Method and apparatus for managing failure of communication network

Abstract

The disclosure provides a fault management method and device of a communication network, and relates to the technical field of artificial intelligence. The method comprises the following steps: acquiring first fault work order data, first complaint work order data and first sudden situation data; inputting the first fault work order data into a preset language model for fault analysis to obtain fault management measures; acquiring real-time fault handling data, and acquiring a network fault report based on the fault handling data and the fault management measures; inputting the first complaint work order data and the first fault work order data into the language model for early warning analysis to obtain complaint early warning information, and acquiring a complaint analysis report based on the complaint early warning information and the fault handling data; inputting the first sudden situation data and the first fault work order data into the language model for management and control analysis to obtain an emergency plan; and performing fault management based on the network fault report, the complaint analysis report and the emergency plan. The disclosure can guarantee smooth operation of the communication network and ensure the effectiveness of management and control.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the technical field of artificial intelligence, infrastructure and Internet technology (IT) support, and particularly relates to a fault management method of a communication network and a device thereof. BACKGROUND

[0002] In the related art, the fault management of a communication network refers to a series of measures and management means taken for various fault conditions that may occur in the communication network. These measures aim to ensure the stable operation of the communication network and timely recovery of services to minimize the impact of faults on users and operators. With the development of network technology and the expansion of network scale, the amount of management methods related to fault management is also rapidly increasing, increasing the difficulty of management and control for maintenance personnel. Relying on the accumulation of maintenance personnel's experience, it is easy to have low efficiency and control failure. Therefore, how to ensure the smooth operation of the communication network, ensure the effectiveness of management and control, and provide data support in the whole process of fault handling for the faults of various communication equipment networks has become one of the important research directions. SUMMARY

[0003] The present disclosure aims to at least partially solve one of the technical problems in the related art. To this end, one object of the present disclosure is to provide a fault management method of a communication network.

[0004] A second object of the present disclosure is to provide a fault management device of a communication network.

[0005] A third object of the present disclosure is to provide an electronic device.

[0006] A fourth object of the present disclosure is to provide a non-transitory computer-readable storage medium.

[0007] A fifth object of the present disclosure is to provide a computer program product.

[0008] To achieve the above objects, an embodiment of the first aspect of the present disclosure provides a fault management method of a communication network, comprising:

[0009] If a fault of the communication network is monitored, real-time first fault ticket data, first complaint ticket data and first burst situation data are acquired, and the first fault ticket data is input into a preset language model for fault analysis to obtain fault management measures;

[0010] Real-time fault handling data is acquired, and a network fault report is obtained based on the fault handling data and the fault management measures;

[0011] The first complaint work order data and the first fault work order data are input into a language model for early warning analysis to obtain complaint early warning information, and a complaint analysis report is obtained based on the complaint early warning information and fault handling data;

[0012] The first emergency situation data and the first fault work order data are input into a language model for management and control analysis to obtain an emergency plan;

[0013] Based on the network fault report, the complaint analysis report, and the emergency plan, fault management and control are performed in multiple scenarios.

[0014] In the embodiments of the present disclosure, by applying the understanding and generation capabilities of large language models for natural language, the process of fault hierarchical management and control is simulated, fault management and control measures and network fault reports are generated, and the process of complaint early warning management and control is simulated by applying the text analysis and question answering capabilities of large language models, and complaint analysis reports are output, which can provide data support for early warning and management of complaints caused by faults. By applying the text analysis and summary generation capabilities of large language models, the process of emergency support is simulated, and emergency plans are output, which can ensure the smooth operation of communication networks and ensure the effectiveness of management and control, and can support multiple communication equipment networks in the fault handling process.

[0015] To achieve the above purpose, a second aspect of the present disclosure provides a fault management and control device for a communication network, comprising:

[0016] The first obtaining module is configured to, if a fault of the communication network is monitored, obtain real-time first fault work order data, first complaint work order data, and first emergency situation data, and input the first fault work order data into a preset language model for fault analysis to obtain fault management and control measures;

[0017] The second obtaining module is configured to obtain real-time fault handling data, and obtain a network fault report based on the fault handling data and the fault management and control measures;

[0018] The third obtaining module is configured to input the first complaint work order data and the first fault work order data into a language model for early warning analysis to obtain complaint early warning information, and obtain a complaint analysis report based on the complaint early warning information and the fault handling data;

[0019] The fourth obtaining module is configured to input the first emergency situation data and the first fault work order data into a language model for management and control analysis to obtain an emergency plan;

[0020] The fault management and control module is configured to perform fault management and control in multiple scenarios based on the network fault report, the complaint analysis report, and the emergency plan.

