Intelligent operation and maintenance method and device and storage medium

Abstract

The invention relates to the field of cluster operation and maintenance, and discloses an intelligent operation and maintenance method and device and a storage medium, and the method comprises the steps: obtaining a current alarm item, and determining a target alarm item corresponding to the current alarm item according to a pre-constructed device topology network; according to the target alarm item, judging whether a target document sequence set corresponding to the target alarm item exists in each candidate document sequence set or not; in response to the existence of the target document sequence set, performing interpretation processing on the target document sequence set according to the main body quality evaluation value corresponding to the alarm processing user, and displaying the target document sequence set on the target equipment; and in response to the absence of the target document sequence set, generating a target question and answer instruction according to the target alarm item, determining target information corresponding to the target alarm item according to the target question and answer instruction and a preset large language model, and displaying the target information on the target device. According to the invention, the feasibility and adaptability of the operation and maintenance document can be improved, and the alarm processing efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the field of cluster operation and maintenance, and in particular to an intelligent operation and maintenance method, device, and storage medium. Background Art

[0002] With the popularization of HPC (High Performance Computing) scientific computing applications and the acceleration of enterprise digital transformation, HPC clusters exhibit the characteristics of "three highs and two fasts", namely high concurrency, high complexity, high coupling, rapid iteration, and rapid fault diffusion.

[0003] Currently, the traditional operation and maintenance system faces three core contradictions: low-value conversion of massive data, difficult precipitation and reuse of implicit knowledge, and strong manual dependence on fault response. When dealing with equipment failures, it is necessary to manually search for relevant information, which takes a lot of time. Existing solutions rely heavily on operation and maintenance documents, which are time-consuming to enter and difficult to iterate. Traditional knowledge bases only support document or keyword retrieval and cannot provide suggestions for occasional failures without precedents. Traditional knowledge bases are difficult to iterate and are prone to mixing invalid solutions, resulting in low fault handling efficiency.

[0004] In view of this, the present invention is specifically proposed. Summary of the Invention

[0005] In order to solve the above technical problems, the present invention provides an intelligent operation and maintenance method, device, and storage medium, achieving the effects of improving the feasibility and adaptability of operation and maintenance documents and improving the alarm handling efficiency.

[0006] An embodiment of the present invention provides an intelligent operation and maintenance method, which includes:

[0007] Obtain the current alarm item, and determine the target alarm item corresponding to the current alarm item according to the pre-constructed device topology network;

[0008] According to the target alarm item, determine whether there is a target document sequence set corresponding to the target alarm item in each candidate document sequence set;

[0009] In response to the existence of the target document sequence set, interpret the target document sequence set according to the subjective quality evaluation value corresponding to the alarm handling user and display it on the target device;

[0010] In response to the non-existence of the target document sequence set, generate a target question-and-answer instruction according to the target alarm item, determine the target information corresponding to the target alarm item according to the target question-and-answer instruction and a preset large language model, and display the target information on the target device.

[0011] An embodiment of the present invention provides an electronic device, which includes:

[0012] Processor and memory;

[0013] The processor is used to execute the steps of the intelligent operation and maintenance method described in any embodiment by calling the program or instruction stored in the memory.

[0014] An embodiment of the present invention provides a computer-readable storage medium storing a program or instruction, and the program or instruction causes a computer to execute the steps of the intelligent operation and maintenance method described in any embodiment.

[0015] The embodiments of the present invention have the following technical effects:

[0016] By obtaining the current alarm item, and determining the target alarm item corresponding to the current alarm item according to the pre-constructed device topology network to analyze the main alarm item and eliminate interference. Furthermore, according to the target alarm item, it is judged whether there is a target document sequence set corresponding to the target alarm item in each candidate document sequence set, so as to find the operation and maintenance documents that can solve the sequence arrangement of the target alarm item through precise and fuzzy retrieval methods. In response to the existence of the target document sequence set, the target document sequence set is interpreted according to the subject quality evaluation value corresponding to the alarm processing user and displayed on the target device, so as to explain the document in combination with the user's ability when providing the operation and maintenance document, reducing the situation that the user does not understand the document content. In response to the non-existence of the target document sequence set, a target question-and-answer instruction is generated according to the target alarm item, and according to the target question-and-answer instruction and the preset large language model, the target information corresponding to the target alarm item is determined and displayed on the target device, so as to expand information retrieval through the preset large language model and provide reference for the alarm processing user, achieving the effects of improving the feasibility and adaptability of the operation and maintenance document and improving the alarm processing efficiency. Description of the Drawings

[0017] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for use in the description of the specific embodiments or the prior art. Obviously, the following drawings are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0018] Figure 1 It is a flowchart of an intelligent operation and maintenance method provided by an embodiment of the present invention;

[0019] Figure 2 It is a schematic structural diagram of an intelligent operation and maintenance system provided by an embodiment of the present invention;

[0020] Figure 3It is a schematic structural diagram of an electronic device provided by an embodiment of the present invention. Detailed implementation manners

[0021] To make the objectives, technical solutions, and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below. Apparently, the described embodiments are only a part rather than all of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0022] The intelligent operation and maintenance method provided by the embodiment of the present invention is mainly applicable to the situation where maintenance personnel need to handle alarms when devices in a cluster have alarms. The intelligent operation and maintenance method provided by the embodiment of the present invention can be executed by an electronic device.

