Method and device for processing cell with poor wireless network quality and medium

Abstract

The invention provides a wireless network poor-quality cell processing method and device and a medium, and the method comprises the steps: compiling a plurality of diagnosis subtasks of a wireless network poor-quality cell and a chain relation of the diagnosis subtasks when being applied to a network optimization expert Agent, distributing each diagnosis subtask to a corresponding poor-quality expert Agent, enabling each poor-quality expert Agent to obtain a respective diagnosis sub-result according to the corresponding diagnosis sub-task, and sending the diagnosis sub-result to the network optimization expert Agent; the diagnosis sub-results sent by the poor-quality expert Agents are received, the diagnosis result of the poor-quality cell of the wireless network is obtained according to the multiple diagnosis sub-results and the corresponding chain relation, and the diagnosis result comprises poor-quality root causes and poor-quality processing suggestions of the poor-quality cell of the wireless network. According to the invention, a network optimization auxiliary tool with relatively high intelligent degree and relatively high processing speed is provided by adopting multi-agent cooperation.

Description

Technical Field

[0001] The present disclosure at least relates to the field of network technology, and in particular to a method, device and medium for processing poor quality cells in a wireless network. Background Art

[0002] Existing analysis of poor wireless network quality issues mostly uses traditional data analysis experience and manual participation of network optimization staff. This results in insufficient intelligence in network optimization analysis, and the degree of optimization depends on the level of the optimization staff. The network optimization work is inefficient, has poor accuracy, and is difficult to meet large-scale network optimization requirements.

[0003] In particular, the analysis of poor wireless network quality problems may involve the analysis of data in multiple dimensions. How to achieve efficient multi-dimensional data analysis and aggregation based on intelligence is one of the technical problems to be solved in this field. Summary of the invention

[0004] The technical problem to be solved by the present disclosure is to provide a method, device and medium for processing poor-quality cells in a wireless network in view of the above-mentioned deficiencies, so as to solve the problem of how to improve the analysis and optimization efficiency of poor-quality cells through automated and intelligent methods.

[0005] In a first aspect, the present disclosure provides a method for processing a poor quality cell in a wireless network, the method being applied to a network optimization expert Agent and comprising:

[0006] Compile several diagnostic subtasks and their chain relationships for poor quality cells in wireless networks, and distribute each diagnostic subtask to the corresponding poor quality expert Agent, so that each poor quality expert Agent obtains its own diagnostic sub-result according to the corresponding diagnostic subtask, and sends the diagnostic sub-result to the network optimization expert Agent;

[0007] Receive the diagnosis sub-results sent by each poor quality expert Agent, and obtain the diagnosis result of the poor quality cell of the wireless network according to the multiple diagnosis sub-results and the corresponding chain relationship. The diagnosis result includes the poor quality root cause of the poor quality cell of the wireless network and the poor quality treatment suggestion;

[0008] Among them, Agent is an intelligent agent, the network optimization expert Agent is an intelligent agent with expert knowledge of wireless network optimization, and the poor quality expert Agent is an intelligent agent with expert knowledge of wireless network poor quality diagnosis.

[0009] Furthermore, several diagnostic subtasks and their chain relationships of poor quality cells of wireless networks are compiled, and each diagnostic subtask is distributed to the corresponding poor quality expert Agent, including:

[0010] receiving user questions through the first network big model, and identifying the user's intention to inquire about the cell with poor wireless network quality in the question;

[0011] Determine the real-time poor wireless network quality type of the poor wireless network quality cell according to the statistical result of the indicator of the poor wireless network quality cell;

[0012] The first network model uses the thinking chain question and answer prompt words to compile several diagnostic subtasks and their chain relationships of the poor quality cells of the wireless network according to the real-time poor quality type of the wireless network, wherein the several diagnostic subtasks include at least one of: fault diagnosis, parameter diagnosis, coverage diagnosis, quality diagnosis, interference diagnosis, capacity diagnosis, switching diagnosis, and hidden fault diagnosis;

[0013] Distribute each diagnosis subtask to a corresponding quality difference expert agent, the quality difference expert agent includes: at least one of a fault diagnosis agent, a parameter diagnosis agent, a coverage diagnosis agent, a quality diagnosis agent, an interference diagnosis agent, a capacity diagnosis agent, a switching diagnosis agent, and a hidden fault diagnosis agent;

[0014] Among them, the first network large model is a large language model used to compile several diagnostic subtasks and their chain relationships after training based on wireless network optimization problems. The thinking chain question and answer prompts are obtained in advance based on the diagnostic process in the historical wireless network poor quality cell diagnosis work order.

[0015] Further, a diagnosis result of a poor quality cell of a wireless network is obtained according to a plurality of diagnosis sub-results and corresponding chain relationships. The diagnosis result includes a poor quality root cause of the poor quality cell of the wireless network and a poor quality processing suggestion, specifically including:

[0016] Obtaining the thinking chain question and answer prompt words and their order corresponding to a number of diagnostic subtasks and their chain relationships, and inputting a number of diagnostic subresults into the question template of the second network macromodel in order to obtain diagnostic results including root causes of poor quality of wireless network poor quality cells and suggestions for poor quality processing, wherein the diagnostic subresults include abnormal real-time wireless network poor quality cell data and its corresponding historical processing experience information;

[0017] Outputting voice and / or text results of the root cause of poor quality of the wireless network poor quality cell and the poor quality treatment suggestion, displaying the wireless network poor quality cell on a map and marking its poor quality root cause, generating a page instruction for repairing the poor quality root cause according to the poor quality treatment suggestion, and automatically or under user triggering executing the page instruction to repair the wireless network poor quality cell;

[0018] Among them, the second network large model is a large language model used to summarize the root causes of poor quality and poor quality processing suggestions of wireless network poor quality cells after the first network large model is trained. The question templates include questions about obtaining the root causes of poor quality based on a number of abnormal real-time wireless network poor quality cell data and questions about obtaining poor quality processing suggestions based on a number of historical processing experiences.

[0019] Furthermore, the method also includes pre-training a large network model, specifically including:

[0020] Collect historical wireless network quality types, historical wireless network quality cell diagnosis work orders and wireless network optimization expert knowledge. Wireless network optimization expert knowledge includes historical diagnosis subtasks and their chain relationships and historical wireless network quality cell diagnosis results. Historical wireless network quality cell diagnosis results include historical poor root causes of historical wireless network quality cells and historical poor quality treatment solutions.

[0021] Control the first network model, extract the thought chain question and answer prompt words according to the diagnosis process in the historical wireless network poor quality cell diagnosis work order corresponding to the historical wireless network poor quality type, compile the historical diagnosis subtasks and their chain relationships under the thought chain question and answer prompt words, and fine-tune the parameters of the first network model until the historical diagnosis subtasks and their chain relationships are accurately output;

[0022] Distribute the historical diagnosis subtask to the corresponding quality expert Agent, so that the quality expert Agent determines the wireless network API interface that needs to be called according to the historical diagnosis subtask, obtains the historical wireless network quality cell data through the wireless network API interface, determines the existence of abnormal historical wireless network quality cell data and its corresponding historical processing experience information according to the wireless network quality diagnosis expert knowledge, and sends the abnormal historical wireless network quality cell data and its corresponding historical processing experience information to the network optimization expert Agent. The API interface is an application programming interface;

[0023] Receive abnormal historical wireless network poor quality cell data and its corresponding historical processing experience information sent by each poor quality expert Agent, control the second network large model, combine the abnormal historical wireless network poor quality cell data and its corresponding historical processing experience information with the chain relationship and the preset problem template, output the historical wireless network poor quality cell diagnosis result, and fine-tune the parameters of the second network large model until the historical wireless network poor quality cell diagnosis result is accurately output.

[0024] In a second aspect, the present disclosure provides a method for processing a poor quality cell in a wireless network, the method being applied to a poor quality expert Agent and comprising:

[0025] Receiving the diagnosis subtask distributed by the network optimization expert Agent, the diagnosis subtask is a number of diagnosis subtasks of poor quality cells of wireless networks compiled by the network optimization expert Agent and the diagnosis subtasks corresponding to the poor quality expert Agent in the chain relationship;

[0026] Obtaining a diagnosis sub-result according to the corresponding diagnosis sub-task, and sending the diagnosis sub-result to the network optimization expert agent, so that the network optimization expert agent obtains a diagnosis result of a poor quality cell of the wireless network according to a plurality of diagnosis sub-results sent by a plurality of poor quality expert agents and a corresponding chain relationship, wherein the diagnosis result includes a poor quality root cause of the poor quality cell of the wireless network and a poor quality treatment suggestion;

[0027] Among them, Agent is an intelligent agent, the network optimization expert Agent is an intelligent agent with expert knowledge of wireless network optimization, and the poor quality expert Agent is an intelligent agent with expert knowledge of wireless network poor quality diagnosis.

[0028] Furthermore, the diagnostic subtasks distributed by the network optimization expert agent are received, including:

[0029] In response to being one of the fault diagnosis agent, parameter diagnosis agent, coverage diagnosis agent, quality diagnosis agent, interference diagnosis agent, capacity diagnosis agent, switching diagnosis agent, and hidden fault diagnosis agent, receiving the corresponding diagnosis subtask distributed by the network optimization expert agent, the diagnosis subtask corresponding to which is one of the fault diagnosis, parameter diagnosis, coverage diagnosis, quality diagnosis, interference diagnosis, capacity diagnosis, switching diagnosis, and hidden fault diagnosis, and is the first network large model of the network optimization expert agent, and is compiled by calling the thinking chain question and answer prompt words according to the real-time poor wireless network quality type of the poor wireless network quality cell;

[0030] Among them, the first network large model is a large language model used to compile several diagnostic subtasks and their chain relationships after training based on wireless network optimization problems. The thinking chain question and answer prompts are obtained in advance based on the diagnostic process in the historical wireless network poor quality cell diagnosis work order.

[0031] Furthermore, the diagnosis sub-result is obtained according to the corresponding diagnosis sub-task, and the diagnosis sub-result is sent to the network optimization expert Agent, which specifically includes:

[0032] Determine the wireless network API interface that needs to be called according to the corresponding diagnostic subtask, and obtain real-time wireless network poor quality cell data through the corresponding wireless network API interface. The API interface is an application programming interface;

[0033] According to the expert knowledge of wireless network quality diagnosis, it is determined that there are abnormal real-time wireless network quality cell data and its corresponding historical processing experience information, and the abnormal real-time wireless network quality cell data and its corresponding historical processing experience information are used as the diagnosis sub-result;

[0034] Sending the diagnosis sub-results to the network optimization expert agent, so that the network optimization expert agent obtains the thinking chain question and answer prompt words and their sequence corresponding to the plurality of diagnosis sub-tasks and their chain relationships, and inputs the plurality of diagnosis sub-results into the question template of the second network macro model in sequence, so as to obtain the diagnosis results including the root causes of the poor quality of the wireless network poor quality cell and the poor quality processing suggestions;

[0035] Among them, the second network large model is a large language model used to summarize the root causes of poor quality and poor quality processing suggestions of wireless network poor quality cells after the first network large model is trained. The question templates include questions about obtaining the root causes of poor quality based on a number of abnormal real-time wireless network poor quality cell data and questions about obtaining poor quality processing suggestions based on a number of historical processing experiences.

[0036] Furthermore, the method also includes assisting in pre-training a large network model, specifically including:

[0037] Receive the historical diagnosis subtask distributed by the network optimization expert agent, where the historical diagnosis subtask is generated after the network optimization expert agent trains the first network model;

[0038] Determine the wireless network API interface that needs to be called according to the historical diagnosis subtask, and obtain historical wireless network poor quality cell data through the wireless network API interface;

[0039] According to the expert knowledge of wireless network poor quality diagnosis, it is judged that there are abnormal historical wireless network poor quality cell data and the corresponding historical processing experience information;

[0040] The abnormal historical wireless network poor quality cell data and the corresponding historical processing experience information are sent to the network optimization expert Agent, so that the network optimization expert Agent trains the second network large model according to the abnormal historical wireless network poor quality cell data and the corresponding historical processing experience information sent by each quality expert Agent.

[0041] In a third aspect, the present disclosure provides a network optimization expert Agent, the network optimization expert Agent comprising:

[0042] The compilation and distribution module is used to compile several diagnostic subtasks of poor quality cells in the wireless network and their chain relationships, and distribute each diagnostic subtask to the corresponding poor quality expert Agent, so that each poor quality expert Agent obtains its own diagnostic subresult according to the corresponding diagnostic subtask, and sends the diagnostic subresult to the network optimization expert Agent;

[0043] The receiving and summarizing module is connected to the compilation and distribution module, and is used to receive the diagnosis sub-results sent by each poor quality expert Agent, and obtain the diagnosis result of the poor quality cell of the wireless network according to a number of diagnosis sub-results and the corresponding chain relationship. The diagnosis result includes the poor quality root cause of the poor quality cell of the wireless network and the poor quality treatment suggestion;

[0044] Among them, Agent is an intelligent agent, the network optimization expert Agent is an intelligent agent with expert knowledge of wireless network optimization, and the poor quality expert Agent is an intelligent agent with expert knowledge of wireless network poor quality diagnosis.

[0045] In a fourth aspect, the present disclosure provides a quality difference expert Agent, wherein the quality difference expert Agent comprises:

[0046] The task receiving module is used to receive the diagnosis subtasks distributed by the network optimization expert Agent. The diagnosis subtasks are the diagnosis subtasks of the poor quality cells of the wireless network compiled by the network optimization expert Agent and the diagnosis subtasks corresponding to the poor quality expert Agent in the chain relationship;

[0047] An execution task module is connected to the receiving task module, and is used to obtain a diagnosis sub-result according to a corresponding diagnosis sub-task, and send the diagnosis sub-result to the network optimization expert agent, so that the network optimization expert agent obtains a diagnosis result of a poor quality cell of the wireless network according to a plurality of diagnosis sub-results sent by a plurality of poor quality expert agents and a corresponding chain relationship, and the diagnosis result includes a poor quality root cause of the poor quality cell of the wireless network and a poor quality treatment suggestion;

[0048] Among them, Agent is an intelligent agent, the network optimization expert Agent is an intelligent agent with expert knowledge of wireless network optimization, and the poor quality expert Agent is an intelligent agent with expert knowledge of wireless network poor quality diagnosis.

[0049] In a fifth aspect, the present disclosure provides a system for processing a poor quality cell in a wireless network, the system comprising:

[0050] A network optimization expert Agent, used to implement the method for processing a poor quality cell in a wireless network as described in the first aspect above;

[0051] A plurality of poor quality expert Agents are connected to the network optimization expert Agent to implement the method for processing poor quality cells in a wireless network as described in the second aspect above.

[0052] In a sixth aspect, the present disclosure provides a computer-readable storage medium, wherein a computer program is stored in the computer-readable storage medium. When the computer program is executed by a processor, the method for processing poor quality cells in a wireless network as described above is implemented.

[0053] The present invention provides a method, device and medium for processing poor-quality cells in a wireless network, and adopts multi-agent collaboration to provide a network optimization auxiliary tool with a high degree of intelligence and a high processing speed. Through the collaboration of a network optimization expert agent and a plurality of poor-quality expert agents, the network optimization expert agent compiles a diagnosis subtask and its chain relationship, and the poor-quality expert agent executes the diagnosis subtask to obtain a diagnosis subresult. Finally, the network optimization expert agent summarizes each diagnosis subresult according to the chain relationship to obtain the poor-quality root cause and poor-quality processing suggestion of the poor-quality cell in the wireless network. The method can be used to assist network optimization engineers in completing network optimization work, and improve network optimization efficiency by completing the diagnosis of each dimension based on multiple agents respectively, and improve the accuracy of diagnosis by obtaining the final diagnosis result based on the multi-dimensional network poor-quality diagnosis subresult. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 is a flow chart of a method for processing a poor quality cell in a wireless network according to an embodiment of the present disclosure;

[0055] Figure 2 is a flow chart of another method for processing a poor quality cell in a wireless network according to an embodiment of the present disclosure;

[0056] Figure 3 is a structural diagram of a network optimization expert Agent according to an embodiment of the present disclosure;

[0057] Figure 4 is a structural diagram of a quality difference expert Agent according to an embodiment of the present disclosure;

[0058] Figure 5 It is a structural schematic diagram of a system for processing poor quality cells in a wireless network according to an embodiment of the present disclosure;

[0059] Figure 6 is a flow chart of another method for processing a poor quality cell in a wireless network according to an embodiment of the present disclosure;

[0060] Figure 7 It is an architecture diagram of a system for processing poor quality cells in a wireless network according to an embodiment of the present disclosure;

[0061] Figure 8 is an example diagram of a poor quality root cause diagnosis chain according to an embodiment of the present disclosure;

[0062] Fig. 9 is an example diagram of a cell PRB-level waveform according to an embodiment of the present disclosure;

[0063] Fig.10 is a flowchart of an execution method of an execution body of an embodiment of the present disclosure;

[0064] Fig.11 is a page schematic diagram of the quality difference analysis auxiliary tool according to an embodiment of the present disclosure;

[0065] Fig.12 This is an example diagram of a page of the quality difference analysis auxiliary tool according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0066] In order to enable those skilled in the art to better understand the technical solution of the present disclosure, the embodiments of the present disclosure will be further described in detail below with reference to the accompanying drawings.

