CN120264329B - 5G private network operation and maintenance method and related device based on large model adaptive capability - Google Patents
5G private network operation and maintenance method and related device based on large model adaptive capabilityInfo
- Publication number
- CN120264329B CN120264329B CN202510729968.6A CN202510729968A CN120264329B CN 120264329 B CN120264329 B CN 120264329B CN 202510729968 A CN202510729968 A CN 202510729968A CN 120264329 B CN120264329 B CN 120264329B
- Authority
- CN
- China
- Prior art keywords
- maintenance
- network operation
- network
- private network
- service
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The application discloses a 5G private network operation and maintenance method based on large model self-adaptation capability and a related device, relating to the field of fusion of large models and communication, wherein the disclosed method comprises the steps of obtaining a network operation and maintenance request for the 5G private network; the method comprises the steps of searching a network operation and maintenance knowledge base based on network operation and maintenance problems in a network operation and maintenance request to obtain context information corresponding to the network operation and maintenance problems, searching application perception data corresponding to the network operation and maintenance problems by utilizing a network operation and maintenance intelligent body, constructing prompt words for a large model based on the application perception data, inputting the network operation and maintenance problems, the context information and the prompt words into the large model, and generating a solution and operation and maintenance strategy for the network operation and maintenance problems based on the context information and the prompt words by utilizing the large model. The application improves the self-adaptive capacity of the large model in the field of intelligent operation and maintenance of the private network in the 5G industry and improves the efficiency and the precision of intelligent operation and maintenance of the private network in the 5G industry.
Description
Technical Field
The application relates to the field of fusion of large models and communication, in particular to a 5G private network operation and maintenance method based on the self-adaptive capacity of a large model and a related device.
Background
With the acceleration of industry digital transformation, the operation and maintenance requirements of an industry 5G (5 th Generation Mobile Communication Technology, fifth generation mobile communication technology) private network in the communication field are increasingly complex, and intelligent operation and maintenance becomes one of key factors for keeping competitiveness in the digital era.
However, with rapid development of services and continuous update of technologies, it has been difficult for conventional operation and maintenance methods to meet the operation and maintenance requirements of communication devices. The large model technology breaks through the intelligent operation and maintenance field, can provide a more humanized man-machine interaction mode, can process massive formatted data, provides high-precision analysis and prediction, and provides technology energization for intelligent operation and maintenance. Compared with the traditional AIOps (ARTIFICIAL INTELLIGENCE for IT Operations), the large model can give further capability to intelligent operation and maintenance, such as simpler interaction, more comprehensive knowledge coverage, more flexible model architecture and the like, has lower use threshold and realizes continuous generalization of operation and maintenance capability. Meanwhile, after the 5G network is deployed by industry users, professional knowledge support and personnel storage are needed for service opening and daily operation and maintenance, so that network cost and investment are increased, and a large model is applied to the field of 5G operation and maintenance, so that intelligent network operation can be realized, and the industry investment can be greatly reduced.
Although the large model has certain application in the intelligent operation and maintenance field of the 5G private network at present, serious defects still exist in terms of how to improve the self-adaptive capacity of the large model in the intelligent operation and maintenance direction of the 5G private network so as to meet the intelligent operation and maintenance requirements in the intelligent operation and maintenance field of the 5G private network.
Disclosure of Invention
In view of the above, the application provides a 5G private network operation and maintenance method and a related device based on the self-adaptive capacity of a large model, which are used for improving the self-adaptive capacity of the large model in the field of intelligent operation and maintenance of the private network in the 5G industry so as to further improve the efficiency and the accuracy of intelligent operation and maintenance of the private network in the 5G industry.
The specific technical scheme is as follows:
A5G private network operation and maintenance method based on large model self-adaption capability comprises the following steps:
acquiring a network operation and maintenance request for the 5G private network, wherein the network operation and maintenance request comprises a network operation and maintenance problem to be solved by the 5G private network;
Searching a network operation and maintenance knowledge base based on the network operation and maintenance problem to obtain context information corresponding to the network operation and maintenance problem, wherein the network operation and maintenance knowledge base is a knowledge base obtained by carrying out structural processing on original operation and maintenance knowledge data of a 5G private network;
Searching application perception data corresponding to the network operation and maintenance problems by utilizing a pre-constructed network operation and maintenance agent, and constructing a prompt word for a large model based on the application perception data;
inputting the network operation and maintenance problem, the context information and the prompt word into a large model to generate a solution and an operation and maintenance strategy for the network operation and maintenance problem based on the context information and the prompt word by using the large model.
Optionally, the obtaining the network operation and maintenance request for the 5G private network includes at least one of:
acquiring a network operation and maintenance request of an operation and maintenance person for a 5G private network initiated by a natural language interaction mode;
Acquiring a network operation and maintenance request which is initiated by the network operation and maintenance agent based on the perceived abnormal event of the service of the 5G private network and is matched with the abnormal event of the service;
The network operation and maintenance agent senses the abnormal business event of the 5G private network by dynamically modeling the relation between the business quality and the user experience of the 5G private network by utilizing a genetic algorithm.