[0021] To achieve the above purpose, a third aspect of the present disclosure provides an electronic device, comprising:

[0022] at least one processor; and

[0023] a memory communicatively connected with the at least one processor; wherein

[0024] The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method for fault management of a communication network provided in the first aspect of the present disclosure.

[0025] To achieve the above object, the fourth aspect of the present disclosure provides a computer readable storage medium having stored thereon computer instructions, wherein the computer instructions are used to cause a computer to execute the method for fault management of a communication network according to the first aspect of the present disclosure.

[0026] To achieve the above object, the fifth aspect of the present disclosure provides a computer program product comprising a computer program, the computer program being executed by a processor to implement the method for fault management of a communication network according to the first aspect of the present disclosure. BRIEF DESCRIPTION OF DRAWINGS

[0027] Figure 1 is a flowchart of the method for fault management of a communication network according to an embodiment of the present disclosure;

[0028] Figure 2 is a schematic diagram of the method for fault management of a communication network according to an embodiment of the present disclosure;

[0029] Figure 3 is a schematic diagram of the method for fault management of a communication network according to an embodiment of the present disclosure;

[0030] Figure 4 is a schematic diagram of the method for fault management of a communication network according to an embodiment of the present disclosure;

[0031] Figure 5 is a structural block diagram of the device for fault management of a communication network according to an embodiment of the present disclosure;

[0032] Figure 6 is a structural schematic diagram of an electronic device according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0033] Embodiments of the present disclosure are described in detail below with reference to the accompanying drawings, examples of which are shown in the drawings, wherein the same or similar notations represent the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to explain the present disclosure, and cannot be understood as a limitation of the present disclosure.

[0034] For ease of understanding, the terms in the embodiments of the present disclosure are explained as follows:

[0035] Large language model: English name is Large Language Model, abbreviated as LLM. Refers to a deep learning model trained using a large amount of text data, which can generate natural language text or understand the meaning of language text. Large language models can handle a variety of natural language tasks such as text classification, question answering, dialogue, etc., and are an important way to artificial intelligence.

[0036] Telecommunication: A communication network is a system that uses switching equipment, transmission equipment to interconnect geographically dispersed user terminal equipment to realize communication and information exchange. The most basic form of communication is to establish a communication system between points, but this cannot be called a communication network. Only when many communication systems (transmission systems) are combined according to a certain topology through a switching system can it be called a communication network.

[0037] The fault management method and device of the communication network of the embodiments of the present disclosure are described below in conjunction with the accompanying drawings.

[0038] Figure 1 The flowchart of the fault management method of the communication network of one embodiment of the present disclosure is shown in Figure 1 The method comprises the following steps:

[0039] S101, if the communication network is monitored to have a fault, real-time first fault ticket data, first complaint ticket data and first emergency situation data are obtained, and the first fault ticket data is input into a preset language model for fault analysis to obtain fault management measures.

[0040] In the embodiments of the present disclosure, Network Fault refers to abnormal situations that occur during the operation of communication network equipment, systems or components, which are usually manifested as communication interruption, communication signal anomaly, etc. The fault may be due to the failure or damage of equipment or system components, or may be due to environmental factors (such as temperature, humidity, electromagnetic interference, etc.) that cause the performance of equipment or system to decrease or become abnormal.

[0041] In some embodiments, if a preset message instruction is monitored, it is determined that the communication network has a fault.

[0042] Optionally, in the embodiments of the present disclosure, the fault ticket data includes fault key information (such as key time points, scenarios, specific network elements, fault areas where the network fault occurs), fault management information, etc. In the embodiments of the present disclosure, the first fault ticket data is real-time fault ticket data.

[0043] Optionally, in the embodiments of the present disclosure, the complaint work order data includes complaint content, complaint object, complaint time and the like. In the embodiments of the present disclosure, the first complaint work order data is real-time complaint work order data.

[0044] Optionally, in some scenarios, an emergency situation may occur, such as a typhoon, which requires a response to information work. In some embodiments, the emergency situation data includes emergency situation key information (such as key time points of the occurrence of the emergency situation, emergency situation scenarios, fault key information). In the embodiments of the present disclosure, the first emergency situation data is real-time emergency situation data.

[0045] In some embodiments, the first fault work order data is input into a preset language model for fault analysis, and a fault classification result is obtained. The language model generates fault control measures corresponding to the fault classification result based on the fault classification result and the preconfigured control task.