[0023] Embodiment 1

[0024] Figure 1 It is a flowchart of an intelligent operation and maintenance method provided by an embodiment of the present invention. Refer to Figure 1 and the intelligent operation and maintenance method specifically includes:

[0025] S110. Obtain a current alarm item, and determine a target alarm item corresponding to the current alarm item according to a pre-constructed device topology network.

[0026] Among them, the current alarm item may be an alarm identifier, and each alarm identifier has a corresponding alarm device and at least one alarm condition. The device topology network is a topology network composed of devices that need to be operated and maintained, and may be a topology network constructed according to the location of the cabinet, the cluster where it is located, etc. The target alarm item is the root cause alarm item among the current alarm items, that is, the alarm source.

[0027] Specifically, a current alarm item can be received. The current alarm item can be an alarm item generated by comparing each alarm condition according to the device signals, device logs, etc. of each basic device. Perform root cause detection on one or more current alarm items in the pre-constructed device topology network to determine one or more target alarm items that cause these current alarm items.

[0028] Based on the above example, a device topology network can also be constructed for each device that needs to be maintained, and other related nodes affected by each device node can be constructed. Specifically, it can be:

[0029] Obtain the basic information of each basic device, and construct a device topology network according to each basic information;

[0030] For each device node in the device topology network, determine the upstream node of the device node, and use the upstream node and each downstream node corresponding to the upstream node as the impact service chain corresponding to the device node.

[0031] Among them, the basic devices are each device that needs to be operated and maintained. The basic information includes device name, location cabinet, location cluster, device role, device IP, manufacturer, serial number, maintenance date, etc. The impact service chain is a set of other device nodes that may be affected by the failure of the device node.

[0032] Specifically, obtain the basic information of each basic device, and construct a device topology network according to the association between each basic device based on the basic knowledge of each basic device. For each device node in the device topology network, use the previous node of the device node as the upstream node of the device node, and use the upstream node and all downstream nodes below the upstream node as the impact service chain corresponding to the device node. If the device node does not have a corresponding upstream node, use the device node and all downstream nodes below the device node as the impact service chain corresponding to the device node.

[0033] Based on the above example, the following method can be used to determine the target alarm item corresponding to the current alarm item according to the pre-constructed device topology network:

[0034] Determine the alarm service chain according to the current alarm item;

[0035] According to the alarm service chain and the impact service chains corresponding to each device node in the pre-constructed device topology network, determine at least one target service chain, and use the device nodes corresponding to the target service chain as the target nodes;

[0036] Use the current alarm item corresponding to the target node as the target alarm item.

[0037] Among them, the alarm service chain is the service chain composed of each basic device corresponding to each current alarm item. The target service chain is a service chain that can cover each impact service chain obtained by disassembling the alarm service chain. The target node is the device node corresponding to each target service chain.

[0038] Specifically, determine the basic devices corresponding to each current alarm item, and form an alarm service chain with these basic devices according to the device association relationship. Match and search the alarm service chain in the impact service chains corresponding to each device node in the pre-constructed device topology network, disassemble the alarm service chain, obtain the target service chain that can cover the entire alarm service chain and has the smallest number, and use the device nodes corresponding to each target service chain as each target node. Furthermore, use at least one current alarm item corresponding to each target node as the target alarm item.

[0039] S120. According to the target alarm item, determine whether there is a target document sequence set corresponding to the target alarm item in each candidate document sequence set.

[0040] Among them, the candidate document sequence set is a set composed of multiple documents that need to be executed continuously written by the operation and maintenance personnel. The target document sequence set is a candidate document sequence combination that may be able to solve the target alarm item.

[0041] Specifically, compare the target alarm item with the alarm items that each candidate document sequence set can solve, and determine whether there is a target document sequence set corresponding to the target alarm item in each candidate document sequence set.