[0067] It should be understood that the specific embodiments and drawings described herein are only used to explain the present disclosure rather than to limit the present disclosure.

[0068] It can be understood that, in the absence of conflict, the various embodiments of the present disclosure and the various features in the embodiments can be combined with each other.

[0069] It can be understood that, for the convenience of description, the drawings of the present disclosure only show parts related to the present disclosure, while parts irrelevant to the present disclosure are not shown in the drawings.

[0070] It can be understood that each module or unit involved in the embodiments of the present disclosure may correspond to only one entity structure, or may be composed of multiple entity structures, or multiple modules or units may be integrated into one entity structure.

[0071] It will be understood that, in the absence of conflict, the functions and steps marked in the flowcharts and block diagrams of the present disclosure may occur in an order different from that marked in the drawings.

[0072] It is understood that the flowcharts and block diagrams of the present disclosure illustrate the possible architectures, functions, and operations of the systems, devices, equipment, and methods according to the various embodiments of the present disclosure. Each box in the flowchart or block diagram may represent a module, unit, program segment, or code, which contains executable instructions for implementing the specified functions. Moreover, each box or combination of boxes in the block diagram and flowchart may be implemented by a hardware-based device that implements the specified functions, or may be implemented by a combination of hardware and computer instructions.

[0073] It can be understood that the modules and units involved in the embodiments of the present disclosure can be implemented by software or hardware, for example, the modules and units can be located in a processor.

[0074] Embodiment 1:

[0075] like Figure 1 As shown, the present disclosure provides a method for processing a poor quality cell in a wireless network, the method is applied to a network optimization expert Agent, and includes:

[0076] S11, compile several diagnosis subtasks of poor quality cells of wireless networks and their chain relationships, distribute each diagnosis subtask to the corresponding poor quality expert Agent, so that each poor quality expert Agent obtains its own diagnosis subresult according to the corresponding diagnosis subtask, and sends the diagnosis subresult to the network optimization expert Agent;

[0077] S12, receiving the diagnosis sub-results sent by each poor quality expert Agent, and obtaining the diagnosis result of the poor quality cell of the wireless network according to the plurality of diagnosis sub-results and the corresponding chain relationship, wherein the diagnosis result includes the poor quality root cause of the poor quality cell of the wireless network and the poor quality treatment suggestion;

[0078] Among them, Agent is an intelligent agent, the network optimization expert Agent is an intelligent agent with expert knowledge of wireless network optimization, and the poor quality expert Agent is an intelligent agent with expert knowledge of wireless network poor quality diagnosis.

[0079] In this embodiment, multi-agent collaboration is adopted to provide a network optimization auxiliary tool with a high degree of intelligence and a fast processing speed. Through the collaboration of a network optimization expert Agent and several quality difference expert Agents, the network optimization expert Agent compiles the diagnosis subtasks and their chain relationships, and the quality difference expert Agent executes the diagnosis subtasks to obtain the diagnosis subresults. Finally, the network optimization expert Agent summarizes the diagnosis subresults according to the chain relationships to obtain the poor quality root causes and quality difference processing suggestions of the poor quality cells in the wireless network. This can be used to assist network optimization engineers in completing network optimization work. The network optimization efficiency is improved by completing the diagnosis of each dimension based on multiple agents respectively. The final diagnosis result is obtained based on the multi-dimensional network quality difference diagnosis subresults to improve the accuracy of the diagnosis.

[0080] and Figure 1 The corresponding method applied to the quality difference expert agent is as follows Figure 2 As shown in the figure, the corresponding structure of the network optimization expert Agent is as follows Figure 3 As shown in the figure, the structure of the quality difference expert Agent is as follows Figure 4 As shown, the corresponding modules execute the steps with corresponding numbers. Figure 2 The corresponding detailed description of the method is not repeated, and those skilled in the art can easily deduce the action of the other end according to the corresponding relationship. In addition, the wireless network poor quality cell processing system of the integrated network optimization expert Agent and the poor quality expert Agent is as follows Figure 5 As shown, the comprehensive method is Figure 6 shown.

[0081] Specifically, a method and device for processing poor quality cells in wireless networks based on multi-agents are provided. Agent, namely AIAgent, refers to an intelligent agent, which is a system driven by a large language model as its brain, and has the ability to autonomously understand, perceive, plan, remember and use tools, and can automatically perform and complete complex tasks. This embodiment uses multi-agent collaboration to analyze the root causes of poor quality cells in wireless networks, and proposes suggestions for processing and repair. A poor quality cell refers to a cell with one or more poor quality problems in network communication quality. The types of poor cell quality problems involve multiple aspects, such as voice service connection rate, call drop rate, packet loss rate, data service drop rate, access success rate, user perception rate, etc.

[0082] At present, the identification of the problem types of poor quality cells mainly adopts the traditional network optimization (abbreviated as: network optimization) method. Specifically, first collect the key performance indicators (KPI) of the cell, and then the network optimization staff sets thresholds for each KPI based on their own knowledge and experience, and then judge the type of poor quality problems of the cell based on the thresholds of these KPIs and the various KPIs of the cell. Then, the network optimization staff analyzes the data, manually checks various network problems such as coverage, interference, and capacity, and conducts root cause analysis and formulates optimization solutions based on experience, so as to optimize the network problems. From the current situation, the root cause analysis of wireless network problems generally relies on manual experience and combines various tool software for analysis, which has many problems: the lack of personnel with professional skills, and the lack of expertise has become an optimization bottleneck; the network problems are complex, the causes are diverse, and the problems are difficult to locate; the problems are repeatedly handled, the processing timeliness is poor, and the cycle is long; the network problem mining has certain limitations, the real pain point is not clearly located, the human investment cost is high, and the benefit is low.

[0083] With the continuous progress and development of the mobile communication network industry, high-speed and high-quality network application content has put forward higher requirements for network operations. Network operators hope to "discover problems before users". In addition to the conventional manual statistics of cell KPIs and the periodic discovery of poor-quality cells defined by manual rules, the design and application of machine learning-oriented mobile communication network poor-quality cell positioning analysis methods are conducive to discovering poor-quality problems in the network, and can predict the degree of poor quality of the network and whether it affects the user's perception of use. With the help of machine learning, the comprehensive analysis of the positioning of poor-quality cells in the mobile communication network can timely predict and deal with network problems in advance, bringing users a better service experience, which has certain practical significance and value.

[0084] Therefore, this embodiment proposes the concept of multi-agent intelligent body, and constructs a wireless poor quality cell analysis and processing intelligent body based on a large model multi-agent architecture. The front-line network optimization engineer throws out the actual wireless poor quality cell problem in the form of natural language. After the network optimization expert agent of the multi-agent architecture reads the problem, it calls the long short-term memory framework as the center of the multi-agent to understand the intention of the problem. After understanding the intention of wireless poor quality cell analysis and processing, it calls the thinking framework to plan the poor quality cell processing task and decompose the task objectives, and then conducts diagnosis and analysis through the poor quality expert agent, where the poor quality expert agent includes multiple dimensions The single-agent collaboration of each dimension can schedule the tool set to execute to complete subtasks, including real-time data acquisition of key indicators, expert knowledge base cases and decision-making small model calls. Through the mutual collaboration between different single-agents in the quality-poor expert Agent, the root cause analysis results and intelligent diagnosis solutions of the poor-quality cells are obtained, and then returned to the network optimization expert Agent for summary and thinking to generate the final answer, which is returned to the network optimization engineer. Finally, a full-cycle closed-loop capability for analysis and processing of wireless poor-quality cells based on natural language input is built, which improves the intelligence of fault diagnosis, improves the work efficiency of front-line network optimization engineers, and reduces the cost of network operation and maintenance investment.

[0085] and Figure 5 The corresponding system solution overall design architecture is as follows Figure 7 As shown, the network optimization expert Agent pre-classifies the diagnosis results (diagnosis work orders) to train the large language model, making it the network large model of this embodiment. The network optimization expert Agent realizes the ability of diagnosis classification and induction and aggregation through thought chain and sub-goal decomposition. It is the hub for processing poor quality cells in wireless networks. Multiple diagnostic agents (poor quality expert agents) are connected to wireless network tools, and can call on wireless network indicators, logs, configurations, reports and other data to realize multi-dimensional diagnosis of poor quality problems in cells, and have the ability to automatically call tools to execute instructions. During use, the user raises the problem of poor quality cells in the actual wireless network language service in the form of natural language. After reading the problem, the network optimization expert agent uses long-term and short-term memory to identify the intention of the problem, thinks according to the content of the intention understanding, and uses the large model thinking chain ability to carry out task decomposition and planning. The network optimization expert agent then forwards it to the poor quality expert agent. Through the mutual cooperation between different single intelligent agents in the poor quality expert agent, the root cause analysis results and intelligent diagnosis solutions of the poor quality cells are obtained, and then returned to the network optimization expert agent for summary and thinking to generate the final answer. After integrating all the information, it is summarized into two types of instructions: visualization instructions and page instructions. After receiving the instructions, the network optimization expert agent executes them and finally presents the complete answer to the user. The overall design process of the solution is as follows Figure 6 As shown, the following will be described in detail Figure 6 Each step in .

[0086] In one implementation, S11 compiles several diagnosis subtasks of poor quality cells in wireless networks and their chain relationships, and distributes each diagnosis subtask to a corresponding poor quality expert Agent, specifically including:

[0087] receiving user questions through the first network big model, and identifying the user's intention to inquire about the cell with poor wireless network quality in the question;

[0088] Determine the real-time poor wireless network quality type of the poor wireless network quality cell according to the statistical result of the indicator of the poor wireless network quality cell;

[0089] The first network model uses the thinking chain question and answer prompt words to compile several diagnostic subtasks and their chain relationships of the poor quality cells of the wireless network according to the real-time poor quality type of the wireless network, wherein the several diagnostic subtasks include at least one of: fault diagnosis, parameter diagnosis, coverage diagnosis, quality diagnosis, interference diagnosis, capacity diagnosis, switching diagnosis, and hidden fault diagnosis;

[0090] Distribute each diagnosis subtask to a corresponding quality difference expert agent, the quality difference expert agent includes: at least one of a fault diagnosis agent, a parameter diagnosis agent, a coverage diagnosis agent, a quality diagnosis agent, an interference diagnosis agent, a capacity diagnosis agent, a switching diagnosis agent, and a hidden fault diagnosis agent;

[0091] Among them, the first network large model is a large language model used to compile several diagnostic subtasks and their chain relationships after training based on wireless network optimization problems. The thinking chain question and answer prompts are obtained in advance based on the diagnostic process in the historical wireless network poor quality cell diagnosis work order.

[0092] In this embodiment, if Figure 6 As shown, S01: using the network large model as an interactive entrance, the user's wireless voice service poor quality cell problem is input into the dialog box.

[0093] The problem of poor quality of wireless voice service cells that front-line users need to analyze and deal with is input into the dialog box through natural language interaction. The relevant indicator types and quality definitions of poor quality of wireless network cells are shown in Table 1 below:

[0094] Table 1. Types of related indicators and definitions of poor quality in poor wireless network cells

[0095]

[0096] After defining the poor quality cells in the table above, users can ask the network model dialog box about the poor quality cells. The network model will judge the poor quality cells according to the definitions of different poor quality types and determine the poor quality type as one of "high dropped calls", "low connection", "high interference", "high dropped calls", "high load", "low switching" and "low traffic". For example, if the following inquiry method is used: What is the poor quality of Shanghai International Exhibition Center West Plaza_1 on October 19? After entering it into the large model dialog box, end step 01 and enter step 02.

[0097] S02: After the user's problem is submitted to the network optimization expert agent for reading through the network big model, the network optimization expert agent, as the central agent, uses long-term and short-term memory to identify the intention and plan the problem. The user raises the actual wireless poor quality cell network problem in the form of natural language. The problem is first received by the network optimization expert agent. The network optimization expert agent, as the central agent, uses long-term and short-term memory to identify the intention and plan the problem, determine which professional fields and related diagnosis agents are involved, and then send it to the poor quality expert agent to start calling different sub-agents in parallel (such as Figure 7 After completion, step 02 ends and step 03 is entered.

[0098] S03: Use the thinking chain capability of the network big model to plan and arrange the tasks after the intent recognition, and determine which professional fields and related diagnosis agents need to be used through the quality difference expert agent. When the network big model receives the intent result completed by the network optimization expert agent, it sends it to the quality difference expert agent. The agent uses the thinking chain capability after incremental pre-training and SFT instruction fine-tuning to plan the analysis task of the quality difference cell problem, and mainly decomposes the task planning from the following aspects: 1) fault diagnosis, 2) parameter diagnosis, 3) coverage diagnosis, 4) quality diagnosis, 5) interference diagnosis, 6) capacity diagnosis, 7) hidden fault diagnosis, such as Figure 8 As shown, go to step 04 after completion.

[0099] S04: The quality difference expert agent decomposes the quality difference problem plan to sub-agents of different dimensions. Each sub-agent conducts quality difference problem analysis through the specified calling tool, and feeds back the analysis and diagnosis results to the network optimization expert agent for problem diagnosis classification and induction. Note that S04 is the action of the quality difference expert agent. After each sub-agent receives the quality difference problem decomposition task sent by the quality difference expert, it conducts quality difference problem analysis through the specified calling tool (such as the docking API interface (Application Programming Interface) of each sub-task to obtain real-time data), including:

[0100] 1. Fault Agent:

[0101] 1) Calling tool: base station side alarm API interface.

[0102] 2) Quality problem analysis process:

[0103] ① If the conditions of step 1 are met, go to step 2 and the fault agent diagnosis ends.

[0104] ② If the condition of step 1 is not met, go to step 3 and the fault agent diagnosis ends.

[0105] Step 1: First call the base station alarm API interface to check whether the base station has any alarms that affect normal business communications, such as station disconnection alarms and cell unavailable alarms.

[0106] Step 2: After checking the alarm on the base station side, the cell generated a cell unavailable fault alarm at XX:XX:XX:XX on X / X / 2024. It is recommended that the maintenance engineer handle the fault and troubleshoot the cause. The fault agent diagnosis is completed.

[0107] Step 3: After checking the alarms on the base station side, no abnormal alarms are found, which is not an alarm problem. It is transferred to other sub-agents for processing (except other sub-agents of the alarm agent), and the fault agent diagnosis is completed.

[0108] The diagnostic analysis results of the fault agent calling the API interface are fed back to the network optimization expert agent for problem diagnosis classification and induction.

[0109] 2. Parameter Agent:

[0110] 1) Calling tool: Cell CM configuration file API interface (including timer parameters, switching parameters and other data).

[0111] 2) Quality problem analysis process:

[0112] ① If the conditions in step 1 are met, proceed to step 2 and the parameter Agent diagnosis ends.

[0113] ② If the conditions in step 1 are not met, go to step 3 and the parameter Agent diagnosis ends.

[0114] Step 1: First call the cell CM configuration file API interface to check whether there is any unreasonable parameter configuration problem in the cell, such as any parameter such as the cell air interface timer, minimum access level, and inter-system switching threshold exceeding the reasonable range.

[0115] Step 2: After querying the parameter configuration in the cell CM configuration file, the air interface timer T300 configuration of the cell parameter exceeds the reasonable range, which has a certain impact on the RRC establishment success rate and data service perception of the cell. It is recommended to reset it within a reasonable range, and the parameter Agent diagnosis is completed.

[0116] Example: During parameter Agent diagnostic analysis, the analysis result is abnormal

[0117] a) T300 = 2000 (reasonable range 0 ~ 1000ms), parameter configuration is abnormal

[0118] b) T304 = 500 (reasonable range 500 ~ 2000ms), parameter configuration is normal

[0119] c) N310 = 10 (reasonable range 6 to 20), parameter configuration is normal

[0120] d) N311 = 1 (reasonable range 1 to 2), parameter configuration is normal

[0121] e) Minimum access level = -126 (reasonable range -110 to -128), parameter configuration is normal

[0122] f) A2 event RSRP threshold (different system data) = -118 (reasonable range -107 to -118), parameter configuration is normal

[0123] g) B2 event RSRP threshold 1 (data) = -116 (reasonable range -110 to -118), parameter configuration is normal

[0124] h) B2 event RSRP threshold 2 (data) = -118 (reasonable range -110 to -118), parameter configuration is normal. Step 3: After querying the parameter configuration in the cell CM configuration file, no abnormal parameter configuration problem is found. It is not a parameter configuration problem and is transferred to other sub-agents for processing (other sub-agents except the parameter agent). The parameter agent diagnosis is completed.

[0125] Example: During parameter Agent diagnosis analysis, the analysis result is normal.