[0046] In some embodiments, data related to fault classification is collected, including historical fault data, pre-set fault classification standards, pre-set fault handling files, and the like. A large language model is used to train the prepared data. The method of transfer learning can be used, and a pre-trained large language model is used as a base model, and fine-tuning is performed based thereon to adapt to specific fault analysis tasks. In some embodiments, the trained model can be applied to actual fault analysis. By inputting the first fault work order data into the model, a fault classification result is obtained. According to the association between the classification result and the control measures, the fault control measures corresponding to the fault classification result are obtained, for example, the control measures can be to allocate different tools, resources, and the like. In some embodiments, the first fault work order data can also be directly input into the model to obtain fault control measures.

[0047] In some embodiments, after obtaining the fault control measures, the fault control measures can be sent to a pre-set application program, terminal device or webpage. In some embodiments, after obtaining the fault control measures, the fault control measures can be displayed, for example, by a display.

[0048] It should be noted that in some embodiments, the method of transfer learning can be used, and a pre-trained large language model is used as a base model, and fine-tuning is performed based thereon, so that the language model can perform fault analysis, early warning analysis, and control analysis.

[0049] S102, real-time fault handling data is obtained, and a network fault report is obtained based on the fault handling data and the fault control measures.

[0050] In some embodiments, the fault control measure includes multiple control steps, and the fault handling progress corresponding to the fault control measure can be obtained based on real-time fault handling data. The real-time fault handling data is accessed through an application programming API interface. When a fault occurs, the interface automatically inputs the data into the language model. In the program, a prompt engineering method is used to set a corresponding network fault report template. The network fault report template is filled based on the fault handling data and the fault control measure to obtain a network fault report. Thus, the fault control measure and the fault control report can be output in real time through the API interface.

[0051] S103, input the first complaint work order data and the first fault work order data into the language model for early warning analysis to obtain complaint early warning information, and obtain a complaint analysis report based on the complaint early warning information and the fault handling data.

[0052] In the embodiments of the present disclosure, a large language model is called to extract text from the first complaint work order data, analyze the complaint centralized problems, and then combine the first fault work order data to make predictions and classifications to obtain complaint early warning information. In some embodiments, a preset complaint analysis report template is filled based on the complaint early warning information and the fault handling data to obtain a complaint analysis report, so that high-risk network fault complaints can be discovered and handled in a timely manner.

[0053] S104, input the first emergency situation data and the first fault work order data into the language model for control analysis to obtain an emergency plan.

[0054] In the embodiments of the present disclosure, a large language model is called to extract text from the first emergency situation data, analyze the emergency situation key information, and then combine the first fault work order data to make predictions and analyses to obtain complaint early warning information.

[0055] S105, based on the network fault report, the complaint analysis report, and the emergency plan, fault control is performed in multiple scenarios.

[0056] Optionally, fault control is performed in a fault grading control scenario based on the network fault report, in a complaint early warning control scenario based on the complaint analysis report, and in an emergency guarantee control scenario based on the emergency plan.

[0057] For example, in the fault grading control scenario, the network fault report is output through an instant messaging software API after being processed by the API. The system operation and maintenance personnel can receive the network fault report through the instant messaging software client to realize fault grading control.

[0058] For example, in the complaint early warning control scene, the first complaint work order data and the first fault work order data are input into the language model for early warning analysis to obtain complaint early warning information, a complaint analysis report is obtained based on the complaint early warning information and the fault handling data, and the complaint analysis report is output through a WEB page. When the system is operated and maintained, the complaint analysis report can be referred to through the WEB page to handle complaints.

[0059] For example, the first emergency data and the first fault work order data are input into the language model through the API for control analysis to obtain an emergency plan, and the emergency plan is output through the WEB page to support the execution of the related plan in an emergency.

[0060] In the embodiments of the present disclosure, by applying the understanding and generation capabilities of large language models for natural language, the process of fault hierarchical control is simulated, fault control measures and network fault reports are generated, and the system has the characteristics of automation. By applying the text analysis and question answering capabilities of large language models, the process of complaint early warning control is simulated, and a complaint analysis report is output. The system can provide data support for the early warning and control of complaints caused by faults. By applying the text analysis and summary generation capabilities of large language models, the process of emergency support is simulated, and an emergency plan is output. The system can ensure the smooth operation of the communication network, ensure the effectiveness of control, and support a variety of communication equipment networks in fault handling. The system can provide data support in the whole process of fault handling.