Optionally, the process of the network operation and maintenance agent for sensing the abnormal business event of the 5G private network includes:
Acquiring service quality index data corresponding to different types of end-to-end service performance of the 5G service respectively;
based on the association relation between network service quality obtained by dynamic modeling of the relation between service quality of the 5G private network and user experience by using a genetic algorithm and the perception of network performance by a user and service quality index data corresponding to the different types of end-to-end service performance, determining the performance perception degree of the user for the different types of end-to-end service performance as the sub-item perception degree corresponding to the different types of end-to-end service performance;
Determining the overall perceptibility of the network performance of the 5G service by the user according to the sub-item perceptibility respectively corresponding to the different types of end-to-end service performance;
And determining whether a service abnormal event of the 5G service occurs or not based on the overall perceptibility of the network performance of the 5G service by the user and a network performance perception threshold obtained by a big data learning mode.
Optionally, before retrieving the network operation and maintenance knowledge base based on the network operation and maintenance problem, the method further includes:
performing at least one of resolution sub-queries, reference resolution and query rewrite on the network operation and maintenance problem;
The decomposing sub-query is used for splitting the network operation and maintenance problem into a plurality of information independent sub-problems; the query rewrite is used for rewriting the network operation and maintenance problem into a plurality of rewrite problems by optimizing and adjusting the language level of the network operation and maintenance problem.
Optionally, the searching the network operation and maintenance knowledge base based on the network operation and maintenance problem to obtain the context information corresponding to the network operation and maintenance problem includes at least one of the following:
Searching the network operation and maintenance knowledge base based on a plurality of sub-problems of the network operation and maintenance problem to obtain sub-search results corresponding to the sub-problems respectively;
Based on different rewrite problems of the network operation and maintenance problems, respectively searching different network operation and maintenance knowledge bases through different search modes to obtain search results respectively corresponding to the rewrite problems, and combining the search results respectively corresponding to the rewrite problems to obtain context information corresponding to the network operation and maintenance problems.
Optionally, the different searching modes include some or all of database query, vector search, question and answer search, knowledge graph search, plug-in search and keyword search.
Optionally, the retrieving, by using a pre-built network operation and maintenance agent, application awareness data corresponding to the network operation and maintenance problem includes:
And searching a user perception evaluation result, service quality data and network performance data corresponding to the network operation and maintenance problem by using a network operation and maintenance agent, wherein the application perception data comprises the user perception evaluation result, the service quality data and the network performance data.
A 5G private network operation and maintenance device based on large model adaptation capability, comprising:
The system comprises an acquisition module, a processing module and a processing module, wherein the acquisition module is used for acquiring a network operation and maintenance request for the 5G private network, wherein the network operation and maintenance request comprises a network operation and maintenance problem to be solved by the 5G private network;
The first retrieval module is used for retrieving a network operation and maintenance knowledge base based on the network operation and maintenance problem to obtain context information corresponding to the network operation and maintenance problem, wherein the network operation and maintenance knowledge base is a knowledge base obtained by structuring original operation and maintenance knowledge data of a 5G private network;
the second retrieval module is used for retrieving application perception data corresponding to the network operation and maintenance problems by utilizing a pre-constructed network operation and maintenance agent, and constructing a prompt word for a large model based on the application perception data;
and the operation and maintenance module is used for inputting the network operation and maintenance problem, the context information and the prompt word into a large model so as to generate a solution and an operation and maintenance strategy for the network operation and maintenance problem based on the context information and the prompt word by using the large model.
An electronic device, comprising:
A memory for storing a computer program;
A processor for implementing the 5G private network operation and maintenance method based on large model adaptation capability as described in any one of the above by calling and executing the computer program in the memory.
A computer readable medium having stored thereon a computer program which, when executed by a processor, is operable to implement a large model adaptation capability based 5G private network operation and maintenance method as claimed in any one of the preceding claims.
According to the scheme, the 5G private network operation and maintenance method and the related device based on the large model self-adaption capability provided by the application are used for constructing the structured network operation and maintenance knowledge base of the 5G private network by carrying out structuring processing on the original operation and maintenance knowledge data of the 5G private network in advance, and constructing the network operation and maintenance agent of the 5G private network in advance, on the basis, the network operation and maintenance knowledge base is searched for the network operation and maintenance request of the 5G private network by carrying out searching on the network operation and maintenance knowledge base based on the network operation and maintenance problem in the request, so that the context information corresponding to the network operation and maintenance problem is obtained, the application perception data corresponding to the network operation and maintenance problem is searched by utilizing the network operation and maintenance agent, and the prompting word for the large model is constructed based on the searched application perception data, so that the large model can be provided with the abundant and high-quality context information and prompting word of the network operation and maintenance problem, the large model is convenient to take the provided context information and prompting word as references, the corresponding solution and operation strategy are generated for the network operation and maintenance problem, thereby, the self-adaption accuracy of the intelligent model 5G private network can be improved, and the self-adaption accuracy of the intelligent network and the service and the intelligent network in the field of the 5G private network can be improved.
Drawings
In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings that are required to be used in the embodiments or the description of the prior art will be briefly described below, and it is obvious that the drawings in the following description are only embodiments of the present application, and that other drawings can be obtained according to the provided drawings without inventive effort for a person skilled in the art.
FIG. 1 is a 5G private network intelligent operation and maintenance framework diagram based on a large model and related intelligent agents;
FIG. 2 is a flow chart of a method for operating and maintaining a 5G private network based on large model self-adaption capability;
FIG. 3 is a schematic diagram of a business perception evaluation variable structure model of a network operation and maintenance agent provided by the application;
FIG. 4 is a schematic diagram of the search enhancement of the network operation and maintenance agent provided by the application;
fig. 5 is a component structure diagram of the 5G private network operation and maintenance device based on the large model adaptive capacity provided by the application;
Fig. 6 is a component configuration diagram of an electronic device provided by the present application.