[0126] a) T300 = 1000 (reasonable range 0 ~ 1000ms), parameter configuration is normal

[0127] b) T304 = 500 (reasonable range 500 ~ 2000ms), parameter configuration is normal

[0128] c) N310 = 10 (reasonable range 6 to 20), parameter configuration is normal

[0129] d) N311 = 1 (reasonable range 1 to 2), parameter configuration is normal

[0130] e) Minimum access level = -126 (reasonable range -110 to -128), parameter configuration is normal

[0131] f) A2 event RSRP threshold (different system data) = -118 (reasonable range -107 to -118), parameter configuration is normal

[0132] g) B2 event RSRP threshold 1 (data) = -116 (reasonable range -110 to -118), parameter configuration is normal

[0133] h) B2 event RSRP threshold 2 (data) = -118 (reasonable range -110 to -118), parameter configuration is normal. The parameter Agent calls the API interface for diagnostic analysis results, which are then fed back to the network optimization expert Agent for problem diagnosis classification and induction.

[0134] 3. Overwrite Agent:

[0135] 1) Calling tool: Community MR indicator API interface.

[0136] 2) Quality problem analysis process:

[0137] ① If the conditions of step 1 are met, go to step 2 and the coverage agent diagnosis ends.

[0138] ② If the conditions in step 1 are not met, go to step 3 and the overwriting Agent diagnosis ends.

[0139] Step 1: First call the cell MR indicator API interface to check whether there is any unreasonable coverage problem in the cell, such as whether any indicator such as weak coverage ratio, over-coverage rate, near-coverage rate, etc. exceeds the reasonable range.

[0140] Step 2: After querying the MR indicators of the cell, the weak coverage ratio (RSRP<=-105 ratio) or weak coverage + over coverage ratio or weak coverage + near coverage ratio of the cell exceeds the reasonable range, which has a certain impact on the RRC establishment success rate and data service perception of the cell. It is recommended to check the cell parameters and make RF optimization adjustments. The coverage Agent diagnosis is completed.

[0141] Example 1: When only the weak coverage ratio exceeds the reasonable range

[0142] (1) Weak coverage ratio (RSRP <= -105 ratio) = 12.6% (reasonable range 0-10%), abnormal indicators

[0143] (2) Total tadv sampling points = 9426

[0144] (3) Number of over-coverage sampling points = 2465

[0145] (4) Over-coverage rate = 16.15% (the reasonable range of TA>1km is 0-25%)

[0146] (5) Number of near-coverage sampling points = 25

[0147] (6) Near coverage rate = 1.28% (reasonable range 0-20%)

[0148] After diagnosis, the cell has a weak coverage problem. It is possible that users with weak signals at the edge of the cell coverage access the cell, which has a certain impact on service perception. Query the cell parameters:

[0149] (1)Standing height = 50 meters

[0150] (2) Mechanical downtilt = 2 degrees, electronic downtilt = 0 degrees, azimuth = 60 degrees

[0151] For weak coverage, we recommend optimizing it from the following aspects:

[0152] (1) Downtilt angle verification: By verifying the electronic downtilt angle and mechanical downtilt angle of the antenna feeder, the downtilt angle configuration can be appropriately adjusted to improve weak downlink coverage.

[0153] (2) Azimuth check: Reasonably plan and adjust the site azimuth based on the base station coverage.

[0154] (3) Power configuration parameter verification: Based on the base station equipment capabilities, evaluate the maximum supported transmit power and modify the parameters based on actual conditions.

[0155] (4) Base station height platform inspection: Evaluate base station coverage, check antenna height, and improve weak downlink coverage by adjusting antenna height.

[0156] (5) Site planning assessment: Evaluate new sites in areas with weak coverage and improve weak downlink coverage through construction plans

[0157] Example 2: When both the weak coverage ratio and the overcoverage ratio are beyond the reasonable range

[0158] (1) Weak coverage ratio (RSRP <= -105 ratio) = 12.6% (reasonable range 0-10%), abnormal indicators

[0159] (2) Total tadv sampling points = 9426

[0160] (3) Number of over-coverage sampling points = 2465

[0161] (4) Over-coverage rate = 28.85% (the reasonable range of TA>1km is 0-25%), and the indicator is abnormal

[0162] (5) Number of near-coverage sampling points = 25

[0163] (6) Near coverage rate = 1.28% (reasonable range 0-20%)

[0164] After diagnosis, it was found that the cell had over-coverage problem. Since the coverage was too far, users at the edge of coverage accessed, which had a certain impact on service perception. The following is the query of the cell's engineering parameter data:

[0165] (1)Standing height = 50 meters

[0166] (2) Mechanical downtilt = 2 degrees, electronic downtilt = 0 degrees, azimuth = 60 degrees

[0167] For over-coverage, we recommend optimizing the following aspects:

[0168] (1) Downtilt angle verification: By verifying the electronic downtilt angle and mechanical downtilt angle of the antenna feeder, the downtilt angle configuration can be appropriately adjusted to improve weak downlink coverage.

[0169] (2) Check power configuration parameters: Reduce the maximum transmit power of the cell

[0170] (3) Base station height platform inspection: Evaluate base station coverage, check antenna height, and reduce cross-area coverage by lowering antenna height.

[0171] (4) Azimuth check: Reasonably plan and adjust the site azimuth based on the base station coverage.

[0172] Example 3: When both the weak coverage ratio and the near coverage ratio are beyond the reasonable range

[0173] (1) Weak coverage ratio (RSRP <= -105 ratio) = 12.6% (reasonable range 0-10%), abnormal indicators

[0174] (2) Total tadv sampling points = 9426

[0175] (3) Number of over-coverage sampling points = 2465

[0176] (4) Over-coverage rate = 16.15% (the reasonable range of TA>1km is 0-25%)

[0177] (5) Number of near-coverage sampling points = 2005

[0178] (6) Near coverage rate = 21.35% (reasonable range 0-20%), abnormal indicator

[0179] After diagnosis, the cell has a near coverage problem, which has a certain impact on user service perception. Query the cell engineering parameters:

[0180] (1)Standing height = 50 meters

[0181] (2) Mechanical downtilt = 2 degrees, electronic downtilt = 0 degrees, azimuth = 60 degrees

[0182] For near coverage, it is recommended to optimize the following aspects:

[0183] (1) Effectively control coverage rationality through RF optimization, adjust azimuth and downtilt angles, and avoid blocking problems in the main direction of the antenna feed.

[0184] (2) On-site inspection to check if there are any billboards or other physical obstructions such as beautification antennas, and make corrections if necessary

[0185] (3) Site relocation: If the problem of too close coverage cannot be improved through rectification due to severe obstruction, site relocation can be carried out.

[0186] (4) Add a new base station. If the base station coverage cannot be optimized and improved, it is recommended to add a new planned base station in the surrounding area. Solution Step 3: After querying the MR indicators of the cell, no weak coverage, near coverage or over-coverage problems are found. It is not a coverage abnormality problem and is transferred to other sub-agents for processing (except the coverage agent). The coverage agent diagnosis is completed.

[0187] Example: Override Agent diagnostic analysis, the analysis result is normal

[0188] (1) Weak coverage ratio (RSRP <= -105 ratio) = 8.28% (reasonable range 0-10%)

[0189] (2) Total tadv sampling points = 9426

[0190] (3) Number of over-coverage sampling points = 2465

[0191] (4) Over-coverage rate = 16.15% (the reasonable range of TA>1km is 0-25%)

[0192] (5) Number of near-coverage sampling points = 25

[0193] (6) Near coverage rate = 1.28% (reasonable range 0-20%)

[0194] Based on the diagnosis and analysis results of the API interface of the coverage agent calling the cell MR indicators, the results are fed back to the network optimization expert agent for problem diagnosis, classification and summary.

[0195] 4. Quality Agent:

[0196] 1) Calling tool: Community KPI performance indicator API interface.

[0197] 2) Quality problem analysis process:

[0198] ① If the conditions of step 1 are met, proceed to step 2 and the quality agent diagnosis ends.

[0199] ② If the conditions in step 1 are not met, proceed to step 3 and the quality agent diagnosis ends.

[0200] Step 1: First call the cell KPI performance indicator API interface to check whether there are unreasonable quality problems in the cell, such as whether any indicator such as the MRO-SINR ≥ 0 sampling point ratio, CQI excellent rate, overlapping coverage ratio, MOD30 interference rate, etc. exceeds the reasonable range.

[0201] Step 2: After querying the KPI performance indicators of the cell, the data of eight situations, including the proportion of MRO-SINR ≥ 0 sampling points, the proportion of MRO-SINR ≥ 0 sampling points + CQI good rate, the proportion of MRO-SINR ≥ 0 sampling points + overlapping coverage, the proportion of MRO-SINR ≥ 0 sampling points + MOD30 interference rate, the proportion of MRO-SINR ≥ 0 sampling points + CQI good rate + overlapping coverage, the proportion of MRO-SINR ≥ 0 sampling points + CQI good rate + MOD30 interference rate, the proportion of MRO-SINR ≥ 0 sampling points + overlapping coverage + MOD30 interference rate, or the proportion of MRO-SINR ≥ 0 sampling points + overlapping coverage + MOD30 interference rate, exceeds a reasonable range, which has a certain impact on the RRC establishment success rate and user service perception of the cell. It is recommended to check the cell engineering parameters and power and optimize the RF and parameter adjustments. The quality agent diagnosis is completed.

[0202] Example 1: When the proportion of sampling points with only cell MRO-SINR ≥ 0 exceeds the reasonable range

[0203] (1) MRO-SINR≥0 sampling point ratio = 88.52% (reasonable range 90% to 100%), the indicator is abnormal

[0204] (2) CQI good rate = 97.19% (reasonable range 95% to 100%)

[0205] (3) Overlap coverage ratio (RSRP = -105) = 14.78% (reasonable range 0% to 20%)

[0206] (4) Number of MOD30 interference sampling points = 1280, MOD30 interference rate = 3.39% (reasonable range 0% to 10%)

[0207] After diagnosis, the proportion of sampling points with SINR ≥ 0 in the cell is low, and there are quality problems, which have a certain impact on service perception. Query the cell working parameters:

[0208] (1)Standing height = 50 meters

[0209] (2) Mechanical downtilt = 2 degrees, electronic downtilt = 0 degrees, azimuth = 60 degrees

[0210] In view of the low proportion of sampling points in the cell where SINR≥0, it is recommended to optimize and adjust from the following aspects:

[0211] (1) Neighboring cell optimization: Optimizing poor SINR quality caused by neighboring cell problems

[0212] (2) Overlap coverage optimization

[0213] (3) MOD30 interference optimization

[0214] (4) Interference problem troubleshooting and optimization

[0215] Example 2: When the cell MRO-SINR≥0 sampling point ratio + CQI good rate exceeds the reasonable range

[0216] (1) MRO-SINR≥0 sampling point ratio = 88.52% (reasonable range 90% to 100%), the indicator is abnormal

[0217] (2) CQI good rate = 93.21% (reasonable range 95% to 100%), abnormal index

[0218] (3) Overlap coverage ratio (RSRP = -105) = 14.78% (reasonable range 0% to 20%)

[0219] (4) Number of MOD30 interference sampling points = 1280, MOD30 interference rate = 3.39% (reasonable range 0% to 10%)

[0220] After diagnosis, the CQI rate of the cell is low, and there are quality problems, which have a certain impact on service perception. Query the cell parameters:

[0221] (1)Standing height = 50 meters

[0222] (2) Mechanical downtilt = 2 degrees, electronic downtilt = 0 degrees, azimuth = 60 degrees

[0223] In view of the low CQI rate of the cell, it is recommended to optimize and adjust from the following aspects:

[0224] (1) RF optimization for low CQI cells

[0225] (2) Modify the CQI configuration parameters and configure them according to the specific parameters of different manufacturers

[0226] (Huawei: CQI reliability optimization switch, PUCCH channel optimization and improvement switch, optimized normalized PDSCH power bias)

[0227] (ZTE: Optimize CQI false detection algorithm, power bias optimization, and PUCCH channel optimization to improve CQI reliability)

[0228] (Nosi: Enable periodic sub-band CQI reporting function, NPDSCHEPREOFF optimization, enable CQI adaptation)

[0229] Example 3: When the cell MRO-SINR ≥ 0 sampling point ratio + overlapping coverage ratio exceeds the reasonable range

[0230] (1) MRO-SINR≥0 sampling point ratio = 88.52% (reasonable range 90% to 100%), the indicator is abnormal

[0231] (2) CQI good rate = 96.28% (reasonable range 95% to 100%)

[0232] (3) Overlap coverage ratio (RSRP = -105) = 24.78% (reasonable range 0% to 20%), abnormal indicator

[0233] (4) Number of MOD30 interference sampling points = 1280, MOD30 interference rate = 3.39% (reasonable range 0% to 10%)

[0234] After diagnosis, the overlapping coverage ratio of the cell is high, and there are quality problems, which have a certain impact on service perception. Query the cell engineering parameters:

[0235] (1)Standing height = 50 meters

[0236] (2) Mechanical downtilt = 2 degrees, electronic downtilt = 0 degrees, azimuth = 60 degrees

[0237] In view of the high proportion of overlapping coverage of cells, it is recommended to optimize and adjust from the following aspects:

[0238] (1) Adjust the antenna downtilt angle: optimize antenna coverage, reduce signal interference, and improve spectrum efficiency.

[0239] (2) Adjust the antenna azimuth: improve signal directivity, improve signal quality, enhance system performance, and reduce overlapping coverage.

[0240] (3) Adjust the antenna height: Improve signal propagation, ensure that the coverage area matches the service cell, and reduce the impact of obstructions.

[0241] (4) Site rectification and relocation: For unreasonable site locations, site rectification or site relocation will be carried out for sites with interference in surrounding areas.

[0242] (5) Change site frequency band: avoid frequency interference, increase data transmission rate, and optimize network performance.

[0243] (6) Adjust cell reference power: balance coverage, optimize signal strength, and reduce interference.

[0244] Example 4: When the cell MRO-SINR≥0 sampling point ratio + MOD30 interference rate exceeds the reasonable range

[0245] (1) MRO-SINR≥0 sampling point ratio = 88.52% (reasonable range 90% to 100%), the indicator is abnormal

[0246] (2) CQI good rate = 96.28% (reasonable range 95% to 100%)

[0247] (3) Overlap coverage ratio (RSRP = -105) = 14.78% (reasonable range 0% to 20%)

[0248] (4) Number of MOD30 interference sampling points = 1280, MOD30 interference rate = 13.39% (reasonable range 0% to 10%), abnormal indicators

[0249] After diagnosis, the MOD30 interference rate of the cell is high, and there are quality problems, which have a certain impact on service perception. Query the cell engineering parameters:

[0250] (1)Standing height = 50 meters

[0251] (2) Mechanical downtilt = 2 degrees, electronic downtilt = 0 degrees, azimuth = 60 degrees

[0252] In view of the high MOD30 interference rate in the community, it is recommended to optimize and adjust from the following aspects:

[0253] (1) PCI re-planning: Combine the PCI planning data of the entire network to reasonably plan the cell PCI to ensure PCI reuse distance and MOD value interference

[0254] (2) Antenna adjustment: Adjust the antennas of cells with the same MOD value to avoid interference between cells with the same MOD value.

[0255] (3) Power modification: reduce the transmission power of the interfering cell and reduce the interference signal level

[0256] Example 5: When the cell MRO-SINR≥0 sampling point ratio + CQI good rate + overlapping coverage ratio exceeds the reasonable range

[0257] (1) MRO-SINR≥0 sampling point ratio = 88.52% (reasonable range 90% to 100%), the indicator is abnormal

[0258] (2) CQI good rate = 92.28% (reasonable range 95% to 100%), abnormal index

[0259] (3) Overlap coverage ratio (RSRP = -105) = 24.78% (reasonable range 0% to 20%), abnormal indicator

[0260] (4) Number of MOD30 interference sampling points = 1280, MOD30 interference rate = 8.39% (reasonable range 0% to 10%), abnormal indicators

[0261] After diagnosis, the cell CQI excellent rate is low, the overlapping coverage ratio is high, and there are quality problems, which have a certain impact on service perception. Query the cell engineering parameters:

[0262] (1)Standing height = 50 meters

[0263] (2) Mechanical downtilt = 2 degrees, electronic downtilt = 0 degrees, azimuth = 60 degrees

[0264] In view of the low CQI rate in the cell, it is recommended to optimize and adjust from the following aspects:

[0265] (1) RF optimization for low CQI cells

[0266] (2) Modify the CQI configuration parameters and configure them according to the specific parameters of different manufacturers

[0267] (Huawei: CQI reliability optimization switch, PUCCH channel optimization and improvement switch, optimized normalized PDSCH power bias)

[0268] (ZTE: Optimize CQI false detection algorithm, power bias optimization, and PUCCH channel optimization to improve CQI reliability)

[0269] (Nosi: Enable periodic sub-band CQI reporting function, NPDSCHEPREOFF optimization, enable CQI adaptation)

[0270] In view of the high proportion of overlapping coverage of cells, it is recommended to optimize and adjust from the following aspects:

[0271] (1) Adjust the antenna downtilt angle: optimize antenna coverage, reduce signal interference, and improve spectrum efficiency.