[0061] Optionally, in the embodiments of the present disclosure, the prepared data is trained using a large language model. The method of transfer learning can be used, a pre-trained large language model is used as a base model, and the base model is fine-tuned to adapt to fault analysis, early warning analysis, control analysis, and other tasks.

[0062] Figure 2 FIG. 1 is a schematic diagram of a fault control method of a communication network according to an embodiment of the present disclosure. Figure 2 As shown in FIG. 1, in some embodiments, data related to fault classification is collected, second fault work order data at a historical time and fault handling specification information are obtained. Optionally, the fault handling specification information can include network fault handling responsibility division, fault classification information, handling principles, and process requirements. The second fault work order data is labeled based on a preset first labeling format to obtain first labeling information. The fault handling specification information is subjected to text preprocessing and knowledge extraction to obtain a structured fault knowledge file. The text embedding mapping is performed based on the fault knowledge file to obtain a text semantic vector. The fault analysis capability of the language model is supervised trained based on the second fault work order data, the first labeling information, and the text semantic vector.

[0063] For example, the first annotation format includes fault key information and fault management information, the fault key information includes key time points, fault scenarios, fault network elements, and fault areas, and the fault management information includes fault classification, processing specialty, and reporting mode.

[0064] As shown in Figure 2 Optionally, in the embodiments of the present disclosure, the fault handling specification information is subjected to text preprocessing and knowledge extraction to obtain a structured fault knowledge file, including using regular rules to match the relationship between the title, subtitle and text paragraphs to improve the accuracy of text search; using optical character recognition (OCR) to convert the text of the charts in the fault handling specification information to improve the completeness of the knowledge base; and performing entity recognition, relationship extraction and event extraction on the fault handling specification information through natural language processing (NLP). The fault handling specification information involves non-generic internal knowledge information such as communication network equipment, network topology relationship, company internal process, professional personnel architecture, etc. Through knowledge extraction, the internal knowledge can be structured to generate a structured fault knowledge file.

[0065] Optionally, in the embodiments of the present disclosure, the text of the fault knowledge file is represented as a series of vectors capable of expressing the semantic of the text in the form of text embedding (Embedding) to obtain a fault specification knowledge base containing text semantic vectors. The text embedding model can use an open source Chinese model m3e-base, which has good retrieval ability for homogeneous and heterogeneous texts, and can help improve the understanding of queries and the indexing of documents, thereby improving the accuracy and efficiency of knowledge base search.

[0066] Deploying an open source large language model and accessing the fault specification knowledge base makes it suitable for knowledge search and question and answer integration in communication network fault handling. In the model pre-training stage, the second fault work order data is annotated based on a preset first annotation format to obtain first annotation information, and the first annotation format is: <fault key information (key time points, fault scenarios, fault network elements, fault areas), fault management information (fault classification, processing specialty, reporting mode)>. Through the above annotation, supervised fine-tuning is performed to adjust the response behavior of the model to adapt to the fault management task. In some embodiments, reinforcement learning can also be performed on the fault analysis ability of the language model, taking the fault classification accuracy of the model as a reward function to improve the accuracy of classification.

[0067] In some implementations, the second fault work order data is divided into a training set and a test set. During the model evaluation phase, the training set is used for training. After pre-training, the test set is input to evaluate the control effect of the optimized model. Based on the second fault work order data, first annotation information, and text semantic vectors in the test set, a language model is invoked to perform fault analysis and obtain candidate fault control measures and candidate network fault reports. The consistency between the candidate fault control measures output by the large language model and the preset reference fault control measures, and the consistency between the candidate network fault reports and the preset reference network fault reports are compared. The fault classification accuracy is used as the judgment standard. If it reaches 80% or above, the model can be put into use; otherwise, the knowledge extraction, vectorization, and pre-training processes are repeated.

[0068] In this embodiment, control tasks and report templates can be pre-configured. First fault work order data and fault handling data can be accessed through the API interface. When a fault occurs, the interface automatically inputs the data into the trained large language model. The corresponding template can be set in the program using the prompt engineering method, and fault control measures and network fault reports can be output in real time through the API interface.

[0069] This disclosure provides data support to facilitate timely and standardized handling of communication network faults.