[0042] Based on the above example, the following method can be used to determine whether there is a target document sequence set corresponding to the target alarm item according to the target alarm item:

[0043] According to the target alarm item, determine whether there is at least one candidate document sequence set bound to the target alarm item;

[0044] In response to the existence of a candidate document sequence set bound to the target alarm item, use the candidate document sequence set bound to the target alarm item as the target document sequence set;

[0045] In response to the non-existence of a candidate document sequence set bound to the target alarm item, determine the candidate matching degrees corresponding to the target alarm item and each candidate document sequence set according to the alarm trigger condition corresponding to the target alarm item and the alarm trigger conditions corresponding to each candidate document sequence set;

[0046] In response to the existence of at least one candidate matching degree greater than or equal to the matching degree threshold, use the candidate document sequence sets corresponding to the candidate matching degrees greater than or equal to the matching degree threshold as the reference document sequence sets, and determine the target document sequence set corresponding to the target alarm item according to the document quality evaluation values corresponding to each reference document sequence set and the preset reference quantity;

[0047] In response to the non-existence of at least one candidate matching degree greater than or equal to the matching degree threshold, determine that there is no target document sequence set corresponding to the target alarm item in each candidate document sequence set.

[0048] Among them, the alarm trigger condition is one or more trigger conditions corresponding to each alarm item. The candidate matching degree is the vector similarity between the vectorized representation of the alarm trigger condition corresponding to the target alarm item and the vectorized representation of the alarm trigger condition corresponding to each candidate document sequence set. The matching degree threshold is a pre-set matching degree value used to determine whether the candidate matching degree meets the preliminary matching requirements. The reference document sequence set is a candidate document sequence set corresponding to each candidate matching degree that is greater than or equal to the matching degree threshold. The document quality assessment value is a comprehensive value of each candidate document sequence set evaluated on each quality assessment dimension, which is used to characterize the quality of each candidate document sequence set. The quality assessment dimensions may include readability, feasibility, clarity, standardization, etc. The preset reference quantity is the maximum number of reference document sequence sets that can be provided in advance.

[0049] Specifically, for each candidate document sequence set, it is determined whether the candidate alarm item bound to the candidate document sequence set is consistent with the target alarm item. If there is at least one consistent, it is determined that there is at least one candidate document sequence set bound to the target alarm item, otherwise, there is no at least one candidate document sequence set bound to the target alarm item. If there is a candidate document sequence set bound to the target alarm item, it means that there is a completely matched alarm item, and the candidate document sequence set bound to the target alarm item can be used as the target document sequence set. If there is no candidate document sequence set bound to the target alarm item, it means that the target alarm item cannot be accurately matched. Therefore, a fuzzy matching method is used to vectorize the alarm trigger conditions corresponding to the target alarm item and the alarm trigger conditions corresponding to each candidate document sequence set, and the vector similarity corresponding to the vectorized representation of the alarm trigger condition corresponding to the target alarm item and the vectorized representation of the alarm trigger condition corresponding to each candidate document sequence set is calculated as the candidate matching degree corresponding to the target alarm item and each candidate document sequence set. Further, the candidate document sequence set is screened by the candidate matching degree. If there is at least one candidate matching degree greater than or equal to the matching degree threshold, the candidate document sequence set corresponding to each candidate matching degree greater than or equal to the matching degree threshold is used as the reference document sequence set, and each reference document sequence set is processed according to the preset reference quantity limit. If the number of reference document sequence sets is greater than the preset reference quantity, they are sorted from large to small according to the document quality evaluation value, and each reference document sequence set within the preset reference quantity is determined as the target document sequence set corresponding to the target alarm item. If there is not at least one candidate matching degree greater than or equal to the matching degree threshold, it is determined that there is no target document sequence set corresponding to the target alarm item in each candidate document sequence set.

[0050] Based on the above examples, the corresponding document quality evaluation values and the preset reference quantity of each reference document sequence set can be used to determine the target document sequence set corresponding to the target alarm item in the following manner:

[0051] In response to the number of sets of the reference document sequence set being less than or equal to the preset reference quantity, each reference document sequence set is used as the target document sequence set corresponding to the target alarm item, and each target document sequence set is sorted in descending order according to the corresponding document quality evaluation value;

[0052] In response to the number of sets of the reference document sequence set being greater than the preset reference quantity, each reference document sequence set is sorted in descending order according to the corresponding document quality evaluation value. If there are at least two reference document sequence sets with the same corresponding document quality evaluation value, the at least two reference document sequence sets are sorted in descending order according to the corresponding main body quality evaluation value; The target document sequence set corresponding to the target alarm item is determined by selecting each sorted reference document sequence set according to the preset reference quantity.

[0053] Among them, the main body quality evaluation value is determined according to the document quality evaluation values of each candidate document sequence set uploaded by the target user corresponding to the reference document sequence set. The main body quality evaluation value is a comprehensive value used to evaluate the quality of each candidate document sequence set uploaded by the operation and maintenance personnel who upload the reference document sequence set.