Detailed Description
The following description of the embodiments of the present application will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present application, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the application without making any inventive effort, are intended to be within the scope of the application.
The embodiment of the application provides a 5G private network operation and maintenance method based on large model self-adaption capability and a related device, which mainly realize automatic intelligent operation and maintenance of a 5G private network by constructing an intelligent body operation and maintenance framework of the 5G private network by utilizing a large model so as to improve the self-adaption capability of the large model in the field of intelligent operation and maintenance of the private network in the 5G industry and improve the efficiency and precision of intelligent operation and maintenance of the private network in the 5G industry.
The large model and the related intelligent agent adopted in the embodiment of the application are used for the operation and maintenance of the 5G private network, and the main purpose of the application is to express the knowledge related to the 5G private network in a computer-understandable form, and enable the computer to understand, reason and apply the knowledge to realize the automation and the intelligent operation and maintenance of the private network in the 5G industry.
Optionally, in the embodiment of the application, the 5G private network operation and maintenance based on the large model and the related intelligent Agent relates to 5G private network professional knowledge representation, knowledge acquisition, knowledge reasoning, knowledge storage, knowledge management and building agents of the 5G industry private network intelligent operation and maintenance to realize the framework structures related to the connection operation and maintenance requirements, LLM (large language model), the 5G industry private network production environment and the like.
Referring to the large model and related agent-based 5G private network intelligent operation and maintenance framework diagram shown in fig. 1, the intelligent operation and maintenance of the 5G private network mainly comprises knowledge retrieval and large model answer generation (corresponding solution and operation and maintenance strategy generation for network operation and maintenance requests) flow executed based on the improved retrieval enhancement generation (RETRIEVAL-Augmented Generation) technology. Before intelligent operation and maintenance of the 5G private network is performed based on the 5G private network intelligent operation and maintenance framework shown in fig. 1, operation and maintenance knowledge construction of the 5G private network needs to be completed in advance.
The operation and maintenance knowledge construction of the 5G private network refers to constructing a structured network operation and maintenance knowledge base for the 5G private network by carrying out structuring treatment on original operation and maintenance knowledge data of the 5G private network. The original operation and maintenance knowledge data of the 5G private network can include, but is not limited to, an internal database, a knowledge graph, unstructured documents of the 5G private network, related operation and maintenance data corresponding to internal and external system plug-ins, and original data such as external operation and maintenance means, question and answer knowledge and the like. Referring to FIG. 1, the structured network operations knowledge base may be implemented, but is not limited to, as a structured knowledge graph, database/table, question-answer knowledge data set, or the like in a variety of structured data formats.
The network operation and maintenance knowledge base construction process of the 5G private network mainly comprises the steps of knowledge data structuring, index establishment and knowledge storage, wherein the steps are mainly used for carrying out unified processing on an original internal database, a knowledge graph, unstructured documents, external operation and maintenance means, question and answer knowledge and the like of the 5G private network, and storing the unified processing and the unified processing into a unified form of the 5G private network operation and maintenance knowledge base so as to finish the processing and construction of the 5G private network intelligent operation and maintenance knowledge data.
Knowledge construction of a 5G private network intelligent operation and maintenance framework based on a large model and an intelligent agent is responsible for converting original knowledge data of the 5G private network into structured knowledge which is easy to store and retrieve and use, and storing the structured knowledge into a knowledge base for management. The private network in the 5G industry has rich and varied knowledge sources including, but not limited to, unstructured documents, knowledge maps, databases, original operation and maintenance data corresponding to internal and external system plug-ins, and the like. Optionally, when a structured network operation and maintenance knowledge base of the 5G private network is constructed through structuring, the embodiment of the present application firstly adopts chapter and title hierarchical structure information of document knowledge itself, uses a segmentation algorithm for segmentation according to chapters and titles, synthesizes title content and text content to slice original operation and maintenance knowledge data, and adds chapter and/or title information of the slice at the head of each slice content, so that the slice content can better maintain semantic information in the original document.
On the basis, unified structuring treatment is further carried out on each slice, a structured network operation and maintenance knowledge base of the 5G private network is taken as a knowledge graph form as an example, optionally, for each slice, LLM (logical level modeling) can be used for carrying out entity recognition and relation extraction on slice contents through preset prompt words (prompt) to construct related structured knowledge of the 5G industry private network operation and maintenance knowledge graph, structured data in the 5G industry private network knowledge graph is stored in a form of triples, and each triplet comprises an entity, an attribute and an attribute value and corresponds to a corresponding triplet index.
Specific implementations of the triplet index may include, but are not limited to, inverted index , prefix tree, hash table , and the like. The inverted index is to record the positions or identifiers of all triples containing the index item by taking each subject, predicate and object as index items, so that relevant triples can be quickly found during query, the prefix tree is to organize the triples of data by using the prefix tree for specific application scenes, particularly when the predicate or object has certain regularity, the query efficiency can be remarkably improved, and the hash table is to map each part of the triples to specific storage positions through a hash function, so that quick search and update operations are realized.