[0272] (2) Adjust the antenna azimuth: improve signal directivity, improve signal quality, enhance system performance, and reduce overlapping coverage.

[0273] (3) Adjust the antenna height: Improve signal propagation, ensure that the coverage area matches the service cell, and reduce the impact of obstructions.

[0274] (4) Site rectification and relocation: For unreasonable site locations, site rectification or site relocation will be carried out for sites with interference in surrounding areas.

[0275] (5) Change site frequency band: avoid frequency interference, increase data transmission rate, and optimize network performance.

[0276] (6) Adjust cell reference power: balance coverage, optimize signal strength, and reduce interference.

[0277] Example 6: When the cell MRO-SINR≥0 sampling point ratio + CQI good rate + MOD30 interference rate exceeds the reasonable range

[0278] (1) MRO-SINR≥0 sampling point ratio = 88.52% (reasonable range 90% to 100%), the indicator is abnormal

[0279] (2) CQI good rate = 92.28% (reasonable range 95% to 100%), abnormal index

[0280] (3) Overlap coverage ratio (RSRP = -105) = 14.78% (reasonable range 0% to 20%)

[0281] (4) Number of MOD30 interference sampling points = 1280, MOD30 interference rate = 12.39% (reasonable range 0% to 10%), abnormal indicators

[0282] After diagnosis, the cell CQI excellent rate is low, MOD30 interference rate is high, there are quality problems, which have a certain impact on service perception. Query the cell working parameters:

[0283] (1)Standing height = 50 meters

[0284] (2) Mechanical downtilt = 2 degrees, electronic downtilt = 0 degrees, azimuth = 60 degrees

[0285] In view of the low CQI rate in the cell, it is recommended to optimize and adjust from the following aspects:

[0286] (1) RF optimization for low CQI cells

[0287] (2) Modify the CQI configuration parameters and configure them according to the specific parameters of different manufacturers

[0288] (Huawei: CQI reliability optimization switch, PUCCH channel optimization and improvement switch, optimized normalized PDSCH power bias)

[0289] (ZTE: Optimize CQI false detection algorithm, power bias optimization, and PUCCH channel optimization to improve CQI reliability)

[0290] (Nosi: Enable periodic sub-band CQI reporting function, NPDSCHEPREOFF optimization, enable CQI adaptation)

[0291] In view of the high MOD30 interference rate in the community, it is recommended to optimize and adjust from the following aspects:

[0292] (1) PCI re-planning: Combine the PCI planning data of the entire network to reasonably plan the cell PCI to ensure PCI reuse distance and MOD value interference

[0293] (2) Antenna adjustment: Adjust the antennas of cells with the same MOD value to avoid interference between cells with the same MOD value.

[0294] (3) Power modification: reduce the transmission power of the interfering cell and reduce the interference signal level

[0295] Example 7: When the cell MRO-SINR≥0 sampling point ratio + overlapping coverage ratio + MOD30 interference rate exceeds the reasonable range

[0296] (1) MRO-SINR≥0 sampling point ratio = 88.52% (reasonable range 90% to 100%), the indicator is abnormal

[0297] (2) CQI good rate = 98.28% (reasonable range 95% to 100%)

[0298] (3) Overlap coverage ratio (RSRP = -105) = 24.78% (reasonable range 0% to 20%), abnormal indicator

[0299] (4) Number of MOD30 interference sampling points = 1280, MOD30 interference rate = 12.39% (reasonable range 0% to 10%), abnormal indicators

[0300] After diagnosis, the cell MOD30 interference rate is high, the overlapping coverage ratio is high, and there are quality problems, which have a certain impact on service perception. Query the cell engineering parameters:

[0301] (1)Standing height = 50 meters

[0302] (2) Mechanical downtilt = 2 degrees, electronic downtilt = 0 degrees, azimuth = 60 degrees

[0303] In view of the high MOD30 interference rate in the community, it is recommended to optimize and adjust from the following aspects:

[0304] (1) PCI re-planning: Combine the PCI planning data of the entire network to reasonably plan the cell PCI to ensure PCI reuse distance and MOD value interference

[0305] (2) Antenna adjustment: Adjust the antennas of cells with the same MOD value to avoid interference between cells with the same MOD value.

[0306] (3) Power modification: reduce the transmission power of the interfering cell and reduce the interference signal level

[0307] In view of the high proportion of overlapping coverage of cells, it is recommended to optimize and adjust from the following aspects:

[0308] (1) Adjust the antenna downtilt angle: optimize antenna coverage, reduce signal interference, and improve spectrum efficiency.

[0309] (2) Adjust the antenna azimuth: improve signal directivity, improve signal quality, enhance system performance, and reduce overlapping coverage.

[0310] (3) Adjust the antenna height: Improve signal propagation, ensure that the coverage area matches the service cell, and reduce the impact of obstructions.

[0311] (4) Site rectification and relocation: For unreasonable site locations, site rectification or site relocation will be carried out for sites with interference in surrounding areas.

[0312] (5) Change site frequency band: avoid frequency interference, increase data transmission rate, and optimize network performance.

[0313] (6) Adjust cell reference power: balance coverage, optimize signal strength, and reduce interference.

[0314] Example 8: When the cell MRO-SINR≥0 sampling point ratio + overlapping coverage ratio + MOD30 interference rate + CQI good rate exceeds the reasonable range

[0315] (1) MRO-SINR≥0 sampling point ratio = 88.52% (reasonable range 90% to 100%), the indicator is abnormal

[0316] (2) CQI good rate = 92.28% (reasonable range 95% to 100%), abnormal index

[0317] (3) Overlap coverage ratio (RSRP = -105) = 24.78% (reasonable range 0% to 20%), abnormal indicator

[0318] (4) Number of MOD30 interference sampling points = 1280, MOD30 interference rate = 12.39% (reasonable range 0% to 10%), abnormal indicators

[0319] After diagnosis, the cell CQI excellent rate is low, MOD30 interference rate is high, and overlapping coverage ratio is high. There are quality problems, which have a certain impact on service perception. Query the cell engineering parameters:

[0320] (1)Standing height = 50 meters

[0321] (2) Mechanical downtilt = 2 degrees, electronic downtilt = 0 degrees, azimuth = 60 degrees

[0322] In view of the low CQI rate in the cell, it is recommended to optimize and adjust from the following aspects:

[0323] (1) RF optimization for low CQI cells

[0324] (2) Modify the CQI configuration parameters and configure them according to the specific parameters of different manufacturers

[0325] (Huawei: CQI reliability optimization switch, PUCCH channel optimization and improvement switch, optimized normalized PDSCH power bias)

[0326] (ZTE: Optimize CQI false detection algorithm, power bias optimization, and PUCCH channel optimization to improve CQI reliability)

[0327] (Nosi: Enable periodic sub-band CQI reporting function, NPDSCHEPREOFF optimization, enable CQI adaptation)

[0328] In view of the high MOD30 interference rate in the community, it is recommended to optimize and adjust from the following aspects:

[0329] (1) PCI re-planning: Combine the PCI planning data of the entire network to reasonably plan the cell PCI to ensure PCI reuse distance and MOD value interference

[0330] (2) Antenna adjustment: Adjust the antennas of cells with the same MOD value to avoid interference between cells with the same MOD value.

[0331] (3) Power modification: reduce the transmission power of the interfering cell and reduce the interference signal level

[0332] In view of the high proportion of overlapping coverage of cells, it is recommended to optimize and adjust from the following aspects:

[0333] (1) Adjust the antenna downtilt angle: optimize antenna coverage, reduce signal interference, and improve spectrum efficiency.

[0334] (2) Adjust the antenna azimuth: improve signal directivity, improve signal quality, enhance system performance, and reduce overlapping coverage.

[0335] (3) Adjust the antenna height: Improve signal propagation, ensure that the coverage area matches the service cell, and reduce the impact of obstructions.

[0336] (4) Site rectification and relocation: For unreasonable site locations, site rectification or site relocation will be carried out for sites with interference in surrounding areas.

[0337] (5) Change site frequency band: avoid frequency interference, increase data transmission rate, and optimize network performance.

[0338] (6) Adjust cell reference power: balance coverage, optimize signal strength, and reduce interference.

[0339] Step 3: After querying the cell performance KPI indicators, no problems such as low SINR ≥ 0 sampling point ratio, low CQI excellent rate, high MOD30 interference rate, and high overlapping coverage ratio were found. These are not quality abnormalities and are transferred to other sub-agents for processing (except the quality agent). The quality agent diagnosis is completed.

[0340] Example: Quality Agent diagnostic analysis, the analysis results are normal

[0341] (1) MRO-SINR ≥ 0 sampling point ratio = 98.52% (reasonable range 90% to 100%)

[0342] (2) CQI good rate = 97.28% (reasonable range 95% to 100%)

[0343] (3) Overlap coverage ratio (RSRP = -105) = 14.78% (reasonable range 0% to 20%)

[0344] (4) Number of MOD30 interference sampling points = 1280, MOD30 interference rate = 2.39% (reasonable range 0% to 10%)

[0345] The quality agent calls the API interface diagnosis and analysis results of the cell performance KPI indicators, and then feeds back to the network optimization expert agent for problem diagnosis classification and induction.

[0346] 5. Interference Agent:

[0347] 1) Calling tool: The system has deployed the uplink interference diagnosis AI small model API interface.

[0348] 2) Interference problem analysis process:

[0349] ① If the conditions of step 1 are met, go to step 2 and the interference agent diagnosis ends.

[0350] ② If the condition of step 1 is not met, go to step 3 and the interference agent diagnosis ends.

[0351] Step 1: First call the uplink interference diagnosis AI small model API interface deployed by the system to check whether there is an uplink interference problem in the cell, such as whether any indicator such as the average interference noise power per prb, uplink interference type, etc. exceeds a reasonable range.

[0352] Step 2: After querying the results of the AI ​​small model for uplink interference diagnosis in the cell, the results of the five situations, such as the average interference noise power per prb of the cell or the uplink interference type = blocking interference or the uplink interference type = narrowband interference or the uplink interference type = spurious interference or the uplink interference type = external interference, are beyond the reasonable range, which has a certain impact on the RRC establishment success rate and user service perception of the cell. It is recommended to do a good job of troubleshooting the uplink interference in the cell according to the different interference types diagnosed by the AI ​​small model, and the interference Agent diagnosis is completed.

[0353] Example 1: When only the average interference noise power per prb of the cell exceeds the reasonable range

[0354] (1) Average interference noise power per prb (dBm) = -98dBm (reasonable range <= -105dBm), abnormal indicator

[0355] (2) Uplink interference type = None

[0356] After diagnosis, it was found that the cell had uplink interference problems, which had a certain impact on service perception. The hourly PRB-level waveform of the cell was queried (such as Fig. 9 shown as an example only).

[0357] It is recommended to conduct a frequency scan on site to determine the interference source, and then observe whether the wireless connection rate indicator returns to normal after eliminating the interference.

[0358] Example 2: When the average interference noise power per prb in the cell exceeds the reasonable range and the uplink interference type = blocking interference

[0359] (1) Average interference noise power per prb (dBm) = -98dBm (reasonable range <= -105dBm), abnormal indicator

[0360] (2) Uplink interference type = blocking interference

[0361] After diagnosis, it is found that there is an uplink interference problem in the cell, which is a blocking interference and has a certain impact on service perception. The hourly PRB-level waveform of the cell is queried.

[0362] It is recommended to follow the steps below to troubleshoot interference:

[0363] (1) Install filters of the corresponding frequency band on the affected base stations to reduce interference.

[0364] (2) Increase the isolation between the two systems, such as raising the antenna height of the interference source base station or the interfered base station to change the horizontal isolation to vertical isolation (in general, the vertical isolation is more than 10 dB greater than the horizontal isolation.

[0365] (3) Check the anti-interference capability of the equipment in the affected cell and replace the interfered equipment with equipment with stronger anti-interference capability.

[0366] Example 3: When the average interference noise power per prb in the cell exceeds the reasonable range and the uplink interference type = spurious interference

[0367] (1) Average interference noise power per prb (dBm) = -98dBm (reasonable range <= -105dBm), abnormal indicator

[0368] (2) Uplink interference type = spurious interference

[0369] After diagnosis, it is found that there is an uplink interference problem in the cell, which is spurious interference and has a certain impact on service perception. The hourly PRB-level waveform diagram of the cell is queried.

[0370] It is recommended to follow the steps below to troubleshoot interference:

[0371] (1) By increasing the isolation between the TD-LTE base station antenna and the interference source base station antenna, the purpose of reducing interference can be achieved. Generally, horizontal isolation can be changed to vertical isolation.

[0372] (2) Reduce spurious interference by installing a bandpass filter at the interference source base station.

[0373] Example 4: When the average interference noise power per prb in the cell exceeds the reasonable range and the uplink interference type = narrowband interference

[0374] (1) Average interference noise power per prb (dBm) = -98dBm (reasonable range <= -105dBm), abnormal indicator

[0375] (2) Uplink interference type = narrowband interference

[0376] After diagnosis, it was found that there was an uplink interference problem in the cell, which was narrowband interference and had a certain impact on service perception. The hourly PRB-level waveform of the cell was queried.

[0377] It is recommended to follow the steps below to troubleshoot interference:

[0378] (1) Use frequency hopping technology to dynamically change the operating frequency to avoid interference sources with fixed frequencies.

[0379] (2) Use direct sequence spread spectrum (DSSS) or frequency hopping spread spectrum (FHSS) technology to improve the signal's anti-interference ability.

[0380] (3) Optimize base station layout and antenna direction to reduce sensitivity to interference sources.

[0381] (4) Adjust the transmit power and receive sensitivity to adapt to the interference environment.

[0382] Example 5: When the average interference noise power per prb in the cell exceeds the reasonable range and the uplink interference type = external interference

[0383] (1) Average interference noise power per prb (dBm) = -98dBm (reasonable range <= -105dBm), abnormal indicator

[0384] (2) Uplink interference type = external interference

[0385] After diagnosis, it is found that there is an uplink interference problem in the cell, which is external interference and has a certain impact on service perception. The hourly PRB-level waveform of the cell is queried.

[0386] It is recommended to follow the steps below to troubleshoot interference:

[0387] (1) Most external interference persists, so the interference source can be found more smoothly through frequency scanning and can be directly coordinated to be shut down.

[0388] (2) For non-continuous interference sources, find out the interference pattern based on hourly indicators and conduct on-site inspections during the interference period. For example, the time when the school exam jammer is turned on coincides with the time of commuting.

[0389] Some methods and techniques for troubleshooting external interference:

[0390] Method 1: Direction determination: Render the background noise value data into an intuitive heat map to guide field engineers to identify the interference location and direction.

[0391] Method 2: The drone quickly and effectively locates the interference location and direction;

[0392] Method 3: Manual carpet inspection to determine the final interference location;

[0393] Method 4: Three-point positioning method

[0394] Step 1: Determine several high points in the interference area. The first point is usually the building where the interfered base station is located. Use a spectrum analyzer to perform a 360-degree test to find the direction with the strongest interference.

[0395] Step 2: Find the second point along the direction of the strongest interference, perform a 360-degree frequency scan, and compare the interference intensity and direction with the first point to determine the approximate location of the interference source.

[0396] Step 3: Sweep the frequency at different high points in turn until the interference source is determined.

[0397] Site selection principles: Select a building with a wide field of view and a high position for frequency scanning to facilitate observation of the surrounding environment, check whether there are antennas, amplifiers and other devices, and avoid interference that may affect judgment.

[0398] Step 3: After querying the cell performance KPI indicators, no average interference noise power per prb and interference type problems are found, which does not belong to abnormal uplink interference problems. It is transferred to other sub-agents for processing (other sub-agents except the interference agent), and the interference agent diagnosis is completed.

[0399] Example: Interference Agent diagnostic analysis, the analysis result is normal

[0400] (1) Average interference noise power per prb (dBm) = -110dBm (reasonable range <= -105dBm)

[0401] (2) Uplink interference type = None

[0402] The interference agent calls the API interface diagnosis and analysis results of the uplink interference model, and then feeds back to the network optimization expert agent for problem diagnosis classification and induction.

[0403] 6. Capacity Agent:

[0404] 1) Calling tool: Community KPI performance indicator API interface.

[0405] 2) Capacity problem analysis process:

[0406] ① If the conditions in step 1 are met, proceed to step 2 and the capacity agent diagnosis ends.

[0407] ② If the conditions in step 1 are not met, proceed to step 3 and the capacity agent diagnosis ends.

[0408] Step 1: First call the cell KPI performance indicator API interface to check whether the cell has insufficient capacity load, such as whether any indicator such as uplink PRB utilization and downlink PRB utilization exceeds the reasonable range.