[0054] Specifically, if the number of sets of the reference document sequence set is less than or equal to the preset reference quantity, it means that each reference document sequence set meets the requirements. Therefore, each reference document sequence set is used as the target document sequence set corresponding to the target alarm item. Furthermore, each target document sequence set is sorted in descending order according to the corresponding document quality evaluation value, so as to facilitate subsequent display according to the sorting, enabling the operation and maintenance personnel to view in descending order of the document quality evaluation value. If the number of sets of the reference document sequence set is greater than the preset reference quantity, each reference document sequence set is sorted in descending order according to the corresponding document quality evaluation value. During the sorting process, if there are at least two reference document sequence sets with the same corresponding document quality evaluation value, the reference document sequence sets with the same document quality evaluation value are sorted in descending order according to the corresponding main body quality evaluation value of the uploading operation and maintenance personnel. Furthermore, according to the preset reference quantity, each sorted reference document sequence set is selected, and the reference document sequence sets within the preset reference quantity are determined as the target document sequence set corresponding to the target alarm item.

[0055] Based on the above examples, it is also possible to perform quality analysis and scoring on the initial document sequence set uploaded by the operation and maintenance personnel, that is, the target users, and reasonably evaluate the capabilities of each target user to determine the corresponding main body quality evaluation value of the target user. Specifically, it can be:

[0056] Receive the initial document sequence set uploaded by the target user and the initial alarm items corresponding to the initial document sequence set, perform structured processing on the initial document sequence set, and update the initial document sequence set;

[0057] According to the evaluation rules corresponding to each quality evaluation dimension, determine the dimension evaluation value of the initial document sequence set in each quality evaluation dimension;

[0058] According to the dimension evaluation values, determine the document quality evaluation value corresponding to the initial document sequence set;

[0059] In response to the document quality evaluation value corresponding to the initial document sequence set being less than the preset evaluation value, generate document modification information, feedback the document modification information and the initial document sequence set to the target user, and when receiving the updated initial document sequence set of the target user, return to execute the step of determining the dimension evaluation value of the initial document sequence set in each quality evaluation dimension according to the evaluation rules corresponding to each quality evaluation dimension;

[0060] In response to the document quality evaluation value corresponding to the initial document sequence set being greater than or equal to the preset evaluation value, use the initial document sequence set as the candidate document sequence set, and use the initial alarm item as the candidate alarm item corresponding to the candidate document sequence set;

[0061] For each target user, determine the corresponding main body quality evaluation value of the target user according to the document quality evaluation value corresponding to each candidate document sequence set uploaded by the target user.

[0062] Among them, the target user is the operation and maintenance personnel who upload the initial document sequence set. The initial document sequence set is a set of operation and maintenance documents uploaded by the target user for processing alarm items arranged in the order of use. The initial alarm item is the alarm item that the initial document sequence set can solve. The evaluation rule is a rule set for each quality evaluation dimension, such as adding a preset value when a certain rule is met, etc. The dimension evaluation value is the final value obtained by analyzing according to the evaluation rule corresponding to the quality evaluation dimension. The preset evaluation value is a value preset for determining whether the initial document sequence set can be stored in the library for selection. The document modification information is information that prompts that the initial document sequence set uploaded by the target user cannot be stored in the library and needs to modify the document content.

[0063] Specifically, receive the initial document sequence set uploaded by the target user and the initial alarm items corresponding to the initial document sequence set. Structurally process the initial document sequence set, such as text extraction, table recognition, drawing parsing, etc., and structurally process the recognized content to obtain an updated initial document sequence set. According to the evaluation rules corresponding to each quality evaluation dimension formulated in advance, score the initial document sequence set on each quality evaluation dimension to obtain the dimension evaluation values on each quality evaluation dimension. Furthermore, fuse the dimension evaluation values, such as summation, weighted summation, fusing using a preset formula, etc., to obtain the document quality evaluation value corresponding to the initial document sequence set. If the document quality evaluation value corresponding to the initial document sequence set is less than the preset evaluation value, it indicates that the initial document sequence set does not meet the warehousing conditions. Therefore, generate document modification information and feedback the document modification information and the initial document sequence set to the target user. The target user can make modifications and re-upload the modified and updated initial document sequence set. Therefore, when receiving the updated initial document sequence set from the target user, return to execute the step of determining the dimension evaluation values of the initial document sequence set on each quality evaluation dimension according to the evaluation rules corresponding to each quality evaluation dimension to re-evaluate the updated initial document sequence set. If the document quality evaluation value corresponding to the initial document sequence set is greater than or equal to the preset evaluation value, use the initial document sequence set as the candidate document sequence set and use the initial alarm item as the candidate alarm item corresponding to the candidate document sequence set. For each target user, comprehensively process the corresponding document quality evaluation values of the candidate document sequence sets uploaded by the target user in a preset manner, such as calculating the mean, calculating the median, etc., to obtain the main body quality evaluation value corresponding to the target user.