And then, directly inquiring the knowledge graph of the intelligent operation and maintenance of the 5G private network by using a triplet index inquiry knowledge graph, and inquiring the triplets based on the network operation and maintenance problems carried in the network operation and maintenance request by the system when the network operation and maintenance request of the 5G private network is initiated so as to find out information related to the problems, and taking the information as the context of the problems to assist a large model in better understanding the problems so as to accurately and efficiently generate a corresponding solution and operation and maintenance strategy for the problems.
In practical application, a related structured operation and maintenance knowledge base such as a structured database, a table and/or a knowledge graph can be built for a large-model-based intelligent operation and maintenance framework of the 5G private network through a relational database theory, a data warehouse theory and the like, and for the case that the original operation and maintenance knowledge data of the 5G private network is in a structured form, the repeated construction of the knowledge data in the intelligent operation and maintenance knowledge management system of the 5G private network is not needed, and only the knowledge data is needed to be incorporated into the knowledge management system so that the system can use the existing operation and maintenance knowledge.
On the basis, the intelligent operation and maintenance of the 5G private network can be realized by executing corresponding knowledge retrieval and a large model answer generation flow based on the knowledge retrieval through the 5G private network operation and maintenance method based on the large model self-adaption capability provided by the embodiment of the application.
Referring to the method flow chart shown in fig. 2, the method for operating and maintaining a 5G private network based on large model adaptation capability according to the embodiment of the present application may at least include the following steps 201 to 204, which are respectively described in detail below.
Step 201, obtaining a network operation and maintenance request for the 5G private network, wherein the network operation and maintenance request comprises a network operation and maintenance problem to be solved by the 5G private network.
The network operation and maintenance agent can specifically acquire a network operation and maintenance request for the 5G private network initiated by operation and maintenance personnel through a natural language interaction mode, or acquire a network operation and maintenance request matched with a business abnormal event initiated by the network operation and maintenance agent based on the business abnormal event of the perceived 5G private network. The network operation and maintenance agent senses the abnormal business event of the 5G private network by dynamically modeling the relationship between the business quality and the user experience of the 5G private network by utilizing a genetic algorithm.
The embodiment of the application constructs the network operation and maintenance agent of the 5G private network in advance, and the constructed network operation and maintenance agent supports the perception of the 5G private network service and actively initiates the network operation and maintenance request of the 5G private network when the 5G private network service is perceived to be abnormal in a body.
The embodiment of the application utilizes a genetic algorithm to dynamically model the KQI (quality of service) and QoE (user experience) relationship of the 5G private network, so as to realize the intelligent perception capability of the 5G private network operation and maintenance agent on the 5G private network service. By utilizing a genetic algorithm to dynamically model the KQI and QoE relations of the 5G private network, a potential association relation between the network service quality of the 5G private network and the perception of the network performance by a user is found, the user perception evaluation is carried out based on the association relation, and when the network performance perceived by the user is not up to a threshold or the service index is deteriorated, an operation and maintenance request is adaptively initiated to solve the corresponding network problem, so that the real intelligent operation and maintenance goal of the 5G private network is realized.
User perception evaluation of the network operation and maintenance agent is performed for the 5G private network service. For any service of the 5G private network, the service can be divided into multiple classes of performance according to the user experience, such as various service performances/network performances of the service, such as network jitter, time delay, rate, occupancy rate, stability rate, etc., where one class of performance corresponds to one or a group of KQI indexes (service quality indexes), which may also be referred to as an end-to-end performance index of the service.
The end-to-end performance index of the service will influence the perception of the corresponding category performance by the user, and the perception of each category performance of the service by the user will determine the overall perception of the network performance of the service by the user, i.e. the overall perception of the service by the user is comprehensively influenced by the perception of the network performance of each category by the user.
Based on the technical thought, a 5G private network service perception evaluation variable structure model of the network operation and maintenance agent is constructed, and the influence relation of variables in the private network service perception evaluation variable structure model is from bottom to top referring to FIG. 3,5G, and the method is specifically characterized in that one class of end-to-end service performance corresponds to one sub-item perception index and one group of KQI indexes. The quality of the KQI index directly influences the quality of the perception of the corresponding sub-item. The quality of the sub-item perceptibility directly influences the quality of the overall perceptibility. Therefore, from the perspective of user perception evaluation, the input variables of the network operation and maintenance agent model comprise KQI indexes of each class to be evaluated, and the output variables comprise the perception degree and the overall perception degree of each sub-item of the service.
The process of the network operation and maintenance agent sensing the abnormal event of the 5G private network can be realized by obtaining service quality index data corresponding to different types of end-to-end service performances of the 5G private network respectively, determining the performance perceptibility of the user to the different types of end-to-end service performances as the sub-item perceptibility corresponding to the different types of end-to-end service performances respectively based on the association relationship between the network service quality obtained by dynamic modeling of the relationship between the service quality of the 5G private network and the user experience by using a genetic algorithm and the perception of the network performance by the user, and determining whether the abnormal event of the 5G service occurs or not based on the overall perceptibility of the user to the network performance of the 5G service and the network performance perception threshold obtained by a big data learning mode.
Optionally, the network operation and maintenance agent of the 5G private network can perform intelligent analysis on big data such as a whole scene telephone traffic model, an intelligent operation and maintenance effect, a KPI trend and the like, and continuously iterate and optimize inflection points of user service perception and network performance on line in a reinforced self-learning mode, so as to obtain a service perception threshold/network performance perception threshold of a user on the 5G private network service, so as to be used for identifying and judging abnormal events of the service.