[0409] Step 2: After querying the cell KPI performance indicators, the uplink PRB utilization rate or uplink PRB utilization rate or uplink and downlink PRB utilization rate of the cell exceeds the reasonable range, which has a certain impact on the RRC establishment success rate and user service perception of the cell. It is recommended to check the cell working parameters and load balancing parameters and optimize the RF and parameters. The capacity agent diagnosis is completed.

[0410] Example: When the cell uplink PRB utilization or downlink PRB utilization or uplink and downlink PRB utilization both exceed the reasonable range

[0411] (1) Uplink PRB utilization = 72% (reasonable range 0-60%), abnormal indicator (take the maximum value of the total 5G traffic in the whole network in 24 hours) or downlink PRB utilization = 85% (reasonable range 0-80%), abnormal indicator (take the maximum value of the total 5G traffic in the whole network in 24 hours) or uplink PRB utilization = 62% (reasonable range 0-60%) and downlink PRB utilization = 87% (reasonable range 0-80%), abnormal indicator

[0412] (2)arfcn=636664

[0413] (3) Average number of 5G users = 826

[0414] (4) 5G traffic (5G uplink traffic + 5G downlink traffic) = 928.5 Gb

[0415] After diagnosis, it was found that the cell had an overload problem, which had a certain impact on service perception. It was recommended to optimize and adjust from the following aspects:

[0416] (1) Check load balancing parameters and configure them appropriately

[0417] (2) RF optimization adjustment

[0418] (3) Expansion of cells with the same or different frequencies

[0419] (4) Newly built base stations share the load

[0420] Step 3: After querying the cell performance KPI indicators, no uplink PRB utilization or downlink PRB utilization is found to be beyond the reasonable range, which is not a capacity load abnormality problem. It is transferred to other sub-agents for processing (except the capacity agent), and the capacity agent diagnosis is completed.

[0421] Example: Capacity Agent diagnostic analysis is in progress and the analysis results are normal

[0422] (1) Uplink PRB utilization = 12% (reasonable range 0-60%) (take the maximum value of the total 5G traffic of the entire network in 24 hours) or downlink PRB utilization = 25% (reasonable range 0-80%) (take the maximum value of the total 5G traffic of the entire network in 24 hours) or uplink PRB utilization = 8% (reasonable range 0-60%) and downlink PRB utilization = 17% (reasonable range 0-80%)

[0423] (2)arfcn=636664

[0424] (3) Average number of 5G users = 16

[0425] (4) 5G traffic (5G uplink traffic + 5G downlink traffic) = 2.8 Gb

[0426] The capacity agent calls the API interface diagnosis and analysis results of the cell KPI performance indicators, and then feeds back to the network optimization expert agent for problem diagnosis, classification and induction.

[0427] 7. Switch Agent:

[0428] 1) Calling tool: Community KPI performance indicator API interface.

[0429] 2) Switching problem analysis process:

[0430] ① If the conditions in step 1 are met, proceed to step 2 and the switch Agent diagnosis ends.

[0431] ② If the condition in step 1 is not met, go to step 3 and switch Agent diagnosis to end.

[0432] Step 1: First call the cell KPI performance indicator API interface to check whether the cell has a high handover failure rate problem, such as whether any indicator such as the 5G intra-system handover success rate, 5G inter-system handover success rate, 5G intra-frequency handover success rate, and 5G inter-frequency handover success rate exceeds a reasonable range.

[0433] Step 2: After querying the KPI performance indicators of the cell, the data of any single situation or combination of the cell's 5G intra-system handover success rate, 5G inter-system handover success rate, 5G intra-frequency handover success rate, or 5G inter-frequency handover success rate exceeds a reasonable range, which has a certain impact on the user service perception of the cell. It is recommended to check the configuration of the cell's neighboring cells and optimize the RF and handover parameters. The handover Agent diagnosis is completed.

[0434] Example: When any single situation or combination of situations such as the 5G intra-system handover success rate, the 5G inter-system handover success rate, the 5G intra-frequency handover success rate, or the 5G inter-frequency handover success rate exceeds a reasonable range

[0435] (1) 5G system handover success rate = 92.35% (reasonable range 98-100%), abnormal indicator

[0436] (2) 5G inter-system handover success rate = 96.45% (reasonable range 98-100%), abnormal indicator

[0437] (3) 5G same-frequency handover success rate = 97.2% (reasonable range 98-100%), abnormal indicator

[0438] (4) 5G inter-frequency handover success rate = 91.48% (reasonable range 98-100%), abnormal indicator

[0439] After diagnosis, the cell switching success rate during this period is low. It is recommended to check the cell switching parameter configuration and neighboring cell configuration:

[0440] Neighboring cell configuration verification:

[0441] (1) Check for missed neighboring areas: timely fill in the gaps in neighboring areas

[0442] (2) Neighboring cell mismatch problem verification: mainly includes external neighboring cell frequency, PCI, cell ID, TAC, and PLMN configuration verification

[0443] (3) Checking the problem of over-allocation of neighboring cells: Based on the cell board model, obtain the maximum number of neighboring cells supported by the main control board, check the configured and planned neighboring cells, and reasonably add or delete neighboring cells.

[0444] Neighborhood class parameter optimization:

[0445] (1) NR_NR neighbor cell self-configuration switch (1 (ON))

[0446] (2) NR_NR Neighborhood self-deletion switch (1 (ON))

[0447] (3) NR self-deletion cycle (days) 1 (days)

[0448] (4) NR neighbor cell self-configuration controlled mode switch 0 (OFF)

[0449] (5) In the NR neighbor cell self-configuration controlled mode, the neighbor cell to be confirmed is allowed to switch to 0 (OFF)

[0450] (6) NR_EUTRAN neighbor cell self-configuration switch 0 (OFF)

[0451] Step 3: After querying the cell performance KPI indicators, no 5G intra-system handover success rate, 5G inter-system handover success rate, 5G intra-frequency handover success rate, and 5G inter-frequency handover success rate were found to be beyond the reasonable range. This is not a capacity handover abnormality problem and is transferred to other sub-agents for processing (except the handover agent). The handover agent diagnosis is completed.

[0452] Example: Switch Agent diagnostic analysis, the analysis result is normal

[0453] (1) 5G system handover success rate = 98.35% (reasonable range 98-100%), the indicator is normal

[0454] (2) 5G inter-system handover success rate = 99.45% (reasonable range 98-100%), normal indicators

[0455] (3) 5G same-frequency handover success rate = 99.2% (reasonable range 98-100%), the indicator is normal

[0456] (4) 5G inter-frequency handover success rate = 99.48% (reasonable range 98-100%), the indicator is normal

[0457] The switching agent calls the API interface diagnosis and analysis results of the cell KPI performance indicators, and then feeds back to the network optimization expert agent for problem diagnosis classification and induction.

[0458] 8. Hidden Fault Agent:

[0459] 1) Calling tools: cell fault, MR, KPI performance indicator API interface.

[0460] 2) Hidden fault problem analysis process:

[0461] ① If the conditions of step 1 are met, go to step 2 and the hidden fault agent diagnosis ends.

[0462] ② If the conditions in step 1 are not met, go to step 3 and the hidden fault agent diagnosis ends.

[0463] Step 1: First call the base station side alarm API interface, cell KPI performance indicator API interface, and cell MR indicator API interface to check whether the cell has base station failure, coverage, interference, quality, capacity, switching and other problems.

[0464] Step 2: After querying the alarms on the base station side, the cell KPI performance indicators, and the cell MR indicators, the data for any single or combined situation of the cell's fault type, coverage type, interference type, quality type, capacity type, switching type, etc. are normal, but it is still a poor-quality cell. It is recommended to conduct a thorough investigation of the cell's hidden faults. The hidden fault Agent diagnosis is completed.

[0465] Example: When querying the alarms on the base station side, the KPI performance indicators of the cell, and the MR indicators of the cell, if the data of any single or combined situation of the cell's fault type, coverage type, interference type, quality type, capacity type, switching type, etc. are normal, it is recommended to check them one by one according to the following steps:

[0466] (1) If the RRU software crashes, it is recommended to power off and reset the RRU in the background and front-end before observing the indicators.

[0467] (2) If the problem is still not solved after resetting, it is necessary to go to the site to check the antenna and feeder system, including whether the antenna is placed in a reasonable position, whether the waterproof tape at the end of the antenna and feeder cable is damaged, and whether the feeder cable is damaged.

[0468] (3) Replace the antenna or swap the antenna.

[0469] (4) If no hidden faults are found during on-site inspection, it is recommended to obtain relevant indicators on the DNS, CDN, core network, and SP server sides according to the end-to-end analysis method, and analyze them section by section to find the problems on the wireless air interface side.

[0470] (5) Check the base station terminal compatibility parameters and change the base station SRS reconfiguration mode switch to 0 (OFF).

[0471] According to the hidden fault agent, the base station alarm API interface, the cell KPI performance index API interface, and the cell MR index API interface are called, and then fed back to the network optimization expert agent for problem diagnosis classification and induction, and the hidden fault agent diagnosis is completed.

[0472] Step 3: After querying the alarms on the base station side, the cell KPI performance indicators, and the cell MR indicators, the data of any single or combined situation of the cell's fault type, coverage type, interference type, quality type, capacity type, switching type, etc. is abnormal, which does not belong to the hidden fault problem. It is transferred to other sub-agents for processing (except the sub-agents other than the hidden fault capsule Agent), and the hidden fault Agent diagnosis is completed.

[0473] When finished, proceed to step 05.

[0474] In one implementation, in S12, a diagnosis result of a poor quality cell of a wireless network is obtained according to a plurality of diagnosis sub-results and corresponding chain relationships, and the diagnosis result includes a poor quality root cause of the poor quality cell of the wireless network and a poor quality processing suggestion, specifically including:

[0475] Obtaining the thinking chain question and answer prompt words and their order corresponding to a number of diagnostic subtasks and their chain relationships, and inputting a number of diagnostic subresults into the question template of the second network macromodel in order to obtain diagnostic results including root causes of poor quality of wireless network poor quality cells and suggestions for poor quality processing, wherein the diagnostic subresults include abnormal real-time wireless network poor quality cell data and its corresponding historical processing experience information;

[0476] Outputting voice and / or text results of the root cause of poor quality of the wireless network poor quality cell and the poor quality treatment suggestion, displaying the wireless network poor quality cell on a map and marking its poor quality root cause, generating a page instruction for repairing the poor quality root cause according to the poor quality treatment suggestion, and automatically or under user triggering executing the page instruction to repair the wireless network poor quality cell;

[0477] Among them, the second network large model is a large language model used to summarize the root causes of poor quality and poor quality processing suggestions of wireless network poor quality cells after the first network large model is trained. The question templates include questions about obtaining the root causes of poor quality based on a number of abnormal real-time wireless network poor quality cell data and questions about obtaining poor quality processing suggestions based on a number of historical processing experiences.

[0478] In this embodiment, if Figure 6 S05 shown: the network optimization expert Agent performs comprehensive summarization on the diagnosis results and obtains two types of instructions: visualization instructions and page instructions. The execution body executes the instructions after receiving them (such as Figure 7 Finally, the complete answer is presented to the user.

[0479] like Figure 7 As shown, the network optimization expert Agent also serves as an inductive hub. Based on the chain of thought and combined with pre-stored knowledge (which can be implemented in the form of a network optimization expert knowledge base), it calls a preset problem template, such as summarizing the root causes of poor network quality in cells with poor network quality according to the following diagnostic sub-results. The second network model of the network optimization expert Agent has been pre-trained and can summarize the root causes of poor quality according to the diagnostic sub-results and their order filled in the problem template. In the same way, effective suggestions for handling poor quality can be summarized, such as giving priority to repairing a certain part of the abnormal causes that cause poor network quality or setting their repair order.

[0480] In such Figure 7In the overall architecture shown, the executor is responsible for using external tools as well as executing client instructions. After the network optimization expert Agent identifies and plans the user's problem intention, it then identifies and plans the processing result, determines the operation type and operation object involved, and then extracts the object unique identifier from the instruction library. After processing all the data in the format agreed with the instruction execution SDK (Software Development Kit), two types of instruction data are obtained: page instructions and visual instructions, such as Fig.10 shown.

[0481] Page instructions are instructions for the agent to present the specific steps of the agent's operation on the existing system page. After receiving the user's instructions, the client uses the page instructions to execute the SDK tool ("complete the user's query / operation goal through a question"). According to the received page instructions, the SDK tool is executed in the order in which the instructions are arranged. There is an error correction mechanism to ensure the correct and complete execution of the instructions. In the event of an instruction execution failure, the error correction mechanism will use a backup plan to correct the error and execute the instruction to ensure that each instruction can be fully executed.

[0482] The visualization instructions are used to visualize the content summarized and summarized by the large model intelligent agent planning on the client. The client extracts the corresponding components from the visualization component library according to the visualization type, inputs the visualization content, and forms a user-friendly visualization page.

[0483] After all instructions are executed, the client will provide feedback to the network optimization expert agent, including the operation results and any abnormal situations. The network optimization expert agent learns based on the feedback and continuously optimizes the accuracy of operation steps and intent recognition. At the same time, in this process, it will also ensure that the entire process complies with data protection regulations and business compliance requirements to protect user data security. The above instruction types and operation objects are all stored in the instruction library.

[0484] Instruction library: instruction operation type, operable objects, and visual components. The operable objects in the instruction library include all operable objects in the system page. In the early stage, all operable objects are stored in the instruction library through manual entry assisted by automated tools. Each operable object is mapped accurately to the business operation process or function name to facilitate the intelligent agent to identify the intent.

[0485] After the network optimization expert Agent plans the operation steps (these steps are detailed to each specific operation action), it will perform intent recognition based on the specific nouns in the specific operation steps, query the instruction library to obtain the object unique identifier corresponding to the semantics, assemble it into the data format agreed with the client and return it to the client. After the client receives the instruction data, if the instruction is a page instruction, it will find the operated object on the page based on the object unique identifier, and perform the corresponding action on the operated object according to the agent operation type.

[0486] The instruction library contains two major categories of instructions: page instructions and visual instructions.

[0487] Page instructions, instructions for agent users to operate the page. The entire instruction content consists of agent operation type (required) + agent operation object (required) + agent content (optional), among which agent operation type and agent operation object are required items, and agent content is optional.

[0488] The proxy operation types include: 1) Click: Use the mouse to click a specific object on the web system page, including left click, right click, and double click. 2) Text filling: Enter text in an editable control, imitating keyboard behavior. 3) Copy. 4) Cut. 5) Paste. 6) Get text: Get the content of the operated object. 7) Get object: Get the operated object; 8) Scroll: Scroll the control item vertically or horizontally, imitating the behavior of the mouse wheel, so that the covered content that is not in the current visible area is visible. 9) Drag: Imitate the behavior of the user dragging the component. 10) Jump: Imitate the user jumping page, including jumping the current page, jumping to a new tab, and jumping to a new window. 11) Page matching: Match to determine whether the current page is the target page; 12) Screenshot: Screenshot the current screen / browser view interface. 13) Confirmation prompt: It can be used to provide a secondary confirmation prompt before performing some irreversible or high-risk operations as a protection mechanism.

[0489] The proxy operation objects include: 1) All operable component objects on the system page: single selection, multiple selection, input box, date, map, chart, icon, table, button, etc. 2) Page scroll bar: horizontal scroll bar and vertical scroll bar.

[0490] Proxy content: some data used to fill in the proxy operation object and assign values ​​to the proxy operation object. The above proxy operation types, proxy operation objects, and proxy contents can be freely combined according to business needs.

[0491] Visualization instructions, including visualization type + visualization content.

[0492] Visualization types include: 1) Chart instructions, which can be subdivided into: bar chart, line chart, discounted bar chart, pie chart, multi-dimensional chart... 2) Tables, which can be subdivided into: no paging, front-end paging, remote paging; support filtering, filtering types include single selection, multiple selection, text, date, label, etc.;... 3) Single selection. 4) Multiple selection. 5) Button. 6) List. 7) Text. 8) Download. 9) Image...

[0493] Visualization content is generally transmitted in the form of JSON data. The specific data fields and hierarchical relationships are related to the visualization type.

[0494] The specific page composition is as follows Fig.11 As shown, the page rendering effect is as follows Fig.12 shown.

[0495] In one embodiment, the method further comprises pre-training a large network model, specifically comprising:

[0496] Collect historical wireless network quality types, historical wireless network quality cell diagnosis work orders and wireless network optimization expert knowledge. Wireless network optimization expert knowledge includes historical diagnosis subtasks and their chain relationships and historical wireless network quality cell diagnosis results. Historical wireless network quality cell diagnosis results include historical poor root causes of historical wireless network quality cells and historical poor quality treatment solutions.