[0064] Based on the above example, after displaying the target document sequence set on the target device according to the document sequence, it is also possible to update the document quality evaluation value in combination with the selection and feedback of the alarm handling user. Specifically, it can be:

[0065] In response to the candidate alarm item bound to the target document sequence set being the same as the target alarm item, receive the processing result corresponding to the target document sequence set;

[0066] In response to the processing result being successful, update the document quality evaluation value corresponding to the target document sequence set according to the first preset incentive value;

[0067] In response to the processing result being a failure, update the document quality evaluation value corresponding to the target document sequence set according to the second preset incentive value, and if the updated document quality evaluation value is less than the preset evaluation value, delete the target document sequence set.

[0068] Among them, the processing result is the result of whether the target alarm item can be resolved using the target document sequence set, which may include processing success and processing failure. The first preset incentive value is the incentive value used to increase the document quality evaluation value, and the second preset incentive value is the incentive value used to decrease the document quality evaluation value. The absolute values of the first preset incentive value and the second preset incentive value may be the same or different.

[0069] Specifically, when the candidate alarm item bound to the target document sequence set is the same as the target alarm item, that is, the target document sequence set can accurately match the target alarm item, then the processing result corresponding to the target document sequence set can be received. If the processing result is processing success, it indicates that the selected target document sequence set can solve the target alarm item. Therefore, the first preset incentive value can be added to the document quality evaluation value of the target document sequence set as the updated document quality evaluation value corresponding to the target document sequence set. If the processing result is processing failure, it indicates that the selected target document sequence set cannot solve the target alarm item. Therefore, it is necessary to decrease the document quality evaluation value of the target document sequence set to reduce its probability of being selected, that is, the second preset incentive value is subtracted from the document quality evaluation value of the target document sequence set as the updated document quality evaluation value corresponding to the target document sequence set. When the updated document quality evaluation value is less than the preset evaluation value, it indicates that the target document sequence set should no longer be selected. Therefore, the target document sequence set is deleted from the library.

[0070] S130. In response to the existence of a target document sequence set, according to the main body quality evaluation value corresponding to the alarm processing user, interpret the target document sequence set and display it on the target device.

[0071] Among them, the target device is the device used by the alarm processing user.

[0072] Specifically, if there is a target document sequence set, it indicates that these documents need to be provided to the alarm processing user. Therefore, first judge the ability level where the main body quality evaluation value corresponding to the alarm processing user is located, perform corresponding interpretation processing on the target document sequence set according to the ability level, and display the interpreted target document sequence set on the target device for the alarm processing user to view, so as to avoid the problem that the alarm processing user is unable to understand the document content due to insufficient ability.

[0073] Based on the above example, the following method can be used to interpret the target document sequence set according to the main body quality evaluation value corresponding to the alarm processing user:

[0074] Determine the ability level corresponding to the alarm processing user according to the main body quality evaluation value corresponding to the alarm processing user;

[0075] Determine each difficult information in the target document sequence set according to the ability level corresponding to the alarm handling user, and determine the explanatory information corresponding to each difficult information according to each difficult information and a preset large language model;

[0076] Perform an explanatory process on the target document sequence set according to the explanatory information corresponding to each difficult information.

[0077] Among them, the ability level is a level used to evaluate the ability of the alarm handling user in writing and understanding operation and maintenance documents. The difficult information is the information that cannot be understood in the target document sequence set analyzed according to the ability level. The preset large language model is a large language model used to retrieve answers. The explanatory information is the information obtained by the preset large language model to answer the difficult information.

[0078] Specifically, match the main body quality evaluation value corresponding to the alarm handling user according to the score intervals divided by each level to obtain the ability level corresponding to the alarm handling user. Based on the ability level corresponding to the alarm handling user, identify the target document sequence set to obtain each difficult information in the target document sequence set. It can be to use the preset large language model combined with the ability level for identification, or to identify based on preset vocabulary, etc., or to match in the target document sequence set according to the query information of other users with the same ability level, and use the successfully matched query information as the difficult information. Furthermore, input each difficult information into the preset large language model for explanation to obtain the explanatory information corresponding to each difficult information respectively. Use the explanatory information corresponding to each difficult information to perform an explanatory process on the target document sequence set.

[0079] S140. In response to the absence of the target document sequence set, generate a target question-and-answer instruction according to the target alarm item, determine the target information corresponding to the target alarm item according to the target question-and-answer instruction and the preset large language model, and display the target information on the target device.

[0080] Among them, the target question-and-answer instruction is a question instruction used to solve the target alarm item. The target information is the answer information fed back by the preset large language model to the target question-and-answer instruction.

[0081] Specifically, if the target document sequence set does not exist, it means that there is no operation and maintenance document corresponding to the target alarm item in the library. Therefore, the alarm handling personnel need to solve the target alarm item without reference information. Therefore, use the target alarm item to generate a target question-and-answer instruction, input the target question-and-answer instruction into the preset large language model, and obtain the answer information for solving the target alarm item through the preset large language model, that is, the target information, and display the target information on the target device for the alarm handling personnel to refer to.