And 202, searching a network operation and maintenance knowledge base based on the network operation and maintenance problem to obtain context information corresponding to the network operation and maintenance problem, wherein the network operation and maintenance knowledge base is a knowledge base obtained by structuring original operation and maintenance knowledge data of a 5G private network.
In order to improve the accuracy of the RAG, the embodiment of the present application adopts a series of preprocessing techniques and methods to preprocess the network operation and maintenance problem in the network operation and maintenance request, so that before the network operation and maintenance knowledge base is searched based on the network operation and maintenance problem, the embodiment of the present application also executes corresponding preprocessing on the network operation and maintenance problem, and the executed preprocessing may include, but is not limited to, decomposing part or all of sub-queries, referring to digestion and query rewrite.
In the knowledge retrieval of the private network intelligent operation and maintenance framework in the 5G industry, a series of preprocessing technology and method are adopted to preprocess the network operation and maintenance problem, and the main purpose is to optimize the network operation and maintenance problem (namely to optimize the query) so as to improve the retrieval efficiency and simultaneously improve the retrieval accuracy.
The method comprises the steps of decomposing sub-queries, wherein the sub-queries are used for splitting a network operation and maintenance problem into a plurality of information independent sub-problems; the query rewrite is used for rewriting the network operation and maintenance problem into a plurality of rewrite problems by optimizing and adjusting the language level of the network operation and maintenance problem.
In particular, decomposing a sub-query may split a complex network operation problem into several smaller, more tractable parts, where each part represents an information independent sub-problem. To achieve this goal, optionally, a multi-query retriever may be employed that automatically generates multiple sub-questions for a given network operation-dimension question from multiple dimensions, which may include, but are not limited to, different dimensions of time dimension, space dimension, and the like, by means of LLM. For a network operation problem in a network operation request, the network operation problem can be regarded as a query (total query), and for each generated sub-problem, the network operation problem can be regarded as a sub-query of the network operation problem, so that the network operation knowledge base can be searched based on each sub-query.
Alternatively, the reference resolution can be implemented in Few-shot Prompt (small sample hint) mode in combination with thinking-action-observation strategy, which enables LLM to analyze and process more complex reference resolution problems by integrating some common data samples of reference resolution scene as Few-shot examples into LLM's Prompt, in combination with CoT (Chain of Thought, thinking chain) method.
The query rewrite is aimed at enhancing the retrieval efficiency and improving the accuracy of the retrieval result by optimizing and adjusting the language level of the network operation and maintenance problem, and can grasp the information requirement of the user more accurately and return more relevant retrieval result by the query rewrite system.
In the implementation, the strong capability of the LLM can be selected, and the LLM can be stimulated to effectively rewrite the network operation and maintenance problem by using high-quality corpus and prompt words. In order to further improve the query rewrite effect, a rewrite component of the service can be introduced, and the rewrite component is used as an auxiliary component to be specially responsible for adjusting and rewriting the network operation and maintenance problems in the network operation and maintenance request, so that the rewrite component is better suitable for the processing requirements of a fixed retriever and LLM.
On the basis of completing the preprocessing of the network operation and maintenance problem, the network operation and maintenance knowledge base can be further searched based on the network operation and maintenance problem, and the process of searching the network operation and maintenance knowledge base based on the network operation and maintenance problem can be realized to comprise at least one of the following steps:
11 Searching the network operation and maintenance knowledge base based on a plurality of sub-problems of the network operation and maintenance problems to obtain sub-search results corresponding to the sub-problems respectively, and merging the sub-search results corresponding to the sub-problems to obtain context information corresponding to the network operation and maintenance problems.
Aiming at each sub-problem of the network operation and maintenance problem, the sub-problem can be regarded as a sub-query, a multi-query retriever is adopted to retrieve a group of knowledge data related to the sub-query from a structured network operation and maintenance knowledge base, and finally, a union operation is adopted to documents retrieved based on all sub-queries, so that a broader knowledge data set with potential correlation with the network operation and maintenance problem is constructed, and the knowledge data set is used as context information of the network operation and maintenance problem to assist a follow-up large model such as LLM to accurately understand and answer the network operation and maintenance problem.
The knowledge retrieval is carried out based on the decomposition sub-query mode, so that certain limitations of the vector distance retrieval method can be broken through, and a group of richer and diversified retrieval results can be obtained.
12 Based on the different rewrite problems of the network operation and maintenance problems, respectively searching different network operation and maintenance knowledge bases through different search modes to obtain search results respectively corresponding to the rewrite problems, and combining the search results respectively corresponding to the rewrite problems to obtain the context information corresponding to the network operation and maintenance problems.
The different search modes may include, but are not limited to, some or all of database queries, vector searches, question and answer searches (QA searches), knowledge-graph searches, plug-in searches, and keyword searches.
After the network operation and maintenance problem is rewritten into a plurality of rewrite problems (i.e., a plurality of queries) through the query rewrite process, the embodiment adopts a hybrid search strategy to search different network operation and maintenance knowledge bases through different search modes based on different rewrite problems of the network operation and maintenance problem.
In implementation, optionally, after the network operation and maintenance problem is rewritten into a plurality of rewrite problems, the different rewrite problems can be distributed to different search method flows through the query routing module, and the search method can comprise database query, vector search, QA search, knowledge graph search, plug-in search, keyword search and the like. Each searching method outputs TopK optimal searching results after searching by the multiple searching methods. And finally, the final search results corresponding to the network operation and maintenance problems can be obtained by combining the search results, and the final search results are used as the context information of the network operation and maintenance problems.