[0070] Figure 3 This is a schematic diagram of a fault management method for a communication network according to an embodiment of this disclosure, as shown below. Figure 3 As shown, in some implementations, data related to user complaints is collected to obtain second complaint work order data at historical moments. The second fault work order data and the second complaint work order data are labeled based on a preset second labeling format to obtain second labeling information. The warning analysis capability of the language model is then trained in a supervised manner based on the second complaint work order data, the second fault work order data, the second labeling information, and the text semantic vector. For example, the second labeling format includes key fault information, key complaint information, and complaint warning judgment information. Key complaint information includes the complaint time, the number of complaining users, the reason for the complaint, and the complaining network element. The complaint warning judgment information includes the correlation between the fault and the complaint.

[0071] In some implementations, actual complaint data can be fed back into the model to continuously monitor its predictive performance and accuracy, and the model can be optimized and adjusted according to the actual situation to improve the effectiveness of early warning and control.

[0072] In some embodiments, text extraction is performed on the second fault work order data, data labeling is performed in combination with the second complaint work order data, a large language model is used, pre-training of the model is performed through reinforcement learning (RLHF), and the model is adapted for early warning analysis, which can perform a communication network fault complaint early warning task. The second labeling format is: <network fault key information (key time point, fault scenario, fault network element, fault area), user complaint key information (complaint time, complaint user number, complaint reason, complaint network element), complaint early warning judgment (correlation between fault and complaint)>.

[0073] In some embodiments, the second fault work order data is divided into a training set and a test set. In the model evaluation stage, the training set is used for training, and after pre-training is completed, model evaluation is performed. The test set of historical fault and complaint work orders is input, and the optimized model is evaluated for control effect. Based on the second complaint work order data, the second fault work order data, the second labeling information, and the text semantic vector of the test set, the language model is called for early warning analysis to obtain candidate complaint early warning information. The candidate complaint early warning information output by the large language model is compared with the preset reference complaint early warning information, respectively. When the accuracy rate reaches 80% or more, the model can be put into use, otherwise the knowledge extraction, vectorization, and pre-training processes are performed again.

[0074] As shown in Figure 3 , the early warning task and the configuration report template can be preconfigured in the embodiments of the present disclosure. When a batch of complaints occurs, the first complaint work order data and the first fault work order data are input into the trained language model through an API interface for early warning analysis to obtain complaint early warning information. Based on the complaint early warning information and the fault handling data, a complaint analysis report is obtained, and the complaint early warning information and the complaint analysis report are output in real time through the API interface.

[0075] In the embodiments of the present disclosure, for complaints caused by network faults, the complaints are controlled and managed according to the user range and growth rate of the complaints, and the fault nodes causing user complaints are analyzed and associated in real time, which can assist in ensuring the accurate positioning of the complaint reasons and providing data support for quickly restoring the network faults affecting users.

[0076] Figure 4 is a schematic diagram of a fault control method of a communication network according to an embodiment of the present disclosure, as Figure 4As shown, in the embodiments of the present disclosure, second burst situation data at a historical time is obtained. The second fault work order data and the second burst situation data are labeled based on a preset third labeling format to obtain third labeling information. The language model is supervised trained based on the second fault work order data, the second burst situation data, the third labeling information, and a text semantic vector. For example, the third labeling format includes burst situation key information and emergency plan key information. The burst situation key information includes a key time point, a burst situation scene, and fault key information. The emergency plan key information includes an emergency response level, a value object, material scheduling information, and burst situation reporting information.

[0077] Emergency guarantee control refers to guaranteeing the safety and smoothness of a communication network through rapid, efficient, and orderly command and dispatch work under various burst situations. A full-network communication guarantee emergency plan is used to train a large language model so that it can quickly, effectively, and completely output a dispatch scheme and command suggestion to realize full coverage of necessary scenes and support rapid plan starting and business recovery under emergency situations.

[0078] In the model pre-training stage, text extraction and data labeling are performed based on the second fault work order data and the second burst situation data. A large language model is used to pre-train the model through reinforcement learning (RLHF) to make it suitable for the control and analysis of communication network faults. The third labeling format is: <burst situation key information (key time point, burst situation scene, fault key information), emergency plan key information (emergency response level, value object, material scheduling, burst situation reporting)>.

[0079] In some embodiments, the second burst situation data is divided into a training set and a test set. In the model evaluation stage, the training set is used for training. After pre-training, the test set is input to evaluate the control effect of the optimized model. The language model is called for control and analysis based on the second fault work order data, the second burst situation data, the third labeling information, and the text semantic vector of the test set to obtain a candidate emergency plan. Whether the candidate emergency plan output by the large language model is consistent with the preset reference emergency plan is compared respectively. When the accuracy rate reaches 80% or more, the model can be put into use. Otherwise, the processes of knowledge extraction, vectorization, and pre-training are performed again.