[0082] Based on the above example, after determining the target alarm item corresponding to the current alarm item, a self-healing script can also be used to process the alarm item simultaneously, so as to introduce document and script processing at the same time and improve the processing efficiency of the alarm item. Specifically, it can be:

[0083] In response to the existence of a target self-healing script corresponding to the target alarm item, execute the self-healing script and determine whether the target alarm item has been processed;

[0084] In response to the target alarm item being processed, cancel the display of the target document sequence set or the target information on the target device.

[0085] Among them, the target self-healing script is an operation and maintenance script that aims to self-heal the target alarm item and can be automatically executed.

[0086] Specifically, if there is a target self-healing script corresponding to the target alarm item, the self-healing script will be automatically executed, and during the execution process, it will be continuously determined whether the target alarm item has been processed, that is, whether the target alarm item has stopped alarming. If the target alarm item has been processed, it means that there is no need for the operation and maintenance personnel (alarm handling user) to manually process it, and thus the target document sequence set or the target information can be cancelled from being displayed on the target device.

[0087] It can be understood that during the execution of the target self-healing script, the target document sequence set or the target information can still be displayed on the target device for the operation and maintenance personnel to view.

[0088] The present invention has the following technical effects: By obtaining the current alarm item and determining the target alarm item corresponding to the current alarm item according to the pre-constructed device topology network, the main alarm items are analyzed to eliminate interference. Furthermore, according to the target alarm item, it is determined whether there is a target document sequence set corresponding to the target alarm item in each candidate document sequence set, so as to find the operation and maintenance documents that can solve the sequence arrangement of the target alarm item through precise and fuzzy retrieval methods. In response to the existence of the target document sequence set, the target document sequence set is interpreted and processed according to the subject quality evaluation value corresponding to the alarm handling user and is displayed on the target device, so as to explain the document in combination with the user's ability when providing the operation and maintenance document, reducing the situation where the user does not understand the content of the document. In response to the non-existence of the target document sequence set, according to the target alarm item, a target Q&A instruction is generated, and according to the target Q&A instruction and the preset large language model, the target information corresponding to the target alarm item is determined and the target information is displayed on the target device, so as to expand information retrieval through the preset large language model and provide reference for the alarm handling user, achieving the effects of improving the feasibility and adaptability of the operation and maintenance document and improving the alarm handling efficiency.

[0089] Embodiment 2

[0090] Figure 2 is a schematic diagram of the structure of an intelligent operation and maintenance system provided by an embodiment of the present invention, such as Figure 2 As shown, it includes infrastructure layer, data collection layer, knowledge processing layer, intelligent application layer and user interaction layer.

[0091] The infrastructure layer is used to build an elastic resource pool based on the cloud-native architecture, including computing clusters, distributed storage and network equipment, and supports resource sharing with existing monitoring platforms. The elastic and scalable hardware resource pool carries the core computing power and storage requirements of the system and can be shared with the monitoring and operation platform. These include: bare metal servers, microservice clusters, object storage, communication equipment, etc.

[0092] The data collection layer is used for global data collection and standardized processing, opening up the connection channel between the physical status of the equipment and the knowledge system. It includes: document parser, log collector, alarm client, operation and maintenance scheduling engine, etc. The document parser uses NLP+OCR technology to parse the unstructured technical documents (initial document sequence set) uploaded by the operation and maintenance personnel in advance, automatically extract knowledge elements, and generate structured knowledge; the log collector collects the device system log in real time; the alarm client is responsible for implementing the standardized conversion interface of multi-source alarms to parse alarm events, generate a unified alarm event stream through standardized protocol conversion, and obtain the target alarm items; the operation and maintenance scheduling engine builds an instruction set warehouse to support the batch distribution of configuration instructions for cross-vendor equipment; this layer is the system's perception nerve, responsible for collecting the equipment's log information, alarm items and text information of the operation and maintenance documents (candidate document sequence set), and transmitting them to the knowledge processing layer for storage so that they can be retrieved when an alarm occurs. Multi-threaded concurrent collection technology is used to achieve second-level data response to ensure the integrity and timeliness of the input data of the knowledge processing layer.

[0093] The knowledge processing layer is used to extract, integrate and dynamically update operation and maintenance knowledge, and build the core carrier of knowledge assets, including: knowledge graph engine, vector database, quality assessment model, etc.