Because the scoring and sorting criteria of the TopK search results respectively generated based on different search methods are different, when the TopK search results respectively generated based on different search methods are combined, they cannot be simply combined together for sorting, and at this time, a new rearrangement algorithm can be introduced to combine and reorder the TopK results, thereby obtaining a final TopK search result, and discarding other search results. These final selected TopK search results are submitted as context information of the network operation and maintenance problem to a large model such as LLM together with the network operation and maintenance problem, so that the large model generates more accurate solutions and operation and maintenance strategies for the network operation and maintenance problem based on the provided context information content.
In combination with the enhanced search schematic diagram of the 5G private network operation and maintenance agent shown in fig. 4, the related processing performed on the search results corresponding to each sub-problem or rewrite problem of the network operation and maintenance problem can be regarded as post-processing of the search, and the post-processing stage is mainly responsible for further optimizing and adjusting the search results of each sub-problem or rewrite problem so as to improve the performance of the search system and the quality of the search results. The search results of each sub-problem or rewrite problem are mainly screened, compressed, reordered, etc., and these operations are performed to refine and sort out a set of final search results for network operation and maintenance problems. These final search results are then submitted to a large model to assist in understanding the network operation and maintenance questions and more accurate answer generation.
According to the embodiment of the application, the required knowledge retrieval is carried out on the network operation and maintenance problems by adopting various pre-and-post processing technologies and adopting mixed knowledge retrieval, so that the accuracy of the RAG can be effectively improved, and the accuracy of answers generated by a final large model for the network operation and maintenance problems can be further improved.
And 203, searching application perception data corresponding to the network operation and maintenance problem by utilizing a pre-constructed network operation and maintenance agent, and constructing a prompt word for a large model based on the application perception data.
The network operation and maintenance agent can be used for searching the user perception evaluation result, the service quality data and the network performance data corresponding to the network operation and maintenance problem. And the retrieved user perception evaluation result, service quality data and network performance data related to the network operation and maintenance problem are used as application perception data corresponding to the network operation and maintenance problem, and prompting words for the large model are constructed based on the application perception data, so that the large model is prompted to mine and analyze the root cause of the network operation and maintenance problem from the aspects of user perception, service quality, network performance and the like, and further accurate generation of answers is rapidly and efficiently carried out for the network operation and maintenance problem.
Step 204, inputting the network operation and maintenance problem, the context information and the prompt word into a large model, so as to generate a solution and an operation and maintenance strategy for the network operation and maintenance problem based on the context information and the prompt word by using the large model.
Finally, the network operation and maintenance problem, the context information and the prompt word can be input into a large model together, for example, the large model is input into LLM, the large model is combined with the input context information and the prompt word to understand the input network operation and maintenance problem, and the root cause of the problem is analyzed, so that a corresponding answer is generated for the network operation and maintenance problem.
Wherein the generated answer includes a solution to the network operation and maintenance problem and an operation and maintenance policy.
Optionally, the solution is a set of high-level solution of decision level which cannot be directly understood and executed by the device, and the operation and maintenance policy is responsible for landing the high-level decisions into specific configuration rules or commands and other operation and maintenance operation information which can be directly understood and executed by the device, so as to realize intelligent operation and maintenance of the 5G private network and automatically solve the network operation and maintenance problem indicated by the initiated network operation and maintenance request.
According to the scheme, the 5G private network operation and maintenance method based on the large model self-adaption capability provided by the application constructs a structured network operation and maintenance knowledge base of the 5G private network by carrying out structuring treatment on the original operation and maintenance knowledge data of the 5G private network in advance, constructs a network operation and maintenance intelligent body of the 5G private network in advance, on the basis, searches the network operation and maintenance knowledge base based on the network operation and maintenance problem in the request to obtain the context information corresponding to the network operation and maintenance problem, and the network operation and maintenance agent is utilized to search application perception data corresponding to the network operation and maintenance problem, and prompt words for the large model are constructed based on the searched application perception data, so that the large model can be provided with rich and high-quality context information and prompt words of the network operation and maintenance problem, the large model can conveniently and efficiently and accurately generate corresponding solutions and operation and maintenance strategies for the network operation and maintenance problem by taking the provided context information and prompt words as references, and the self-adaptive capacity of the large model in the intelligent operation and maintenance field of the private network in the 5G industry is improved, and the intelligent operation and maintenance efficiency and precision of the private network in the 5G industry are improved.
Besides, the application also embeds the 5G private network operation and maintenance agent obtained by dynamic self-adaptive modeling of the KQI-QoE relation based on a large model framework, and can mine the association relation between the service quality and the network performance perception of the user based on the agent, and when the user perception corresponding performance index does not reach the threshold or the service index is deteriorated, the agent can self-adaptively initiate the operation and maintenance request to solve the corresponding network operation and maintenance problem, thereby realizing the real objective of the intelligent operation and maintenance of the 5G private network.