[0080] As shown in FIG. 1, Figure 4 In the embodiments of the present disclosure, a control task can be pre-configured. The first burst situation data and the first fault work order data are accessed through an API interface. The interface automatically inputs the data into the trained language model. When the model determines that the emergency control process needs to be started, a real-time emergency plan can be output in real time through the API interface.

[0081] The communication emergency guarantee control process is started in the burst situation in the embodiments of the present disclosure, a rapid reaction mechanism for early warning and guaranteeing emergency communication is established, and the communication network is ensured to be safe and smooth. In the case of network failure, emergency communication command and dispatch work can be carried out quickly, efficiently and orderly.

[0082] Figure 5 is a structural block diagram of a fault control device of a communication network in an embodiment of the present disclosure, as Figure 5 shown, the fault control device 500 of the communication network comprises:

[0083] The first acquisition module 510 is configured to acquire real-time first fault work order data, first complaint work order data and first burst situation data if it is monitored that the communication network has a fault, and input the first fault work order data into a preset language model for fault analysis to acquire fault control measures;

[0084] The second acquisition module 520 is configured to acquire real-time fault handling data, and acquire a network fault report based on the fault handling data and the fault control measures;

[0085] The third acquisition module 530 is configured to input the first complaint work order data and the first fault work order data into a language model for early warning analysis to acquire complaint early warning information, and acquire a complaint analysis report based on the complaint early warning information and the fault handling data;

[0086] The fourth acquisition module 540 is configured to input the first burst situation data and the first fault work order data into a language model for control analysis to acquire an emergency plan;

[0087] The fault control module 550 is configured to perform fault control in multiple scenarios based on the network fault report, the complaint analysis report and the emergency plan.

[0088] In some embodiments, the fault control device 500 of the communication network further comprises a model training module 560 configured to:

[0089] acquire second fault work order data and fault handling specification information at a historical moment, label the second fault work order data based on a preset first labeling format to acquire first labeling information;

[0090] perform text preprocessing and knowledge extraction on the fault handling specification information to obtain a structured fault knowledge file, and perform text embedding mapping based on the fault knowledge file to obtain a text semantic vector;

[0091] supervise training of the language model based on the second fault work order data, the first labeling information and the text semantic vector;

[0092] The first annotation format includes fault key information and fault management information, the fault key information includes a key time point, a fault scene, a fault network element, and a fault area, and the fault management information includes fault classification, processing specialty, and reporting mode.

[0093] In some embodiments, the first acquisition module 510 is further configured to:

[0094] input the first fault work order data into a preset language model to perform fault analysis and obtain a fault classification result;

[0095] invoke the language model to generate fault management measures corresponding to the fault classification result based on the fault classification result and preconfigured management tasks.

[0096] In some embodiments, the model training module 560 is further configured to:

[0097] acquire second complaint work order data at a historical time point;

[0098] annotate the second fault work order data and the second complaint work order data based on a preset second annotation format to obtain second annotation information;

[0099] perform supervised training on the language model based on the second complaint work order data, the second fault work order data, the second annotation information, and a text semantic vector;

[0100] The second annotation format includes fault key information, complaint key information, and complaint early warning judgment information, the complaint key information includes complaint time, complaint user number, complaint reason, and complaint network element, and the complaint early warning judgment information includes a correlation between the fault and the complaint.

[0101] In some embodiments, the model training module 560 is further configured to:

[0102] acquire second emergency situation data at a historical time point;

[0103] annotate the second fault work order data and the second emergency situation data based on a preset third annotation format to obtain third annotation information;

[0104] perform supervised training on the language model based on the second fault work order data, the second emergency situation data, the third annotation information, and a text semantic vector;

[0105] The third annotation format includes emergency situation key information and emergency plan key information, the emergency situation key information includes a key time point, an emergency situation scene, and fault key information, and the emergency plan key information includes an emergency response level, a value object, material scheduling information, and emergency situation reporting information.

[0106] In the embodiments of the present disclosure, by applying the understanding and generation capabilities of large language models for natural language, the process of fault hierarchical management and control is simulated, and fault management measures and network fault reports are generated, which has the characteristics of automation. By applying the text analysis and question and answer capabilities of the large language model, the process of complaint early warning management is simulated, and a complaint analysis report is output, which can provide data support for early warning and management of complaints caused by faults. By applying the text analysis and summary generation capabilities of the large language model, the process of emergency support is simulated, and an emergency plan is output, which can ensure the smooth operation of the communication network and ensure the effectiveness of management and control, and is compatible with various communication equipment network faults, and provides data support in the whole process of fault handling.