[0094] In the data acquisition layer, the device digital twin warehouse stores the basic information of all basic devices in the cluster, including device name, location cabinet, location cluster, device role, device IP, manufacturer, serial number, maintenance date, etc. The knowledge graph engine uses the basic information of the basic devices in the device warehouse to construct a device topology relationship network (device topology network) based on the existing device digital twin warehouse, and realizes association reasoning such as "the failure of switch A affects the service chain" through query language; adopts a workflow engine to realize streaming orchestration of documents, and binds alarm items (candidate alarm items) to the orchestrated document set (candidate document order set) in the document orchestration module to form an automated document recommendation function; in addition, in the self-healing orchestration module, streaming orchestration is also realized for the operation and maintenance script (self-healing script), and the orchestration result is associated with the alarm information as a template to form an automated operation and maintenance process. The knowledge graph learns the association relationship of alarm self-healing through reasoning, and establishes an operation and maintenance regulation model for alarm items without bound self-healing scripts; the vector database stores the document semantic vectors provided by the document collector in the data acquisition layer, supports thousand-dimensional vector similarity retrieval, and responds to fuzzy requirements such as "network latency optimization plan", or fuzzy requirements for alarm conditions corresponding to different alarm items; the quality evaluation model integrates the document scoring algorithm to calculate the document quality score, including scores for different quality evaluation dimensions such as readability, feasibility, clarity, and standardization, and dynamically adjusts the knowledge node weights of the operation and maintenance documents (candidate document order set). This layer is the intelligent center of the system, achieving a double breakthrough in the accuracy and coverage of knowledge query through a hybrid retrieval mechanism (exact retrieval + fuzzy semantic retrieval), so that regardless of whether there is an exact match for the target document order set, the system can generate operation and maintenance knowledge based on the existing data in the system, that is, for the target document order set with an exact match for the target alarm item, this set is preferentially recommended, and for the target document order set without an exact match, the corresponding target document order set is obtained through fuzzy matching reasoning.

[0095] The intelligent application layer is used to generate decision-making solutions and drive automated operation and maintenance, manage, call, and provide knowledge integration feedback for generative / inferential models, and achieve the closed-loop transformation from data to action. It includes: a generative question-and-answer engine, a model management center, an alarm rule management center, a document management scheduling center, an integral incentive system, etc. The generative question-and-answer engine, as the middleware for outputting operation and maintenance knowledge, connects the knowledge processing layer and the user interaction layer. Operation and maintenance personnel input natural language through the intelligent assistant dialog box provided on the front-end page as parameters for the API (Application Programming Interface) provided by the generative question-and-answer engine. After the engine obtains the user's query statement, it determines the source of operation and maintenance knowledge through a query analyzer, combines and obtains knowledge according to the order of each target document recommended by the knowledge processing layer, fuses it into multiple versions of operation and maintenance knowledge through a result fusion module, scores and ranks the knowledge according to accuracy and relevance, and the knowledge with the highest score is returned as a return value to the front-end dialog box; the model management center is responsible for managing generative models and inferential models and providing model switching and calling functions for the generative question-and-answer engine; the alarm rule management center is mainly responsible for storing alarm trigger conditions and corresponding alarm items. Operation and maintenance personnel can associate alarm items with a set of candidate document orders to achieve the precise document recommendation function mentioned in the knowledge processing layer; the document management scheduling center is used to store the set of candidate document orders uploaded by operation and maintenance personnel for knowledge sources and training corpora; the integral incentive system refers to calling the quality assessment model in the knowledge processing layer to score the document quality assessment value of the set of candidate document orders uploaded by operation and maintenance personnel, generating a subject quality assessment value for operation and maintenance personnel, dividing the ability level through the subject quality assessment value, and using it as a reference for evaluating the operation and maintenance knowledge production ability.

[0096] The user interaction layer provides a naturalized human-machine collaboration interface for the system, improving the efficiency and experience of operation and maintenance operations. It includes: an intelligent assistant, model management, alarm self-healing configuration, document management, an intelligent dashboard, etc. It supports operations such as natural language interaction and configuration between operation and maintenance personnel and models, also provides an alarm self-healing procedure configuration and a document management entry, and provides an intelligent dashboard including but not limited to a document contribution list. The intelligent assistant is a front-end component similar to a customer service dialog box, and the back-end uses the API provided by the generative question-and-answer engine in the intelligent application layer, allowing different models to be selected for communication to achieve the effect of learning while operating and maintaining during the conversation; model management, as the front-end module of the model management center in the intelligent application layer, provides functions such as uploading, configuring, updating, training, and deleting models; the alarm self-healing configuration provides a front-end module for associating alarm items with operation and maintenance scripts into a streaming operation; document management corresponds to the document scheduling center in the intelligent application layer and is responsible for implementing the streaming arrangement of adding, deleting, modifying, querying operation and maintenance documents and associating alarm items on the front-end interaction page; the intelligent dashboard is responsible for displaying functions such as the document score list, the operation and maintenance knowledge production scoreboard of operation and maintenance personnel, and the evaluation of operation and maintenance knowledge production ability.