Corresponding to the above method, the embodiment of the present application further provides a 5G private network operation and maintenance device based on the large model adaptive capability, referring to a schematic composition structure of fig. 5, where the device includes:
an obtaining module 501, configured to obtain a network operation and maintenance request for a 5G private network, where the network operation and maintenance request includes a network operation and maintenance problem to be solved by the 5G private network;
The first retrieval module 502 is configured to retrieve a network operation and maintenance knowledge base based on the network operation and maintenance problem to obtain context information corresponding to the network operation and maintenance problem, where the network operation and maintenance knowledge base is a knowledge base obtained by performing structural processing on original operation and maintenance knowledge data of a 5G private network;
a second retrieving module 503, configured to retrieve application awareness data corresponding to the network operation and maintenance problem by using a pre-constructed network operation and maintenance agent, and construct a prompt word for a large model based on the application awareness data;
an operation and maintenance module 504, configured to input the network operation and maintenance problem, the context information and the prompt word into a large model, so as to generate a solution and an operation and maintenance policy for the network operation and maintenance problem based on the context information and the prompt word by using the large model.
In an alternative embodiment, the obtaining module 501 is specifically configured to:
acquiring a network operation and maintenance request of an operation and maintenance person for a 5G private network initiated by a natural language interaction mode;
Acquiring a network operation and maintenance request which is initiated by the network operation and maintenance agent based on the perceived abnormal event of the service of the 5G private network and is matched with the abnormal event of the service;
The network operation and maintenance agent senses the abnormal business event of the 5G private network by dynamically modeling the relation between the business quality and the user experience of the 5G private network by utilizing a genetic algorithm.
In an alternative embodiment, the process of the network operation and maintenance agent for sensing the abnormal business event of the 5G private network includes:
Acquiring service quality index data corresponding to different types of end-to-end service performance of the 5G service respectively;
Determining the performance perceptibility of the user for the different types of end-to-end service performance based on the association relationship between the network service quality obtained by dynamically modeling the relationship between the service quality of the 5G private network and the user experience by using a genetic algorithm and the perception of the network performance by the user and the service quality index data corresponding to the different types of end-to-end service performance respectively, and taking the determined performance perceptibility of the user for the different types of end-to-end service performance as the sub-item perceptibility corresponding to the different types of end-to-end service performance respectively;
Determining the overall perceptibility of the network performance of the 5G service by the user according to the sub-item perceptibility respectively corresponding to the different types of end-to-end service performance;
And determining whether a service abnormal event of the 5G service occurs or not based on the overall perceptibility of the network performance of the 5G service by the user and a network performance perception threshold obtained by a big data learning mode.
In an alternative embodiment, the apparatus further comprises a preprocessing module for performing at least one of decomposing sub-queries, referring to digestion, and query rewrite on the network operation and maintenance questions prior to retrieving a network operation and maintenance knowledge base based on the network operation and maintenance questions;
The decomposing sub-query is used for splitting the network operation and maintenance problem into a plurality of information independent sub-problems; the query rewrite is used for rewriting the network operation and maintenance problem into a plurality of rewrite problems by optimizing and adjusting the language level of the network operation and maintenance problem.
In an alternative embodiment, the first search module 502 is specifically configured to:
Searching the network operation and maintenance knowledge base based on a plurality of sub-problems of the network operation and maintenance problem to obtain sub-search results corresponding to the sub-problems respectively;
Based on different rewrite problems of the network operation and maintenance problems, respectively searching different network operation and maintenance knowledge bases through different search modes to obtain search results respectively corresponding to the rewrite problems, and combining the search results respectively corresponding to the rewrite problems to obtain context information corresponding to the network operation and maintenance problems.
In an alternative embodiment, the different search modes include some or all of database query, vector search, question and answer search, knowledge graph search, plug-in search and keyword search.
In an alternative embodiment, the second retrieving module 503 is specifically configured to:
And searching a user perception evaluation result, service quality data and network performance data corresponding to the network operation and maintenance problem by using a network operation and maintenance agent, wherein the application perception data comprises the user perception evaluation result, the service quality data and the network performance data.
For the 5G private network operation and maintenance device based on the large model self-adaptation capability disclosed in the embodiment of the present application, since the device corresponds to the 5G private network operation and maintenance method based on the large model self-adaptation capability disclosed in the above method embodiment, the description is simpler, and the relevant similarities are only required to refer to the description of the above method embodiments, and are not described in detail herein.
The embodiment of the application also discloses an electronic device, and the composition structure of the electronic device, as shown in fig. 6, at least comprises:
Memory 10 for storing a computer program.
And the processor 20 is used for realizing the 5G private network operation and maintenance method based on the large model self-adaption capability and provided by the embodiment of the method Wen Ren by calling and executing the computer program in the memory.
The processor 20 may be a central processing unit (Central Processing Unit, CPU), application-specific integrated circuit (ASIC), digital Signal Processor (DSP), application-specific integrated circuit (ASIC), field-programmable gate array (FPGA), neural Network Processor (NPU), deep learning processor (DPU), or other programmable logic device, etc.
In addition, the electronic device may include communication interfaces, communication buses, and the like. The memory, processor and communication interface communicate with each other via a communication bus.
The communication interface is used for communication between the electronic device and other devices. The communication bus may be a peripheral component interconnect standard (PERIPHERAL COMPONENT INTERCONNECT, PCI) bus, or an extended industry standard architecture (Extended Industry Standard Architecture, EISA) bus, etc., and may be classified as an address bus, a data bus, a control bus, etc.
In addition, the present application also provides a computer readable medium having stored thereon a computer program comprising program code for performing a 5G private network operation and maintenance method based on large model adaptation capability as disclosed in any of the method embodiments above. Accordingly, the computer program, when executed by a processor, is operable to implement a 5G private network operation and maintenance method based on large model adaptation capability as disclosed in any of the method embodiments above.