[0497] Control the first network model, extract the thought chain question and answer prompt words according to the diagnosis process in the historical wireless network poor quality cell diagnosis work order corresponding to the historical wireless network poor quality type, compile the historical diagnosis subtasks and their chain relationships under the thought chain question and answer prompt words, and fine-tune the parameters of the first network model until the historical diagnosis subtasks and their chain relationships are accurately output;

[0498] Distribute the historical diagnosis subtask to the corresponding quality expert Agent, so that the quality expert Agent determines the wireless network API interface that needs to be called according to the historical diagnosis subtask, obtains the historical wireless network quality cell data through the wireless network API interface, determines the existence of abnormal historical wireless network quality cell data and its corresponding historical processing experience information according to the wireless network quality diagnosis expert knowledge, and sends the abnormal historical wireless network quality cell data and its corresponding historical processing experience information to the network optimization expert Agent. The API interface is an application programming interface;

[0499] Receive abnormal historical wireless network poor quality cell data and its corresponding historical processing experience information sent by each poor quality expert Agent, control the second network large model, combine the abnormal historical wireless network poor quality cell data and its corresponding historical processing experience information with the chain relationship and the preset problem template, output the historical wireless network poor quality cell diagnosis result, and fine-tune the parameters of the second network large model until the historical wireless network poor quality cell diagnosis result is accurately output.

[0500] In this embodiment, the network model used is a large language model that has been fine-tuned for the quality difference analysis task, such as Figure 7 As shown, it includes two major functions: decomposing tasks and summarizing results. Both functions are realized based on the thinking chain capability of the large network model. The thinking chain capability is mainly realized through incremental pre-training and SFT (supervised fine-tuning) instruction fine-tuning.

[0501] Incremental pre-training: learn new knowledge from a large-scale "unlabeled" text corpus through training, collect and organize more than ten types of front-line business corpora and data including those from the wireless network air interface side, core network side, base station transmission side, bearer network, IP (Internet Protocol) network, backbone network, complaint handling, road testing, service quality analysis, network maintenance, etc., perform text purification, calculate PPL (perplexity) to filter low-quality text, form a network private professional knowledge data set, perform incremental pre-training on the basic model, and inject network expert knowledge.

[0502] SFT instruction fine-tuning: For the specific task of analyzing voice service poor quality cells, the label data such as voice service diagnosis work orders, voice service fault diagnosis knowledge graphs, and voice service performance indicators KPI, perception indicators KQI (Key Quality Indicators) are cleaned into the professional private domain data (network private professional knowledge data set) of the voice service poor quality cell analysis process, including the answer content of the voice service diagnosis work order, etc., to form a structured voice service poor quality diagnosis process text. Subsequently, these texts are used to use the big model in combination with Prompt Engineering to form the ability of thinking chain, and the sub-goals of the task are decomposed based on intent understanding, so that the big model has the ReAct-Agent (reaction intelligent body) ability to analyze, diagnose, perceive, and make flexible decisions on the problem of voice service poor quality cells.

[0503] Among them, the keywords for the ability to form thought chains (the thought chain prompt words of the first network large model) and the problem templates used for summary by the second network large model can be derived from diagnostic work orders. Diagnostic work orders refer to work orders formed by manual repairs of poor quality cells in the past. The large language model is trained to learn human experience in diagnostic work orders, so that it has the ability to imitate humans to analyze and judge poor quality failures.

[0504] More specifically, the following empirical knowledge is used to design the network model:

[0505] 1. Experience in judging and handling poor quality communities:

[0506] 1. Ideas for dealing with poor quality communities

[0507] First, define relevant indicators and design KPI monitoring templates on the network management system based on the assessment requirements. Screen out problem cells based on the traffic statistics data. Then, classify and locate different problems. Track and record the process. Track and compare the results after the processing is completed. The loop can only be closed after the problem is solved.

[0508] The following six categories are commonly used in the processing of poor quality cells in LTE (Long Term Evolution) networks. In daily work, voice service / data service performance analysis, resource evaluation, network structure, and neighboring cell parameter verification are given priority, as shown in Table 2 below:

[0509] Table 2 Examples of poor quality cell analysis and processing types

[0510]

[0511] 2. Recommendations for the threshold for judging poor quality cells:

[0512] 2.1) RRC (Radio Resource Control) establishment failure: The number of RRC connection establishment failures is greater than 50 times, and the RRC connection establishment success rate is less than 95%;

[0513] 2.2) E-RAB (Evolved Radio Access Bearer) establishment failure: The number of E-RAB establishment failures is greater than 50 times, and the E-RAB establishment success rate is less than 95%;

[0514] 2.3) Drop rate: The number of abnormal releases of UE Context (user equipment context) is greater than 50 times, and the drop rate is greater than 5%;

[0515] 2.4) Handover success rate: The number of cell handover failures is greater than 300 times, and the handover success rate is less than 80%;

[0516] 2.5) Capacity resources: sites involved in key areas

[0517] Condition 1: The average utilization of downlink PRB (Physical Resource Block) is greater than 50%, the average number of valid RRC connections is greater than 30, and the cell downlink busy throughput is greater than 5G (gigabyte, GB);

[0518] Condition 2: The average uplink PRB utilization is greater than 50%, the average number of valid RRC connections is greater than 30, and the cell uplink busy throughput is greater than 1G (gigabyte, GB);

[0519] Condition 3: The maximum number of valid RRC connections is greater than 200;

[0520] Note: Take the average data of the community’s busiest system time for 7 days, and any one of the conditions will suffice.

[0521] 3. Suggestions for checking parameters of poor quality cells are shown in Table 3 below:

[0522] Table 3 Poor quality cell parameter check table

[0523]

[0524] 4. Ideas and steps for locating the problem of poor quality cells:

[0525] There are usually three main reasons for poor quality of problem cells: radio side parameter configuration problems, core network side configuration problems, and channel environment impacts. Therefore, when you encounter a problem cell, you can roughly follow the steps below to troubleshoot.

[0526] 4.1) Confirm whether the indicators of the entire network have deteriorated. If so, it is necessary to check whether there are any network changes and upgrades in the operations and alarms.

[0527] 4.2) If the indicators of some sites deteriorate and affect the indicators of the entire network, it is necessary to find the TOP poor-quality cells (the cells with the worst quality).

[0528] 4.3) Query the TOP cells (the cells with the worst quality) and the TOP occurrence time period (the time period when the quality difference occurs most frequently) with low RRC connection establishment success rate, low ERAB establishment success rate, low handover success rate, and high dropped call rate.

[0529] 4.4) Check the TOP cell site alarms, board status, RRU (Remote Radio Unit) status, cell operation status, and whether the OMC (Operation & Management Center) operation configuration is normal.

[0530] 4.5) Extract CHR log (log file reflecting the problem of the base station) and analyze the channel quality during access and whether the SINR (Signal to Interference plus Noise Ratio) of SRS (Sounding Reference Signal) is poor (weak coverage) and whether there are TOP users (users with the worst SINR).

[0531] 4.6) Conduct targeted standard signaling tracking and interference detection analysis for TOP cell sites. If there are no abnormalities in standard signaling and interference detection, return the one-click log, standard port tracking and interference detection results to the developer for analysis.

[0532] 4.7) Based on the analysis results of all the above data, the first-line network optimization engineers analyze and eliminate each problem one by one, formulate optimization solutions based on expert experience, and then arrange RF optimization personnel and tower workers to implement the optimization solutions until the problem of poor quality TOP cells is finally solved.

[0533] 2. Experience in designing machine learning algorithms:

[0534] 1. Algorithm level

[0535] The positioning of poor-quality cells in mobile communication networks guided by machine learning needs to be combined with the corresponding KQI data of mobile communication network service perception at the algorithm level, comprehensively analyze the perception portrait of the service, and use machine learning as the guidance algorithm to effectively improve the actual accuracy of the portrait, and build a machine learning-oriented positioning of poor-quality cells in mobile communication networks and a comprehensive perception evaluation and analysis system for poor-quality users. The user portrait is formed and iteratively optimized based on quantitative data such as telephone return visits, user usage perception, and network satisfaction, and the user group and poor-quality content are gradually identified. The impact of poor-quality cells on users and the root cause analysis of poor user perception satisfaction are further optimized, and the optimization solution for poor-quality cells is output. Machine learning pushes the analysis results to the network maintenance unit and combines them with network data analysis, including network performance indicators, network complaint data, coverage and road test quality data, equipment failure rate and impact duration, etc., to enable the mapping of basic elements of customer perception and network elements.

[0536] 1.1) Business perception-related feature portrait

[0537] The characteristic portrait of business perception is mainly based on business models and time models. The network services of mobile communications are mainly streaming media, web browsing, and instant messaging services. The streaming media level services are mainly video and games. The poor perception level scenarios are mainly unable to access the Internet, video freezes, game freezes, and slow web page opening. They have the highest frequency of occurrence in current network complaints, accounting for about 90% of the total complaints. Different services often have different corresponding perception factors W. Among them, the time model implements hour-level perception quantitative comprehensive scoring for different stages of the Internet user's business process to accurately describe the specific situation of user perception.

[0538] 1.2) Basic Quantitative Algorithms for Mobile Communication Network Perception Guided by Machine Learning

[0539] By conducting correlation analysis on user sample data through online KQI and online surveys, we use machine learning as the basic algorithm to effectively construct the indicator set, realize user-business perception coupling, promote the initial formation of the portrait model, implement data verification analysis and model tuning, and finally form a complete perception portrait model, realizing quantitative evaluation under customer perception and effective identification of root causes.

[0540] 2. Identify the level of poor quality users

[0541] 2.1) Identify poor quality services

[0542] For business scenarios such as game freezes, video freezes, inability to access the Internet, slow web page opening, etc., perception indicators are determined separately to achieve a quantitative definition of poor quality services. For services at the level of inability to access the Internet, control plane indicators are used to achieve rapid definition; for services at the perception level related to video freezes, corresponding algorithms are introduced manually, such as playback freezes due to insufficient loading data rate in the client buffer, which contains many perception indicators. Guided by logistic regression machine learning, the threshold of the threshold output by the iterative algorithm is calibrated, and business identification is implemented based on the user's perception of the threshold. For example, if the actual number of video freezes is 0.3 times / minute, the actual waiting time for video playback is more than 5 seconds, and the total duration of video freezes accounts for 30%, it can be determined that the user has a poor perception of one use.

[0543] 2.2) User Perception Rating Model

[0544] Combined with the basic model of wireless Internet perception quantification, the actual number of times that users in the existing network experience poor perception per unit time is obtained, and different dimensions of scoring are implemented in combination with different types of poor quality services to obtain the final score of user perception. For example, the following formula:

[0545] S=∑(K1W1+K2W2+K3W3)

[0546] S represents the user perception score; K1 represents the web browsing service score, and W1 represents the web browsing service weight; K2 represents the streaming media service score, and W2 represents the streaming media service weight; K3 represents the instant messaging service score, and W3 represents the instant messaging service weight.

[0547] From the statistical analysis, we know that user perception varies from good to bad, and most of them are maintained in the middle state, that is, the perception is general or good, which is consistent with the normal distribution or chi-square distribution. In the typical distribution situation, the focus is on the probability distribution. The horizontal axis represents the total number of failures of a single user in various scenarios at the hourly granularity, and the vertical axis is the actual proportion of the number of users corresponding to the number. The vertical axis is the one-dimensional actual variable X, which is the probability of the total number of failures at the user hourly granularity. Combined with the curve of the actual density function change of the random variable X probability, it can be understood that the number of failures can be divided into different intervals, and the corresponding interval users have different perception scores. You can set the user score to be projected into 5 different intervals to obtain the perception score of the current network users. For example, set the filter output to output users with a perception score below 3 points as poor quality users.

[0548] 3. Identify poor quality cells

[0549] 3.1) Basic model of cell perception scoring

[0550] Combined with the number of poor-quality services obtained for users, the user's permanent residence cells are gathered, the actual number of poor-quality services for different users in the cell is input, and the actual number of poor perceptions for each user and the total number of services occurring are obtained. This will reveal that the proportion of poor perception single indicators is higher than X (poor-perception users). Statistical analysis is performed on the number of poor-perception users Ki and the total number of users Mi in each cell, and the proportion of poor-perception users is obtained as Zi = Ki / Mi.

[0551] 3.2) The proportion of users with poor perception can be divided into five different ranges

[0552] There are differences in the perception scores of users in each corresponding interval. Based on the perception algorithm, the cell scores are effectively mapped to 5 different intervals to obtain the cell scores, that is, 5 points for the best perception and 0 points for the worst perception.

[0553] 3.3) Identifying poor quality cells

[0554] Based on this model, the cell scores are obtained, and low-scoring cells with perception scores below 3 points are filtered out, which are poor-quality cells.

[0555] 4. Uncover the root causes of poor quality cells and output solutions

[0556] Network quality and user perception complement each other. User perception can be evaluated comprehensively from aspects such as "station availability, stable residence, and high rate". When a cell has problems such as low access, high disconnection rate, and low rate, the cell is considered to be a poor quality cell that affects user perception. Although these poor quality cells can be identified through indicators such as disconnection rate, wireless access success rate, and rate, it is still very difficult for front-line personnel to locate the causes of poor quality cells because of the variety of reasons that affect user perception, which may even be caused by multiple factors. At present, it is mainly based on expert experience to divide network problems into types such as coverage, interference, capacity, and faults. Combined with CM (Configuration Management), PM (Performance Management), MR (Measurement Report) and other data, the root cause of poor quality cells is marked by formulating corresponding root cause algorithms.

[0557] 4.1) Coverage Class Root Cause Location Algorithm

[0558] Wireless network coverage problems are caused by unreasonable network coverage. In general, network coverage problems in macro base station scenarios can be divided into weak coverage, over-coverage and overlapping coverage. The coverage problems such as overlapping coverage and weak coverage are jointly located through indicators such as RSRP (Reference Signal Received Power), TA (Time Advanced), and overlapping coverage rate.

[0559] ① Analysis of root causes of over-coverage

[0560] At present, the scale construction of 4G (fourth generation mobile communication technology) network has become mature, but the coverage of sites in different scenarios such as urban areas, suburbs / county towns, towns / rural areas is different: urban sites are dense, the distance between suburban / county town sites is large, and the planning of towns / rural sites is mostly wide coverage. Therefore, when formulating the root cause algorithm, different TA thresholds need to be set according to different coverage scenarios. The coverage of urban sites is generally planned to be within 1km. Considering the differences in wireless environment in urban areas, the ratio of TA>1km>25% is used as a prerequisite for determining over-coverage in urban areas; while the coverage area of ​​suburban / county town sites is not too far due to the large distance between sites, but it cannot be too close at the same time. The ratio of TA>2.5km>25% is used as a prerequisite for determining over-coverage in suburban / county towns; although the coverage of towns / rural sites is wide, considering the problem of wireless propagation loss, the farther the signal coverage is, the greater the attenuation and the weaker the signal penetration ability, so the coverage distance should not be too wide, and the ratio of TA>3.5km>25% is used as a prerequisite for determining over-coverage in towns / rural areas.

[0561] ②Analysis of root causes of over-coverage

[0562] Too close coverage means that the signal coverage is far below the planned coverage range. For example, the antenna downtilt angle of the poor quality cell is too large, there are obstructions in the coverage direction, and there are hidden faults in the antenna feed system, which leads to too close coverage. Comprehensive judgment can be made through indicators such as TA and RSRP. The accuracy of using a single indicator to judge the root cause of poor quality is low. Through indicators such as TA and MR, combined with the station spacing and average cell coverage distance in different scenarios, root cause positioning algorithms such as weak coverage, overlapping coverage, over-coverage, and too close coverage can be formulated, as shown in Table 4 below:

[0563] Table 4 Coverage problem judgment basis table

[0564]

[0565]

[0566] ③Optimize solution output

[0567] For macro base station weak coverage, over-coverage and other coverage problems, there are generally five solutions, provided that the base station and antenna feed system are working properly and the parameters are set reasonably: adjusting the antenna height, adjusting the antenna downtilt angle and azimuth angle, adjusting the RS (Reference Signal) power, relocating the site, and adding RRU. Among them, the solution of adjusting the antenna downtilt angle and azimuth angle is given priority.

[0568] 4.2) Interference root cause location algorithm

[0569] Wireless network interference problems generally include external interference and internal interference. In the root cause algorithm proposed in this paper, the interference problem is located by using the average interference noise power per PRB, which is an indicator that represents the interference intensity.

[0570] ①Interference outside the system

[0571] Generally referred to as external interference, it is caused by external interference sources. External interference is continuous and persistent. When judging external interference, it is necessary to exclude the situation where interference is increased due to factors such as user use during the day. Generally, the interference index at 03:00 when the business is idle is considered for judgment.

[0572] ②Interference within the system

[0573] Refers to interference caused by co-frequency interference within the system, GPS (Global Positioning System) failure, data configuration errors, etc. The time period for interference within the system is not fixed and is greatly affected by factors such as user behavior. Generally, the interference index for the whole day is considered. The root cause location algorithm for external and internal interference is shown in Table 5 below:

[0574] Table 5 Interference problem location basis table

[0575]

[0576] ③Optimize solution output

[0577] For interference problems, there are generally five solutions, provided that the base station and antenna feed system are working properly and the parameters are set reasonably: check for external interference sources, adjust antenna height, adjust antenna downtilt angle and azimuth angle, and adjust RS power. Among them, the solution of checking for external interference sources is given priority.

[0578] 4.3) Root cause location algorithm for poor quality

[0579] Indicators related to poor wireless network quality can directly characterize network quality. For example, CQI (Channel Quality Indicator) characterizes channel quality, and MCS (Modulation and Coding Scheme) characterizes resource scheduling. In general macro base station scenarios, poor network quality problems are mainly caused by weak coverage, over-coverage, and overlapping coverage. In the root cause algorithm based on expert experience, poor quality problems can be jointly located through indicators such as CQI, MCS, and TA.

[0580] ①Quality difference analysis based on coverage type

[0581] The coverage judgment conditions for urban, suburban / county, and township / rural sites in the poor quality category are consistent with the coverage category analysis process. In the poor quality category root cause location algorithm, CQI, MCS, TA indicators, average cell coverage distance, average station spacing in different scenarios and other indicators are combined to locate the root cause of poor quality cells. Please see Table 6 below:

[0582] Table 6 Coverage type quality difference analysis basis table

[0583]

[0584] ②Optimize solution output

[0585] Solving poor quality problems is similar to solving coverage problems. While ensuring that the base station and antenna feed system are working properly and the parameters are set reasonably, optimization is performed by adjusting the antenna height, antenna downtilt angle and azimuth angle, and RS power.

[0586] 4.4) Capacity-based root cause location algorithm

[0587] High load directly affects various network indicators. In the root cause algorithm based on expert experience, load problems such as unbalanced traffic load between carriers and unbalanced traffic load between sectors are jointly located through indicators that characterize the load, such as PRB utilization and RS power.

[0588] ① Load imbalance analysis

[0589] Under the same RRU, the maximum power of two logical cells with the same bandwidth and different frequencies is the same. At this time, if the difference in their PRB utilization is >20%, it is load imbalance; in the actual situation, under the same physical cell, the maximum power of two logical cells with the same coverage may not be the same to avoid interference and other factors, but under normal circumstances, the difference will not exceed 3dB. Combined with the actual situation, the PRB utilization and RS power are used to jointly locate the load balancing problem between carriers. At the same time, this method is also applicable to the load balancing judgment between sectors.

[0590] Based on indicators such as PRB utilization, CCE (Control Channel Element) utilization, number of users, traffic, RS power, and cell merging, root cause algorithms such as unbalanced traffic load between carriers, unbalanced traffic load between sectors, large number of users, insufficient bandwidth, high load on logical cells in merged state, unbalanced traffic load between license-restricted sectors, and high CCE utilization are formulated, as shown in Table 7 below:

[0591] Table 7 Capacity quality difference analysis basis table

[0592]

[0593]

[0594] ②Optimize solution output

[0595] In view of the high load problem of macro base stations, there are generally five solutions, provided that the base station and antenna feeder system work normally and the parameters are set reasonably: cell expansion, cell splitting and merging, load balancing between carriers, load balancing between stations, and license expansion. Among them, load balancing between carriers and load balancing between stations are preferred.

[0596] 5. Verification of optimization effect

[0597] An application experiment of the root cause location algorithm was carried out in the pilot area to verify the effectiveness of the above algorithm.

[0598] This embodiment uses the definition of 4 / 5G poor quality cells in wireless networks and builds a wireless poor quality cell analysis and processing agent based on a large model multi-agent architecture. This method can be designed based on any network or higher evolution network cell. This solution can be applied in the poor quality analysis assistant product, and the application effect is as follows: Fig.12 shown.

[0599] Embodiment 2:

[0600] like Figure 2 As shown, the present disclosure provides a method for processing poor quality cells in a wireless network, the method is applied to a poor quality expert Agent, and includes:

[0601] S21, receiving the diagnosis subtask distributed by the network optimization expert Agent, where the diagnosis subtask is a diagnosis subtask corresponding to the poor quality expert Agent in a chain relationship of several diagnosis subtasks of poor quality cells of wireless networks compiled by the network optimization expert Agent;

[0602] S22, obtaining a diagnosis sub-result according to the corresponding diagnosis sub-task, and sending the diagnosis sub-result to the network optimization expert agent, so that the network optimization expert agent obtains a diagnosis result of the poor quality cell of the wireless network according to the plurality of diagnosis sub-results sent by the poor quality expert agents and the corresponding chain relationship, and the diagnosis result includes the poor quality root cause of the poor quality cell of the wireless network and the poor quality treatment suggestion;

[0603] Among them, Agent is an intelligent agent, the network optimization expert Agent is an intelligent agent with expert knowledge of wireless network optimization, and the poor quality expert Agent is an intelligent agent with expert knowledge of wireless network poor quality diagnosis.

[0604] In one implementation, receiving the diagnosis subtask distributed by the network optimization expert Agent in S21 specifically includes:

[0605] In response to being one of the fault diagnosis agent, parameter diagnosis agent, coverage diagnosis agent, quality diagnosis agent, interference diagnosis agent, capacity diagnosis agent, switching diagnosis agent, and hidden fault diagnosis agent, receiving the corresponding diagnosis subtask distributed by the network optimization expert agent, the diagnosis subtask corresponding to which is one of the fault diagnosis, parameter diagnosis, coverage diagnosis, quality diagnosis, interference diagnosis, capacity diagnosis, switching diagnosis, and hidden fault diagnosis, and is the first network large model of the network optimization expert agent, and is compiled by calling the thinking chain question and answer prompt words according to the real-time poor wireless network quality type of the poor wireless network quality cell;

[0606] Among them, the first network large model is a large language model used to compile several diagnostic subtasks and their chain relationships after training based on wireless network optimization problems. The thinking chain question and answer prompts are obtained in advance based on the diagnostic process in the historical wireless network poor quality cell diagnosis work order.

[0607] In one embodiment, obtaining a diagnosis sub-result according to the corresponding diagnosis sub-task in S22 and sending the diagnosis sub-result to the network optimization expert Agent specifically includes:

[0608] Determine the wireless network API interface that needs to be called according to the corresponding diagnostic subtask, and obtain real-time wireless network poor quality cell data through the corresponding wireless network API interface. The API interface is an application programming interface;

[0609] According to the expert knowledge of wireless network quality diagnosis, it is determined that there are abnormal real-time wireless network quality cell data and its corresponding historical processing experience information, and the abnormal real-time wireless network quality cell data and its corresponding historical processing experience information are used as the diagnosis sub-result;

[0610] Sending the diagnosis sub-results to the network optimization expert agent, so that the network optimization expert agent obtains the thinking chain question and answer prompt words and their sequence corresponding to the plurality of diagnosis sub-tasks and their chain relationships, and inputs the plurality of diagnosis sub-results into the question template of the second network macro model in sequence, so as to obtain the diagnosis results including the root causes of the poor quality of the wireless network poor quality cell and the poor quality processing suggestions;

[0611] Among them, the second network large model is a large language model used to summarize the root causes of poor quality and poor quality processing suggestions of wireless network poor quality cells after the first network large model is trained. The question templates include questions about obtaining the root causes of poor quality based on a number of abnormal real-time wireless network poor quality cell data and questions about obtaining poor quality processing suggestions based on a number of historical processing experiences.

[0612] In one embodiment, the method further includes assisting in pre-training a large network model, specifically including:

[0613] Receive the historical diagnosis subtask distributed by the network optimization expert agent, where the historical diagnosis subtask is generated after the network optimization expert agent trains the first network model;

[0614] Determine the wireless network API interface that needs to be called according to the historical diagnosis subtask, and obtain historical wireless network poor quality cell data through the wireless network API interface;

[0615] According to the expert knowledge of wireless network poor quality diagnosis, it is judged that there are abnormal historical wireless network poor quality cell data and the corresponding historical processing experience information;

[0616] The abnormal historical wireless network poor quality cell data and the corresponding historical processing experience information are sent to the network optimization expert Agent, so that the network optimization expert Agent trains the second network large model according to the abnormal historical wireless network poor quality cell data and the corresponding historical processing experience information sent by each quality expert Agent.

[0617] Embodiment 3:

[0618] like Figure 3 As shown, the present disclosure provides a device for processing a poor quality cell in a wireless network, wherein the device is a network optimization expert Agent, and the network optimization expert Agent includes:

[0619] The compilation and distribution module 11 is used to compile several diagnostic subtasks of poor quality cells of wireless networks and their chain relationships, and distribute each diagnostic subtask to the corresponding poor quality expert Agent, so that each poor quality expert Agent obtains its own diagnostic subresult according to the corresponding diagnostic subtask, and sends the diagnostic subresult to the network optimization expert Agent;

[0620] The receiving and summarizing module 12 is connected to the compilation and distribution module 11, and is used to receive the diagnosis sub-results sent by each poor quality expert Agent, and obtain the diagnosis result of the poor quality cell of the wireless network according to the plurality of diagnosis sub-results and the corresponding chain relationship, wherein the diagnosis result includes the poor quality root cause of the poor quality cell of the wireless network and the poor quality treatment suggestion;

[0621] Among them, Agent is an intelligent agent, the network optimization expert Agent is an intelligent agent with expert knowledge of wireless network optimization, and the poor quality expert Agent is an intelligent agent with expert knowledge of wireless network poor quality diagnosis.

[0622] In one embodiment, the compilation distribution module 11 specifically includes:

[0623] An intention recognition unit, configured to receive a user question through the first network large model, and recognize the inquiry intention of the user question regarding a cell with poor wireless network quality;

[0624] A type determination unit, connected to the intention identification unit, is used to determine the real-time poor wireless network quality type of the poor wireless network quality cell according to the statistical results of the indicators of the poor wireless network quality cell;

[0625] The compilation unit is connected to the type judgment unit, and is used to compile several diagnostic subtasks and their chain relationships of the wireless network poor quality cell by calling the thinking chain question and answer prompt words according to the real-time wireless network poor quality type through the first network large model, wherein the several diagnostic subtasks include: at least one of fault diagnosis, parameter diagnosis, coverage diagnosis, quality diagnosis, interference diagnosis, capacity diagnosis, switching diagnosis, and hidden fault diagnosis;

[0626] A distribution unit connected to the compilation unit, for distributing each diagnosis subtask to a corresponding quality difference expert agent, wherein the quality difference expert agent includes at least one of a fault diagnosis agent, a parameter diagnosis agent, a coverage diagnosis agent, a quality diagnosis agent, an interference diagnosis agent, a capacity diagnosis agent, a switching diagnosis agent, and a hidden fault diagnosis agent;

[0627] Among them, the first network large model is a large language model used to compile several diagnostic subtasks and their chain relationships after training based on wireless network optimization problems. The thinking chain question and answer prompts are obtained in advance based on the diagnostic process in the historical wireless network poor quality cell diagnosis work order.

[0628] In one embodiment, the receiving and summarizing module 12 includes:

[0629] A summarizing unit is specifically used to obtain the thinking chain question and answer prompt words and their order corresponding to a number of diagnostic subtasks and their chain relationships, and input a number of diagnostic sub-results into the question template of the second network large model in order to obtain diagnostic results including the root cause of the poor quality of the wireless network poor quality cell and the poor quality processing suggestions, wherein the diagnostic sub-results include abnormal real-time wireless network poor quality cell data and its corresponding historical processing experience information;

[0630] an output unit connected to the summarizing unit, configured to output voice and / or text results of the root cause of poor quality of the wireless network poor quality cell and the poor quality treatment suggestion, display the wireless network poor quality cell on a map and mark its poor quality root cause, generate a page instruction for repairing the root cause of poor quality according to the poor quality treatment suggestion, and execute the page instruction automatically or under user triggering to repair the wireless network poor quality cell;

[0631] Among them, the second network large model is a large language model used to summarize the root causes of poor quality and poor quality processing suggestions of wireless network poor quality cells after the first network large model is trained. The question templates include questions about obtaining the root causes of poor quality based on a number of abnormal real-time wireless network poor quality cell data and questions about obtaining poor quality processing suggestions based on a number of historical processing experiences.

[0632] In one embodiment, the network optimization expert Agent further includes a training module, specifically including:

[0633] A collection unit, used to collect historical wireless network poor quality types, historical wireless network poor quality cell diagnosis work orders and wireless network optimization expert knowledge, the wireless network optimization expert knowledge includes historical diagnosis subtasks and their chain relationships and historical wireless network poor quality cell diagnosis results, the historical wireless network poor quality cell diagnosis results include historical poor quality root causes and historical poor quality treatment solutions for historical wireless network poor quality cells;

[0634] The first training unit is connected to the acquisition unit and is used to control the first network large model, extract the thinking chain question and answer prompt words according to the diagnosis process in the historical wireless network poor quality cell diagnosis work order corresponding to the historical wireless network poor quality type, compile the historical diagnosis subtasks and their chain relationships under the thinking chain question and answer prompt words, and fine-tune the parameters of the first network large model until the historical diagnosis subtasks and their chain relationships are accurately output;

[0635] The distribution unit is also connected to the first training unit, and is also used to distribute the historical diagnosis subtask to the corresponding quality expert Agent, so that the quality expert Agent determines the wireless network API interface that needs to be called according to the historical diagnosis subtask, obtains the historical wireless network quality cell data through the wireless network API interface, determines the existence of abnormal historical wireless network quality cell data and its corresponding historical processing experience information according to the wireless network quality diagnosis expert knowledge, and sends the abnormal historical wireless network quality cell data and its corresponding historical processing experience information to the network optimization expert Agent. The API interface is an application programming interface;

[0636] The second training unit is connected to the distribution unit, and is used to receive abnormal historical wireless network poor quality cell data and corresponding historical processing experience information sent by each poor quality expert Agent, control the second network large model, combine the abnormal historical wireless network poor quality cell data and corresponding historical processing experience information with the chain relationship and the preset problem template, output the historical wireless network poor quality cell diagnosis result, and fine-tune the parameters of the second network large model until the historical wireless network poor quality cell diagnosis result is accurately output.

[0637] Embodiment 4:

[0638] like Figure 4 As shown, the present disclosure provides a device for processing poor quality cells in a wireless network, the device is a poor quality expert Agent, and the poor quality expert Agent includes:

[0639] The task receiving module 21 is used to receive the diagnosis subtask distributed by the network optimization expert Agent, where the diagnosis subtask is a diagnosis subtask of a poor quality cell of a wireless network compiled by the network optimization expert Agent and a chain relationship thereof, and a diagnosis subtask corresponding to the poor quality expert Agent;

[0640] The execution task module 22 is connected to the receiving task module 21, and is used to obtain the diagnosis sub-result according to the corresponding diagnosis sub-task, and send the diagnosis sub-result to the network optimization expert agent, so that the network optimization expert agent obtains the diagnosis result of the poor quality cell of the wireless network according to the multiple diagnosis sub-results sent by the poor quality expert agents and the corresponding chain relationship, and the diagnosis result includes the poor quality root cause of the poor quality cell of the wireless network and the poor quality treatment suggestion;

[0641] Among them, Agent is an intelligent agent, the network optimization expert Agent is an intelligent agent with expert knowledge of wireless network optimization, and the poor quality expert Agent is an intelligent agent with expert knowledge of wireless network poor quality diagnosis.

[0642] In one implementation, the task receiving module 21 is specifically configured to:

[0643] In response to being one of the fault diagnosis agent, parameter diagnosis agent, coverage diagnosis agent, quality diagnosis agent, interference diagnosis agent, capacity diagnosis agent, switching diagnosis agent, and hidden fault diagnosis agent, receiving the corresponding diagnosis subtask distributed by the network optimization expert agent, the diagnosis subtask corresponding to which is one of the fault diagnosis, parameter diagnosis, coverage diagnosis, quality diagnosis, interference diagnosis, capacity diagnosis, switching diagnosis, and hidden fault diagnosis, and is the first network large model of the network optimization expert agent, and is compiled by calling the thinking chain question and answer prompt words according to the real-time poor wireless network quality type of the poor wireless network quality cell;

[0644] Among them, the first network large model is a large language model used to compile several diagnostic subtasks and their chain relationships after training based on wireless network optimization problems. The thinking chain question and answer prompts are obtained in advance based on the diagnostic process in the historical wireless network poor quality cell diagnosis work order.

[0645] In one embodiment, executing the task module 22 specifically includes:

[0646] A data acquisition unit, used to determine the wireless network API interface that needs to be called according to the corresponding diagnosis subtask, and obtain real-time wireless network poor quality cell data through the corresponding wireless network API interface, where the API interface is an application programming interface;

[0647] an abnormality judgment unit connected to the data acquisition unit, and configured to judge the presence of abnormal real-time poor-quality wireless network cell data and its corresponding historical processing experience information based on the expert knowledge of poor-quality wireless network diagnosis, and use the abnormal real-time poor-quality wireless network cell data and its corresponding historical processing experience information as the diagnosis sub-result;

[0648] A result sending unit is connected to the abnormality judgment unit and is used to send the diagnosis sub-result to the network optimization expert agent, so that the network optimization expert agent obtains the thinking chain question and answer prompt words and their sequence corresponding to the plurality of diagnosis sub-tasks and their chain relationships, and inputs the plurality of diagnosis sub-results into the question template of the second network large model in sequence, so as to obtain the diagnosis result including the root cause of the poor quality of the wireless network poor quality cell and the poor quality processing suggestion;

[0649] Among them, the second network large model is a large language model used to summarize the root causes of poor quality and poor quality processing suggestions of wireless network poor quality cells after the first network large model is trained. The question templates include questions about obtaining the root causes of poor quality based on a number of abnormal real-time wireless network poor quality cell data and questions about obtaining poor quality processing suggestions based on a number of historical processing experiences.

