Intelligent operation and maintenance method and system based on natural language processing technology

Abstract

The invention relates to an intelligent operation and maintenance method and system based on a natural language processing technology. The method comprises the following steps: collecting operation and maintenance related data in real time; analyzing the operation and maintenance related data by using a large language model, and if an analysis result is a fault, comparing the analysis result with data in a historical fault database; if the analysis result is related to the historical fault data, executing a historical solution of the related fault; if the analysis result is irrelevant to the historical fault data, generating a solution according to the analysis result by using a large language model, tracking a scheme execution state, and performing real-time feedback; and in combination with historical fault data and real-time feedback, optimizing and improving the large language model, and updating the large language model. Compared with the prior art, operation and maintenance logs, monitoring alarms and user feedback can be automatically analyzed, rapid recognition and processing of faults are achieved, the operation and maintenance efficiency is improved, the historical fault database is self-updated in real time, and the operation and maintenance capacity can be continuously improved.

Description

Technical Field

[0001] The present invention relates to the field of information technology, and in particular to a smart operation and maintenance management (SOMM) method and system based on natural language processing (NLP) technology. Background Art

[0002] With the continuous development of natural language processing technology, its application scenarios are becoming more and more extensive, especially for scenarios that require high repetition and high investment, such as operation and maintenance scenarios. Traditional operation and maintenance mainly rely on manual intervention, which has high communication and manpower investment costs, and does not greatly improve work efficiency, and is prone to cause irreparable losses. Figure 1 As shown, the main capabilities of each subsystem in the entire operation and maintenance management system in the framework are as follows.

[0003] Monitoring system: Responsible for real-time monitoring of the status of IT infrastructure, including servers, networks, and applications.

[0004] Configuration management: Manage and maintain system configurations to ensure configuration consistency and traceability.

[0005] Deployment systems: Automate the deployment, update, and rollback process of applications and services.

[0006] Alerting system: Sends notifications to the operations team when problems are detected and provides necessary contextual information.

[0007] Log management: Collect, store, and analyze system logs to facilitate troubleshooting and performance monitoring.

[0008] Performance Analysis: Monitor and analyze system performance to identify bottlenecks and optimization opportunities.

[0009] Automation scripts: Automated scripts that execute daily operation and maintenance tasks to improve efficiency and accuracy.

[0010] Security systems: Protect systems from security threats, including access control, vulnerability management, and intrusion detection

[0011] The existing technical solutions have the following shortcomings:

[0012] Insufficient integration: Different operation and maintenance tools are not well integrated, resulting in data silos and inconsistent operations.

[0013] Lack of flexibility: The system may have difficulty adapting to rapidly changing business needs and technology environments.

[0014] Limited automation: Many operations tasks still rely on manual operations, resulting in inefficiency and human errors.

[0015] Incomplete monitoring coverage: The monitoring system may not cover all key components, making it difficult to detect potential problems in a timely manner.

[0016] Alarm flooding: Too many alarm messages may lead to alarm fatigue, causing important alarms to be ignored.

[0017] Difficult fault diagnosis: Lack of effective tools and processes to quickly locate and resolve faults.

[0018] Poor knowledge sharing: The knowledge base is not updated in a timely manner, and new employee training and problem solving are inefficient.

[0019] Inadequate security performance: Security measures may not be perfect and may be vulnerable to external attacks and internal misoperation.

[0020] Chinese patent application CN118365311A proposes an intelligent operation and maintenance management system based on a large language model. The method obtains the natural language questions of the operation and maintenance personnel and / or the current operation and maintenance data of the target application system, and then uses the large language model to understand and analyze them, and calls the operation and maintenance knowledge base to obtain the natural language questions to obtain the answers or operation and maintenance management suggestions, and finally achieves the purpose of intelligent management of the system. However, this method is limited to the fixed operation and maintenance knowledge base for the management of operation and maintenance, and requires operation and maintenance engineers to maintain the data regularly. When faced with fault data that has not appeared in the past, it will lack adaptability and cannot respond. Summary of the invention

[0021] The purpose of the present invention is to provide an intelligent operation and maintenance method and system based on natural language processing technology, which is used to improve operation and maintenance efficiency and automated fault response, and to self-update the historical fault database in real time, so that the operation and maintenance capabilities can be continuously improved.

[0022] The purpose of the present invention can be achieved by the following technical solutions:

[0023] An intelligent operation and maintenance method based on natural language processing technology, the method comprising:

[0024] Collect operation and maintenance related data in real time;

[0025] Use a large language model to analyze operation and maintenance related data. If the analysis result is a fault, compare the analysis result with the data in the historical fault database;

[0026] If the analysis result is related to historical fault data, the historical solution to the related fault is executed;

[0027] If the analysis results are irrelevant to historical fault data, a large language model is used to generate solutions based on the analysis results and track the execution status of the solutions for real-time feedback;

[0028] Combine historical fault data and real-time feedback to optimize and improve the large language model and update the large language model.

[0029] Furthermore, the collection objects of the operation and maintenance related data include server logs, network monitoring, application performance indicators and user feedback.

[0030] Furthermore, during the collection of the operation and maintenance related data, standardized data collection tools and protocols are used to collect the target data to ensure the integrity and accuracy of the data.

[0031] Furthermore, the process of analyzing the data by the large language model includes:

[0032] Preprocess the collected operation and maintenance related data, including text cleaning, stemming, word form restoration, stop word removal, standardization, and processing of missing values;

[0033] Decompose the text content in operation and maintenance related data into words or phrases and perform feature extraction;

[0034] Evaluate the sentiment trends in operation-related data of user feedback types, including satisfaction, anger, and disappointment;

[0035] Identify unusual patterns and trends in operational-related data such as server logs, monitoring data, and application performance metrics;

[0036] Organize features, sentiment tendencies, abnormal patterns and trends, analyze whether it is a fault and output the analysis results.

[0037] Furthermore, the large language model is trained by using a historical fault database.

[0038] Furthermore, the data in the historical fault database includes:

[0039] Historical failure data and corresponding historical solutions;

[0040] New fault data collected in real time during use and the corresponding feasible new fault solutions.

[0041] Furthermore, the solution includes repair steps, optimization suggestions, operation and maintenance task arrangements and / or resource allocation methods.

[0042] Furthermore, the real-time feedback includes the performance status of the system and user feedback after the execution of the solution. The real-time feedback is displayed through a user interaction interface. The functions of the user interaction interface include real-time monitoring of the system status, receiving fault reports and / or performing manual intervention. The interaction methods include web interface, mobile application and / or instant messaging tool. The user interaction interface sends alarms via email, text message and / or application notification.

[0043] Furthermore, the optimization and improvement of the large language model includes:

[0044] If the real-time feedback is positive, the new fault data and the new fault solution are added to the historical fault database, and the large language model is trained using the updated historical fault database;

[0045] If the real-time feedback is negative feedback, the fault solution generated for the new fault data will be deleted, and the solution will be regenerated and executed.

[0046] An intelligent operation and maintenance system based on natural language processing technology, comprising:

[0047] Data collection module, real-time collection of operation and maintenance related data;

[0048] The natural language processing module uses a large language model to analyze operation and maintenance related data;

[0049] Fault diagnosis module, if the analysis result is a fault, the analysis result is compared with the data in the historical fault database;

[0050] The automated scheduling module executes historical solutions if the analysis results are related to historical fault data. If the analysis results are not related to historical fault data, a large language model is used to generate solutions based on the analysis results and track the execution status of the solutions for real-time feedback.

[0051] The user interaction module provides a user interaction interface, monitors the system status in real time, receives fault reports, and performs manual intervention;

[0052] Optimize the learning module, combine historical fault data and real-time feedback, optimize and improve the large language model, and update the large language model.

[0053] Compared with the prior art, the present invention has the following beneficial effects:

[0054] 1. The present invention uses natural language processing technology to quickly parse a large amount of text data, including log files, monitoring alarms and user feedback, thereby improving the speed and accuracy of understanding the system status, collecting operation and maintenance related data, timely discovering and responding to potential faults and problems, reducing system downtime and the risk of business interruption, using a large language model to plan solutions, automatically allocate resources and tasks, reduce dependence on manual intervention, improve the efficiency and reliability of operation and maintenance work, and reduce the cost of operation and maintenance work. At the same time, the large language model has the ability of self-learning and updating, and continuously updates the historical fault database, so that the operation and maintenance capabilities can be continuously improved.

[0055] 2. The present invention provides an intuitive user interaction interface, enabling users to monitor system status, manage tasks and receive alarms in real time, thereby improving the work efficiency of the operation and maintenance team. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 This is the main architecture diagram of the existing operation and maintenance management system;

[0057] Figure 2 is a flow chart of the method of the present invention;

[0058] Figure 3 It is a system structure diagram of the present invention. DETAILED DESCRIPTION

[0059] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. This embodiment is implemented based on the technical solution of the present invention, and provides a detailed implementation method and specific operation process, but the protection scope of the present invention is not limited to the following embodiments.

[0060] Example 1

[0061] This embodiment discloses an intelligent operation and maintenance method based on natural language processing technology, which is used to improve operation and maintenance efficiency and automate fault response. The specific steps are as follows: Figure 2 As shown:

[0062] S1, real-time collection of operation and maintenance related data;

[0063] S2, uses a large language model to analyze operation and maintenance related data;

[0064] S3, if the analysis result is a fault, the analysis result is compared with the data in the historical fault database;

[0065] S4: If the analysis result is related to the historical fault data, the historical solution is executed. If the analysis result is not related to the historical fault data, a large language model is used to generate a solution based on the analysis result and track the execution status of the solution for real-time feedback.

[0066] S5, combines historical fault data and real-time feedback to optimize and improve the large language model and update the large language model.

[0067] The operation and maintenance related data includes the collection of operation and maintenance related data from various IT resources. This includes but is not limited to server logs, network monitoring data, application performance indicators and user feedback. During the collection, standardized data collection tools and protocols such as Syslog, SNMP, API calls, etc. are used to interact with the target system to ensure the integrity and accuracy of the data.

[0068] The use of a large language model to analyze the operation and maintenance data includes:

[0069] Preprocess the collected operation and maintenance related data, including text cleaning, stemming, word form restoration, stop word removal, standardization, and processing of missing values;

[0070] Decompose the text content in operation and maintenance related data into words or phrases and perform feature extraction;

[0071] Evaluate the sentiment trends in operation-related data of user feedback types, including satisfaction, anger, and disappointment;

[0072] Identify unusual patterns and trends in operational-related data such as server logs, monitoring data, and application performance metrics;

[0073] Organize features, sentiment tendencies, abnormal patterns and trends, analyze whether it is a fault and output the analysis results.

[0074] The large language model includes BERT, GPT, etc., which is trained by using a historical fault database.

[0075] The data in the historical fault database includes: historical fault data and corresponding historical solutions, new fault data collected in real time during use and corresponding feasible new fault solutions. During subsequent use, the historical fault database will continuously collect fault data and fault solutions and be updated.

[0076] The solution includes repair steps, optimization suggestions, operation and maintenance task arrangements and / or resource allocation methods.

[0077] Among them, the operation and maintenance task arrangement and resource allocation method are determined by optimization algorithms to minimize the system downtime and impact.

[0078] The real-time feedback includes the performance status of the system and user feedback after the solution is executed, and the real-time feedback is displayed through the user interaction interface.

[0079] The functions of the user interaction interface include real-time monitoring of system status, receiving fault reports and / or performing manual intervention. The interaction methods include web interface, mobile application and / or instant messaging tool. The user interaction interface sends alerts via email, SMS and / or application notification.

[0080] Furthermore, the optimization and improvement of the large language model includes:

[0081] If the real-time feedback is positive, the new fault data and the new fault solution are added to the historical fault database, and the large language model is trained using the updated historical fault database;

[0082] If the real-time feedback is negative feedback, the fault solution generated for the new fault data will be deleted, and the solution will be regenerated and executed.

[0083] The specific usage examples of this method are as follows:

[0084] When the system is running, it collects server log data, network monitoring data, application performance indicator data and user feedback data of the IT infrastructure in real time, and uses the large language model to analyze the operation and maintenance related data. If the analysis result is a fault, the analysis result is compared with the data in the historical fault database. If the analysis result is related to the historical fault data, the historical solution is executed. If the analysis result is not related to the historical fault data, the large language model is used to generate a solution based on the analysis result, and repair, optimization, operation and maintenance tasks and resource allocation are performed according to the solution. The execution status is tracked and real-time feedback is provided through the user interaction interface. Finally, the large language model is optimized and improved based on the historical fault data and real-time feedback, and the large language model is updated. The updated large language model is redeployed to the system, and the new fault data and solution are entered into the historical fault database to complete the update of the historical fault database, followed by the next round of data collection and fault analysis.

[0085] Example 2

[0086] This embodiment discloses an intelligent operation and maintenance system based on natural language processing technology. Figure 3 As shown, including:

[0087] Data collection module 1, real-time collection of operation and maintenance related data;

[0088] Natural language processing module 2 uses a large language model to analyze operation and maintenance related data;

[0089] Fault diagnosis module 3, if the analysis result is a fault, the analysis result is compared with the data in the historical fault database;

[0090] Automated scheduling module 4: If the analysis result is related to the historical fault data, the historical solution is executed; if the analysis result is not related to the historical fault data, a large language model is used to generate a solution based on the analysis result and track the execution status of the solution to provide real-time feedback;

[0091] User interaction module 5, providing a user interaction interface, monitoring the system status in real time, receiving fault reports, and performing manual intervention;

[0092] The optimization learning module 6 optimizes and improves the large language model by combining historical fault data and real-time feedback, and updates the large language model.

[0093] In the data acquisition module 1, key components include:

[0094] Log Collector: Used to collect log files from servers and applications.

[0095] Monitoring agent: Deployed in networks and systems to monitor performance indicators in real time.

[0096] User feedback interface: integrates social media, customer service systems, etc. to collect user feedback.

[0097] In the natural language processing module 2, key components include:

[0098] Tokenizer: Breaks text into words or phrases for further analysis.

[0099] Sentiment Analyzer: Evaluates the emotional tendency in user feedback, such as satisfaction, anger, or disappointment.

[0100] Pattern Recognizer: Identify unusual patterns and trends in log and monitoring data.

[0101] In the fault diagnosis module 3, key components include:

[0102] Fault matching engine: compares the extracted features with historical fault cases.

[0103] Cause Analyzer: Infers the root cause of a failure based on the failure mode and system status.

[0104] Solution Recommender: Provides repair steps or optimization suggestions based on the diagnosis results.

[0105] In the automated scheduling module 4, key components include:

[0106] Resource allocator: Allocate appropriate operation and maintenance resources based on the fault type and urgency.

[0107] Task Scheduler: Create and schedule operational tasks, such as restarting services or deploying updates.

[0108] Status Tracker: Monitors the progress of task execution and provides feedback after task completion.

[0109] In the user interaction module 5, key components include:

[0110] Dashboard: Displays key performance indicators and status information of the system.

[0111] Notification system: Send alerts and updates via email, SMS or app notifications.

[0112] Manual intervention tools: Allow operators to execute specific commands or change system configurations.

[0113] In the optimization learning module 6, key components include:

[0114] Online learning engine: Updates models and strategies in real time to adapt to new data and trends.

[0115] Performance Analyzer: Evaluate system performance metrics to identify potential bottlenecks and areas for improvement.

[0116] Policy adjuster: adjusts resource allocation and task scheduling strategies based on analysis results.

[0117] The specific details of the above modules and components can be understood by referring to the relevant descriptions and effects in Example 1.

[0118] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present invention, and these modifications or replacements should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. An intelligent operation and maintenance method based on natural language processing technology, characterized in that: The method comprises: Collect operation and maintenance related data in real time; Use a large language model to analyze operation and maintenance related data. If the analysis result is a fault, compare the analysis result with the data in the historical fault database; If the analysis result is related to historical fault data, the historical solution to the related fault is executed; If the analysis results are irrelevant to historical fault data, a large language model is used to generate solutions based on the analysis results and track the execution status of the solutions for real-time feedback; Combine historical fault data and real-time feedback to optimize and improve the large language model and update the large language model.

2. According to claim 1, the intelligent operation and maintenance method based on natural language processing technology is characterized in that: The collection objects of the operation and maintenance related data include server logs, network monitoring, application performance indicators and user feedback.

3. According to claim 1, the intelligent operation and maintenance method based on natural language processing technology is characterized in that: During the collection of the operation and maintenance related data, standardized data collection tools and protocols are used to collect the target data to ensure the integrity and accuracy of the data.

4. According to claim 1, the intelligent operation and maintenance method based on natural language processing technology is characterized in that: The process of analyzing data by the large language model includes: Preprocess the collected operation and maintenance related data, including text cleaning, stemming, word form restoration, stop word removal, standardization, and processing of missing values; Decompose the text content in operation and maintenance related data into words or phrases and perform feature extraction; Evaluate the sentiment trends in operation-related data of user feedback types, including satisfaction, anger, and disappointment; Identify unusual patterns and trends in operational-related data such as server logs, monitoring data, and application performance metrics; Organize features, sentiment tendencies, abnormal patterns and trends, analyze whether it is a fault and output the analysis results.

5. The intelligent operation and maintenance method based on natural language processing technology according to claim 1 is characterized in that: The large language model is trained by using a historical fault database.

6. The intelligent operation and maintenance method based on natural language processing technology according to claim 1, characterized in that: The data in the historical fault database includes: Historical failure data and corresponding historical solutions; New fault data collected in real time during use and the corresponding feasible new fault solutions.

7. The intelligent operation and maintenance method based on natural language processing technology according to claim 1 is characterized in that: The solution includes repair steps, optimization suggestions, operation and maintenance task arrangements and / or resource allocation methods.

8. The intelligent operation and maintenance method based on natural language processing technology according to claim 1 is characterized in that: The real-time feedback includes the performance status of the system and user feedback after the solution is executed. The real-time feedback is displayed through a user interaction interface. The functions of the user interaction interface include real-time monitoring of the system status, receiving fault reports and / or performing manual intervention. The interaction methods include a web interface, a mobile application and / or an instant messaging tool. The user interaction interface sends alerts via email, text messages and / or application notifications.

9. The intelligent operation and maintenance method based on natural language processing technology according to claim 1 is characterized in that: The optimization and improvement of the large language model includes: If the real-time feedback is positive, the new fault data and the new fault solution are added to the historical fault database, and the large language model is trained using the updated historical fault database; If the real-time feedback is negative feedback, the fault solution generated for the new fault data will be deleted, and the solution will be regenerated and executed.

10. An intelligent operation and maintenance system based on natural language processing technology, characterized in that: include: Data collection module, real-time collection of operation and maintenance related data; The natural language processing module uses a large language model to analyze operation and maintenance related data; Fault diagnosis module, if the analysis result is a fault, the analysis result is compared with the data in the historical fault database; The automated scheduling module executes historical solutions if the analysis results are related to historical fault data. If the analysis results are not related to historical fault data, a large language model is used to generate solutions based on the analysis results and track the execution status of the solutions for real-time feedback. The user interaction module provides a user interaction interface, monitors the system status in real time, receives fault reports, and performs manual intervention; Optimize the learning module, combine historical fault data and real-time feedback, optimize and improve the large language model, and update the large language model.

Images