[0107] In the technical solutions of the present disclosure, the acquisition, transmission, storage, use, processing, etc. of data comply with relevant provisions of national laws and regulations.

[0108] It should be noted that in the embodiments of the present disclosure, some existing industry solutions, components, models, etc. may be mentioned, which should be considered as exemplary, and the purpose is only to illustrate the feasibility of the technical solution implementation of the present application, but it does not mean that the applicant has or will necessarily use the solution.

[0109] Figure 6 FIG. 1 is a structural schematic diagram of an electronic device according to an embodiment of the present disclosure.

[0110] As shown in FIG. 8, the electronic device 800 includes: Figure 6

[0111] The memory 801 and the processor 802, the bus 803 connecting different components (including the memory 801 and the processor 802), the memory 801 stores a computer program, and when the processor 802 executes the program, the communication network fault management method of the embodiments of the present disclosure is realized.

[0112] The bus 803 represents one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, a graphics acceleration port, a processor, or a local bus using any of a variety of bus structures. For example, these architectures include but are not limited to industry standard architecture (ISA) bus, micro channel architecture (MAC) bus, enhanced ISA bus, video electronics standards association (VESA) local bus, and peripheral component interconnect (PCI) bus.

[0113] The electronic device 800 typically includes a variety of electronic device readable media. These media can be any available media that can be accessed by the electronic device 800, including volatile and non-volatile media, removable and non-removable media.

[0114] ​The storage 806, which can be implemented as a non- volatile storage device such as a magnetic disk drive and / or optical disk drive, is coupled to the bus 803 by a storage interface 807. The storage 806 can include a database implemented with, for example, a Structured Query Language (SQL) database, an object-oriented database, or another type of database. The storage 806 stores data in one or more databases for use by the electronic device 800. Figure 6 Not shown are typically called "hard disk drive" (HDD). Although Figure 6 not shown in the FIG. 1, a magnetic hard disk drive (e.g., a "hard drive"), and / or a solid state drive, can be provided for reading from and writing to non-removable, nonvolatile magnetic media (typically to three-dimensional arrays of magnetic disks), and a disk drive can be provided for reading from and writing to a removable, nonvolatile media (e.g., an optical disk, such as a CD-ROM, a DVD-ROM, or another optical medium). In these instances, each drive can be connected to the bus 803 by one or more data media interfaces. The storage 806 can include one or more program product(s) 808 having sets of instructions 807 executable by the electronic device 800. A set of the instructions 807 can include an operating system (OS), an application, other program modules, and program data, and have been stored in the storage 806 by read only memory, programmable read only memory, erasable programmable read only memory, or flash memory, among others.

[0115] The programs 808 that implement the programs / modules 807 can be stored in the storage 806 and implemented with, for example, an interpreter that executes instructions in the program 808, or with another program or set of instructions carried in the program 808.

[0116] The electronic device 800 can also communicate with one or more external devices 809 such as a keyboard or a pointing device, displays 811, etc.; other devices Figure 6 associated with the electronic device 800; and / or one or more devices in a communications system. Communication, for example, from and to the electronic device 800 can be enabled by an I / O interface 812. In addition, the electronic device 800 can communicate with one or more networks, such as a local area network (LAN), a general

[0117] The processor 802 performs functions of various embodiments by executing program code that is tangibly stored in the memory 801.

[0118] It should be noted that the implementation process and technical principles of the electronic device in this embodiment are referred to the foregoing explanation of the communication network fault management method of the embodiments of the present disclosure, which will not be repeated here.

[0119] In order to realize the above-mentioned embodiments, the present disclosure further proposes a computer readable storage medium.

[0120] The instructions in the computer readable storage medium are executed by the processor of the electronic device, so that the electronic device can execute the foregoing communication network fault management method. Optionally, the computer readable storage medium can be a ROM, a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, and an optical data storage device, etc.

[0121] Other embodiments of the present disclosure will be apparent to those skilled in the art from consideration of the specification and practice of the application disclosed herein. The present disclosure is intended to cover any variations, uses or adaptations of the present disclosure following the general principles thereof and including the general principles thereof disclosed in the specification and examples. The specification and examples are to be regarded as illustrative only, and the true scope and spirit of the present disclosure are indicated by the following claims.

[0122] It should be understood that the present disclosure is not limited to the precise structures described and shown in the drawings, and various modifications and changes can be made without departing from its scope. The scope of the present disclosure is limited only by the appended claims.

Claims

1. A method for failure handling of a communication network, characterized in that, The method comprises the following steps: If a communication network failure is monitored, real-time first failure work order data, first complaint work order data and first emergency situation data are obtained, and the first failure work order data is input into a preset language model for failure analysis to obtain failure control measures; Real-time failure handling data is obtained, and a network failure report is obtained based on the failure handling data and the failure control measures; The first complaint work order data and the first failure work order data are input into the language model for early warning analysis to obtain complaint early warning information, and a complaint analysis report is obtained based on the complaint early warning information and the failure handling data; The first emergency situation data and the first failure work order data are input into the language model for control analysis to obtain an emergency plan; Based on the network failure report, the complaint analysis report and the emergency plan, failure control is performed in multiple scenarios.

2. The method of claim 1, wherein, The training process of the language model comprises the following steps: Second failure work order data and failure handling specification information at a historical time are obtained, the second failure work order data is labeled based on a preset first labeling format to obtain first labeling information; Text preprocessing and knowledge extraction are performed on the failure handling specification information to obtain a structured failure knowledge file, and text semantic vectors are obtained based on text embedding mapping of the failure knowledge file; The language model is supervised trained based on the second failure work order data, the first labeling information and the text semantic vectors. The first labeling format comprises failure key information and failure control information, the failure key information comprises a key time point, a failure scenario, a failure network element and a failure area, and the failure control information comprises failure classification, handling specialty and reporting mode.

3. The method of claim 2, wherein, The language model is trained based on the second complaint work order data, the second failure work order data, the second labeling information and the text semantic vectors. The training process of the language model comprises the following steps: Second complaint work order data at a historical time is obtained; 4. The method of claim 2, wherein, The second failure work order data and the second complaint work order data are labeled based on a preset second labeling format to obtain second labeling information; The language model is supervised trained based on the second complaint work order data, the second failure work order data, the second labeling information and the text semantic vectors. The second labeling format comprises failure key information, complaint key information and complaint early warning judgment information, the complaint key information comprises complaint time, complaint user number, complaint reason and complaint network element, and the complaint early warning judgment information comprises the correlation between failure and complaint. The training process of the language model comprises the following steps: Second emergency situation data at a historical time is obtained; 5. The method of claim 2, wherein, The second failure work order data and the second emergency situation data are labeled based on a preset third labeling format to obtain third labeling information; ​ ​ The language model is supervised trained based on the second fault work order data, the second emergency data, the third annotation information and the text semantic vector. The third annotation format includes emergency key information and emergency plan key information, the emergency key information includes key time point, emergency scene and fault key information, and the emergency plan key information includes emergency response level, on-duty object, material scheduling information and emergency reporting information.

6. A failure management apparatus of a communication network, characterized by comprising: The method comprises: The first acquisition module is configured to acquire real-time first fault work order data, first complaint work order data and first emergency data if a fault of a communication network is monitored, and input the first fault work order data into a preset language model for fault analysis to acquire fault control measures. The second acquisition module is configured to acquire real-time fault handling data, and acquire a network fault report based on the fault handling data and the fault control measures. The third acquisition module is configured to input the first complaint work order data and the first fault work order data into the language model for early warning analysis to acquire complaint early warning information, and acquire a complaint analysis report based on the complaint early warning information and the fault handling data. The fourth acquisition module is configured to input the first emergency data and the first fault work order data into the language model for control analysis to acquire an emergency plan. The fault control module is configured to perform fault control in multiple scenes based on the network fault report, the complaint analysis report and the emergency plan.

7. The apparatus of claim 6, wherein, The model training module is configured to: acquire second fault work order data and fault handling specification information at a historical time, and acquire first annotation information by annotating the second fault work order data based on a preset first annotation format; perform text preprocessing and knowledge extraction on the fault handling specification information to obtain a structured fault knowledge file, and acquire a text semantic vector by performing text embedding mapping based on the fault knowledge file; supervised train the language model based on the second fault work order data, the first annotation information and the text semantic vector; The first annotation format includes fault key information and fault control information, the fault key information includes key time point, fault scene, fault network element and fault area, and the fault control information includes fault classification, handling specialty and reporting mode. 8.An electronic device comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-5. The computer instructions are used to enable the computer to perform the steps of the method according to any one of claims 1-5.

9. A non-transitory computer readable storage medium having computer instructions stored therein, wherein, 10.A computer program product comprising a computer program which, when executed by a processor, implements the method according to any one of claims 1-5. ​

Images