[0650] In one embodiment, the quality difference expert Agent also assists in pre-training the large network model, specifically including:

[0651] The task receiving module 21 is also used to receive the historical diagnosis subtask distributed by the network optimization expert Agent, where the historical diagnosis subtask is generated after the network optimization expert Agent trains the first network large model;

[0652] The data acquisition unit is further used to determine the wireless network API interface that needs to be called according to the historical diagnosis subtask, and obtain the historical wireless network poor quality cell data through the wireless network API interface;

[0653] The abnormality judgment unit is further used to judge the abnormal historical wireless network poor quality cell data and its corresponding historical processing experience information according to the wireless network poor quality diagnosis expert knowledge;

[0654] The result sending unit is also used to send abnormal historical wireless network poor quality cell data and its corresponding historical processing experience information to the network optimization expert Agent, so that the network optimization expert Agent can train the second network large model according to the abnormal historical wireless network poor quality cell data and its corresponding historical processing experience information sent by each quality expert Agent.

[0655] Embodiment 5:

[0656] like Figure 5 As shown, the present disclosure provides a device for processing a poor quality cell in a wireless network, the device is a system for processing a poor quality cell in a wireless network, and the system includes:

[0657] The network optimization expert Agent as described in Example 3 is used to implement the method for processing a poor quality cell in a wireless network as described in Example 1;

[0658] The several poor quality expert Agents described in Example 4 are connected to the network optimization expert Agent to implement the method for processing poor quality cells in a wireless network described in Example 2.

[0659] Embodiment 6:

[0660] Embodiment 6 of the present disclosure provides a computer-readable storage medium, in which a computer program is stored. When the computer program is executed by a processor, the method for processing a wireless network cell with poor quality as described in Embodiment 1 or 2 is implemented, or the device for processing a wireless network cell with poor quality as described in any one of Embodiments 3-5 is implemented.

[0661] The computer-readable storage medium includes volatile or non-volatile, removable or non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, computer program elements or other data). Computer-readable storage media include, but are not limited to, RAM (Random Access Memory), ROM (Read-Only Memory), EEPROM (Electrically Erasable Programmable read only memory), flash memory or other memory technology, CD-ROM (Compact Disc Read-Only Memory), digital versatile disk (DVD) or other optical disk storage, magnetic cassettes, magnetic tapes, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to store the desired information and can be accessed by a computer.

[0662] In addition, the present disclosure may also provide a computer device, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor runs the computer program stored in the memory, the processor executes the method for processing a poor-quality wireless network cell as described in Embodiment 1 or 2. The computer device may be a device for processing a poor-quality wireless network cell as described in any one of Embodiments 3-5.

[0663] The memory is connected to the processor, the memory may be a flash memory or a read-only memory or other memory, and the processor may be a central processing unit or a single-chip microcomputer.

[0664] Embodiments 1-6 of the present disclosure provide a method, device and medium for processing poor quality cells in a wireless network, and adopt multi-agent collaboration to provide a network optimization auxiliary tool with a high degree of intelligence and a fast processing speed. Through the collaboration of a network optimization expert Agent and several poor quality expert Agents, the network optimization expert Agent compiles diagnostic subtasks and their chain relationships, and the poor quality expert Agent executes the diagnostic subtasks to obtain diagnostic subresults. Finally, the network optimization expert Agent summarizes the various diagnostic subresults according to the chain relationship to obtain the poor quality root causes and poor quality processing suggestions for the poor quality cells in the wireless network. This can be used to assist network optimization engineers in completing network optimization work, and improve network optimization efficiency by completing the diagnosis of each dimension based on multiple agents. The final diagnostic result is obtained based on the multi-dimensional network poor quality diagnostic subresults to improve the accuracy of the diagnosis.

[0665] It is to be understood that the above embodiments are merely exemplary embodiments used to illustrate the principles of the present disclosure, but the present disclosure is not limited thereto. For those of ordinary skill in the art, various modifications and improvements can be made without departing from the spirit and substance of the present disclosure, and these modifications and improvements are also considered to be within the scope of protection of the present disclosure.

Claims

1. A method for processing a poor quality cell in a wireless network, characterized in that: The method is applied to a network optimization expert Agent and includes: Compile several diagnostic subtasks and their chain relationships for poor quality cells in wireless networks, and distribute each diagnostic subtask to the corresponding poor quality expert Agent, so that each poor quality expert Agent obtains its own diagnostic sub-result according to the corresponding diagnostic subtask, and sends the diagnostic sub-result to the network optimization expert Agent; Receive the diagnosis sub-results sent by each poor quality expert Agent, and obtain the diagnosis result of the poor quality cell of the wireless network according to the multiple diagnosis sub-results and the corresponding chain relationship. The diagnosis result includes the poor quality root cause of the poor quality cell of the wireless network and the poor quality treatment suggestion; Among them, Agent is an intelligent agent, the network optimization expert Agent is an intelligent agent with expert knowledge of wireless network optimization, and the poor quality expert Agent is an intelligent agent with expert knowledge of wireless network poor quality diagnosis.

2. The method according to claim 1, characterized in that Compile several diagnostic subtasks and their chain relationships for poor quality cells in wireless networks, and distribute each diagnostic subtask to the corresponding poor quality expert Agent, including: receiving user questions through the first network big model, and identifying the user's intention to inquire about the cell with poor wireless network quality in the question; Determine the real-time poor wireless network quality type of the poor wireless network quality cell according to the statistical result of the indicator of the poor wireless network quality cell; The first network model uses the thinking chain question and answer prompt words to compile several diagnostic subtasks and their chain relationships of the poor quality cells of the wireless network according to the real-time poor quality type of the wireless network, wherein the several diagnostic subtasks include at least one of: fault diagnosis, parameter diagnosis, coverage diagnosis, quality diagnosis, interference diagnosis, capacity diagnosis, switching diagnosis, and hidden fault diagnosis; Distribute each diagnosis subtask to a corresponding quality difference expert agent, the quality difference expert agent includes: at least one of a fault diagnosis agent, a parameter diagnosis agent, a coverage diagnosis agent, a quality diagnosis agent, an interference diagnosis agent, a capacity diagnosis agent, a switching diagnosis agent, and a hidden fault diagnosis agent; Among them, the first network large model is a large language model used to compile several diagnostic subtasks and their chain relationships after training based on wireless network optimization problems. The thinking chain question and answer prompts are obtained in advance based on the diagnostic process in the historical wireless network poor quality cell diagnosis work order.

3. The method according to claim 2, characterized in that The diagnosis result of the poor quality cell of the wireless network is obtained according to a number of diagnosis sub-results and corresponding chain relationships. The diagnosis result includes the poor quality root cause of the poor quality cell of the wireless network and the poor quality treatment suggestion, specifically including: Obtaining the thinking chain question and answer prompt words and their order corresponding to a number of diagnostic subtasks and their chain relationships, and inputting a number of diagnostic subresults into the question template of the second network macromodel in order to obtain diagnostic results including root causes of poor quality of wireless network poor quality cells and suggestions for poor quality processing, wherein the diagnostic subresults include abnormal real-time wireless network poor quality cell data and its corresponding historical processing experience information; Outputting voice and / or text results of the root cause of poor quality of the wireless network poor quality cell and the poor quality treatment suggestion, displaying the wireless network poor quality cell on a map and marking its poor quality root cause, generating a page instruction for repairing the poor quality root cause according to the poor quality treatment suggestion, and automatically or under user triggering executing the page instruction to repair the wireless network poor quality cell; Among them, the second network large model is a large language model used to summarize the root causes of poor quality and poor quality processing suggestions of wireless network poor quality cells after the first network large model is trained. The question templates include questions about obtaining the root causes of poor quality based on a number of abnormal real-time wireless network poor quality cell data and questions about obtaining poor quality processing suggestions based on a number of historical processing experiences.

4. The method according to claim 3, characterized in that The method also includes pre-training a large network model, specifically including: Collect historical wireless network quality types, historical wireless network quality cell diagnosis work orders and wireless network optimization expert knowledge. Wireless network optimization expert knowledge includes historical diagnosis subtasks and their chain relationships and historical wireless network quality cell diagnosis results. Historical wireless network quality cell diagnosis results include historical poor root causes of historical wireless network quality cells and historical poor quality treatment solutions. Control the first network model, extract the thought chain question and answer prompt words according to the diagnosis process in the historical wireless network poor quality cell diagnosis work order corresponding to the historical wireless network poor quality type, compile the historical diagnosis subtasks and their chain relationships under the thought chain question and answer prompt words, and fine-tune the parameters of the first network model until the historical diagnosis subtasks and their chain relationships are accurately output; Distribute the historical diagnosis subtask to the corresponding quality expert Agent, so that the quality expert Agent determines the wireless network API interface that needs to be called according to the historical diagnosis subtask, obtains the historical wireless network quality cell data through the wireless network API interface, determines the existence of abnormal historical wireless network quality cell data and its corresponding historical processing experience information according to the wireless network quality diagnosis expert knowledge, and sends the abnormal historical wireless network quality cell data and its corresponding historical processing experience information to the network optimization expert Agent. The API interface is an application programming interface; Receive abnormal historical wireless network poor quality cell data and its corresponding historical processing experience information sent by each poor quality expert Agent, control the second network large model, combine the abnormal historical wireless network poor quality cell data and its corresponding historical processing experience information with the chain relationship and the preset problem template, output the historical wireless network poor quality cell diagnosis result, and fine-tune the parameters of the second network large model until the historical wireless network poor quality cell diagnosis result is accurately output.

5. A method for processing a poor quality cell in a wireless network, characterized in that: The method is applied to a quality difference expert Agent and includes: Receiving the diagnosis subtask distributed by the network optimization expert Agent, the diagnosis subtask is a number of diagnosis subtasks of poor quality cells of wireless networks compiled by the network optimization expert Agent and the diagnosis subtasks corresponding to the poor quality expert Agent in the chain relationship; Obtaining a diagnosis sub-result according to the corresponding diagnosis sub-task, and sending the diagnosis sub-result to the network optimization expert agent, so that the network optimization expert agent obtains a diagnosis result of a poor quality cell of the wireless network according to a plurality of diagnosis sub-results sent by a plurality of poor quality expert agents and a corresponding chain relationship, wherein the diagnosis result includes a poor quality root cause of the poor quality cell of the wireless network and a poor quality treatment suggestion; Among them, Agent is an intelligent agent, the network optimization expert Agent is an intelligent agent with expert knowledge of wireless network optimization, and the poor quality expert Agent is an intelligent agent with expert knowledge of wireless network poor quality diagnosis.

6. The method according to claim 5, characterized in that Receive the diagnostic subtasks distributed by the network optimization expert agent, including: In response to being one of the fault diagnosis agent, parameter diagnosis agent, coverage diagnosis agent, quality diagnosis agent, interference diagnosis agent, capacity diagnosis agent, switching diagnosis agent, and hidden fault diagnosis agent, receiving the corresponding diagnosis subtask distributed by the network optimization expert agent, the diagnosis subtask corresponding to which is one of the fault diagnosis, parameter diagnosis, coverage diagnosis, quality diagnosis, interference diagnosis, capacity diagnosis, switching diagnosis, and hidden fault diagnosis, and is the first network large model of the network optimization expert agent, and is compiled by calling the thinking chain question and answer prompt words according to the real-time poor wireless network quality type of the poor wireless network quality cell; Among them, the first network large model is a large language model used to compile several diagnostic subtasks and their chain relationships after training based on wireless network optimization problems. The thinking chain question and answer prompts are obtained in advance based on the diagnostic process in the historical wireless network poor quality cell diagnosis work order.

7. The method according to claim 6, characterized in that Obtain the diagnosis sub-results according to the corresponding diagnosis sub-tasks and send them to the network optimization expert agent, including: Determine the wireless network API interface that needs to be called according to the corresponding diagnostic subtask, and obtain real-time wireless network poor quality cell data through the corresponding wireless network API interface. The API interface is an application programming interface; According to the expert knowledge of wireless network quality diagnosis, it is determined that there are abnormal real-time wireless network quality cell data and its corresponding historical processing experience information, and the abnormal real-time wireless network quality cell data and its corresponding historical processing experience information are used as the diagnosis sub-result; Sending the diagnosis sub-results to the network optimization expert agent, so that the network optimization expert agent obtains the thinking chain question and answer prompt words and their sequence corresponding to the plurality of diagnosis sub-tasks and their chain relationships, and inputs the plurality of diagnosis sub-results into the question template of the second network macro model in sequence, so as to obtain the diagnosis results including the root causes of the poor quality of the wireless network poor quality cell and the poor quality processing suggestions; Among them, the second network large model is a large language model used to summarize the root causes of poor quality and poor quality processing suggestions of wireless network poor quality cells after the first network large model is trained. The question templates include questions about obtaining the root causes of poor quality based on a number of abnormal real-time wireless network poor quality cell data and questions about obtaining poor quality processing suggestions based on a number of historical processing experiences.

8. The method according to claim 7, characterized in that The method also includes assisting in pre-training a large network model, specifically including: Receive the historical diagnosis subtask distributed by the network optimization expert agent, where the historical diagnosis subtask is generated after the network optimization expert agent trains the first network model; Determine the wireless network API interface that needs to be called according to the historical diagnosis subtask, and obtain historical wireless network poor quality cell data through the wireless network API interface; According to the expert knowledge of wireless network poor quality diagnosis, it is judged that there are abnormal historical wireless network poor quality cell data and the corresponding historical processing experience information; The abnormal historical wireless network poor quality cell data and the corresponding historical processing experience information are sent to the network optimization expert Agent, so that the network optimization expert Agent trains the second network large model according to the abnormal historical wireless network poor quality cell data and the corresponding historical processing experience information sent by each quality expert Agent.

9. A network optimization expert Agent, characterized in that: The network optimization expert Agent includes: The compilation and distribution module is used to compile several diagnostic subtasks of poor quality cells in the wireless network and their chain relationships, and distribute each diagnostic subtask to the corresponding poor quality expert Agent, so that each poor quality expert Agent obtains its own diagnostic subresult according to the corresponding diagnostic subtask, and sends the diagnostic subresult to the network optimization expert Agent; The receiving and summarizing module is connected to the compilation and distribution module, and is used to receive the diagnosis sub-results sent by each poor quality expert Agent, and obtain the diagnosis result of the poor quality cell of the wireless network according to a number of diagnosis sub-results and the corresponding chain relationship. The diagnosis result includes the poor quality root cause of the poor quality cell of the wireless network and the poor quality treatment suggestion; Among them, Agent is an intelligent agent, the network optimization expert Agent is an intelligent agent with expert knowledge of wireless network optimization, and the poor quality expert Agent is an intelligent agent with expert knowledge of wireless network poor quality diagnosis.

10. A quality difference expert Agent, characterized in that: The quality difference expert Agent includes: The task receiving module is used to receive the diagnosis subtasks distributed by the network optimization expert Agent. The diagnosis subtasks are the diagnosis subtasks of the poor quality cells of the wireless network compiled by the network optimization expert Agent and the diagnosis subtasks corresponding to the poor quality expert Agent in the chain relationship; An execution task module is connected to the receiving task module, and is used to obtain a diagnosis sub-result according to a corresponding diagnosis sub-task, and send the diagnosis sub-result to the network optimization expert agent, so that the network optimization expert agent obtains a diagnosis result of a poor quality cell of the wireless network according to a plurality of diagnosis sub-results sent by a plurality of poor quality expert agents and a corresponding chain relationship, and the diagnosis result includes a poor quality root cause of the poor quality cell of the wireless network and a poor quality treatment suggestion; Among them, Agent is an intelligent agent, the network optimization expert Agent is an intelligent agent with expert knowledge of wireless network optimization, and the poor quality expert Agent is an intelligent agent with expert knowledge of wireless network poor quality diagnosis.

11. A system for processing poor quality cells in a wireless network, characterized in that: The system comprises: A network optimization expert Agent, used to implement the method for processing a poor quality cell in a wireless network as claimed in any one of claims 1 to 4; A plurality of poor quality expert Agents are connected to the network optimization expert Agent and are used to implement the method for processing poor quality cells in a wireless network as described in any one of claims 5-8.

12. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by the processor, the method for processing a poor quality cell in a wireless network as described in any one of claims 1-4 or 5-8 is implemented.

Images