In the context of the present application, a computer-readable medium (machine-readable medium) can be a tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. The machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
It should be noted that, in the present specification, each embodiment is described in a progressive manner, and each embodiment is mainly described as different from other embodiments, and identical and similar parts between the embodiments are all enough to be referred to each other.
For convenience of description, the above system or apparatus is described as being functionally divided into various modules or units, respectively. Of course, the functions of each element may be implemented in the same piece or pieces of software and/or hardware when implementing the present application.
From the above description of embodiments, it will be apparent to those skilled in the art that the present application may be implemented in software plus a necessary general hardware platform. Based on such understanding, the technical solution of the present application may be embodied essentially or in a part contributing to the prior art in the form of a software product, which may be stored in a storage medium, such as a ROM/RAM, a magnetic disk, an optical disk, etc., including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method described in the embodiments or some parts of the embodiments of the present application.
Finally, it is further noted that relational terms such as first, second, third, and the like, if any, are used solely to distinguish one from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises an element.
The foregoing is merely a preferred embodiment of the present application and it should be noted that modifications and adaptations to those skilled in the art may be made without departing from the principles of the present application, which are intended to be comprehended within the scope of the present application.
Claims (9)
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202510729968.6A CN120264329B (en) | 2025-06-03 | 2025-06-03 | 5G private network operation and maintenance method and related device based on large model adaptive capability |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202510729968.6A CN120264329B (en) | 2025-06-03 | 2025-06-03 | 5G private network operation and maintenance method and related device based on large model adaptive capability |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN120264329A CN120264329A (en) | 2025-07-04 |
| CN120264329B true CN120264329B (en) | 2025-09-16 |
Family
ID=96183713
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202510729968.6A Active CN120264329B (en) | 2025-06-03 | 2025-06-03 | 5G private network operation and maintenance method and related device based on large model adaptive capability |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN120264329B (en) |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN119967457A (en) * | 2024-12-26 | 2025-05-09 | 天翼物联科技有限公司 | Signaling analysis method, system, electronic device and storage medium |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113038508A (en) * | 2021-02-26 | 2021-06-25 | 浙江方大通信有限公司 | 5G communication network operation and maintenance system and method based on data perception fusion evaluation |
| CN118945047A (en) * | 2023-05-10 | 2024-11-12 | 华为技术有限公司 | Network operation and maintenance method, device, equipment and storage medium based on large language model |
| CN118916453A (en) * | 2024-07-09 | 2024-11-08 | 平安健康保险股份有限公司 | Intelligent operation and maintenance method based on self-developed GPT model and related equipment thereof |
| CN118916618A (en) * | 2024-10-10 | 2024-11-08 | 卓望信息技术（北京）有限公司 | Network operation and maintenance analysis method, device and storage medium |
| CN119835668A (en) * | 2024-12-26 | 2025-04-15 | 中国联合网络通信集团有限公司 | Wireless network quality difference processing method, device and medium |
-
2025
- 2025-06-03 CN CN202510729968.6A patent/CN120264329B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN119967457A (en) * | 2024-12-26 | 2025-05-09 | 天翼物联科技有限公司 | Signaling analysis method, system, electronic device and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN120264329A (en) | 2025-07-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110377715A (en) | Reasoning type accurate intelligent answering method based on legal knowledge map | |
| US9720944B2 (en) | Method for facet searching and search suggestions | |
| CN112800170A (en) | Question matching method and device, and question answering method and device | |
| CN118839021B (en) | A dynamic relevance enhanced retrieval generation system and method driven by intelligent knowledge graph | |
| CN105045875B (en) | Personalized search and device | |
| CN119476458A (en) | A knowledge management system and method for building a large speech model | |
| CN112084347B (en) | A data retrieval method and system based on knowledge representation learning | |
| US20160171057A1 (en) | Non-exclusionary search within in-memory databases | |
| CN118643134A (en) | Retrieval enhancement generation system and method based on knowledge graph | |
| CN118364087A (en) | Database-based retrieval enhancement and question-answering method and system | |
| CN112597285B (en) | Man-machine interaction method and system based on knowledge graph | |
| CN113254630A (en) | Domain knowledge map recommendation method for global comprehensive observation results | |
| CN109471889A (en) | Report accelerated method, system, computer equipment and storage medium | |
| CN119336900B (en) | Retrieval optimization method based on hierarchical expert routing model and CoT reasoning | |
| CN119597862A (en) | A method and system for enhancing information retrieval based on multi-channel recall mode | |
| CN120264329B (en) | 5G private network operation and maintenance method and related device based on large model adaptive capability | |
| CN109460506B (en) | A User Demand-Driven Resource Matching Push Method | |
| CN116881538A (en) | Short text information retrieval system based on big data application | |
| CN114780700A (en) | Intelligent question-answering method, device, equipment and medium based on machine reading understanding | |
| Chen | English translation template retrieval based on semantic distance ontology knowledge recognition algorithm | |
| CN111897932A (en) | Query processing method and system for text big data | |
| CN120046711B (en) | Logical reasoning driven knowledge graph enhanced large model generation method | |
| CN118070925B (en) | Model training method, device, electronic equipment, storage medium and program product | |
| CN118445300B (en) | Query statement rewriting method, rewriting platform, electronic device and storage medium | |
| CN117971888B (en) | Method, device, equipment, storage medium and program product for determining data engine |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |