perform deep reasoning on complex questions and generate logical and explanatory answers; The following operations are performed through the knowledge base layer: Core knowledge of the storage industry.

9. A computer program product, characterized in that The method comprises a computer program, which implements the method according to any one of claims 8 when executed by a processor.

10. A computing device, characterized in that include: processor; as well as A memory storing a computer program, wherein when the computer program is executed by the processor, the method according to any one of claims 8 is implemented.

Images