[0097] The above system has a dual-mode knowledge engine, namely a knowledge graph + vector database, which can achieve seamless connection between precise and fuzzy retrieval. It has a generative decision-making center, combines a preset large language model to drive the generation of automation solutions, and reaches a second-level response speed. Moreover, it can perform multi-source data fusion, align the device signals, documents, and logs in terms of time and space, and eliminate information silos. It also conducts immersive interaction design, reduces the cognitive load through dialogue / visualization multimodal operations, has intelligent parsing capabilities, can automatically extract parameters for unstructured documents, and improves the conversion efficiency. In terms of use, it has a closed-loop feedback mechanism, can reverse-optimize the weights through the document quality evaluation value, form an automated system, and realizes a contribution incentive ecosystem, quantifies the knowledge value (document quality evaluation value and subject quality flatness value), and promotes the output of high-quality documents.

[0098] Embodiment III

[0099] Figure 3 is a schematic structural diagram of an electronic device provided by an embodiment of the present invention. As Figure 3 shown, the electronic device 300 includes one or more processors 301 and a memory 302.

[0100] The processor 301 can be a central processing unit (CPU) or other forms of processing units with data processing capabilities and / or instruction execution capabilities, and can control other components in the electronic device 300 to perform desired functions.

[0101] The memory 302 can include one or more computer program products, and the computer program products can include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory can include, for example, random access memory (RAM) and / or cache memory, etc. The non-volatile memory can include, for example, read-only memory (ROM), hard disk, flash memory, etc. One or more computer program instructions can be stored on the computer-readable storage media, and the processor 301 can run the program instructions to implement the intelligent operation and maintenance method of any embodiment of the present invention described above and / or other desired functions. Various contents such as initial external parameters and thresholds can also be stored in the computer-readable storage media.

[0102] In one example, the electronic device 300 may further include: an input device 303 and an output device 304, and these components are interconnected through a bus system and / or other forms of connection mechanisms (not shown). The input device 303 may include, for example, a keyboard, a mouse, and so on. The output device 304 may output various information to the outside, including warning prompt information, braking force, etc. The output device 304 may include, for example, a display, a speaker, a printer, and a communication network and its connected remote output devices, and so on.

[0103] Of course, for simplicity, Figure 3 only some of the components related to the present invention in the electronic device 300 are shown, and components such as a bus, an input / output interface, and so on are omitted. In addition, according to specific application scenarios, the electronic device 300 may further include any other appropriate components.

[0104] Embodiment 4

[0105] In addition to the above methods and devices, an embodiment of the present invention may also be a computer program product, which includes computer program instructions, and when the computer program instructions are run by a processor, the processor is caused to execute the steps of the intelligent operation and maintenance method provided by any embodiment of the present invention.

[0106] The computer program product may be written in any combination of one or more programming languages to write program code for performing the operations of the embodiments of the present invention. The programming languages include object-oriented programming languages, such as Java, C++, etc., and also include conventional procedural programming languages, such as the "C" language or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, executed as an independent software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0107] In addition, an embodiment of the present invention may also be a computer-readable storage medium, on which computer program instructions are stored, and when the computer program instructions are run by a processor, the processor is caused to execute the steps of the intelligent operation and maintenance method provided by any embodiment of the present invention.

[0108] The computer-readable storage medium may adopt any combination of one or more readable media. The readable media may be a readable signal medium or a readable storage medium. The readable storage medium may include, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples (non-exhaustive list) of the readable storage medium include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0109] It should be noted that the terms used in the present invention are only for describing specific embodiments and do not limit the scope of the present application. As shown in the specification of the present invention, unless the context clearly indicates otherwise, words such as "a", "an", "one", and / or "the" do not specifically refer to the singular and may also include the plural. The term "comprising", "including", or any other variant thereof is intended to cover non-exclusive inclusion, such that a process, method, or device comprising a series of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, or device. Without further limitation, an element defined by the statement "comprising an..." does not exclude the existence of additional identical elements in the process, method, or device comprising the said element.

[0110] It should also be noted that the orientation or positional relationship indicated by terms such as "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc. is based on the orientation or positional relationship shown in the drawings, and is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus should not be construed as a limitation to the present invention. Unless otherwise clearly specified and defined, terms such as "installed", "connected", "coupled" should be understood in a broad sense. For example, it may be a fixed connection, a detachable connection, or an integral connection; it may be a mechanical connection or an electrical connection; it may be directly connected or indirectly connected through an intermediate medium, and it may be the internal communication of two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0111] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some or all of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the technical solutions of the embodiments of the present invention.

Claims

1. An intelligent operation and maintenance method, characterized in that Including: Obtain the current alarm item, and determine the target alarm item corresponding to the current alarm item according to the pre-constructed device topology network; According to the target alarm item, determine whether there is a target document sequence set corresponding to the target alarm item in each candidate document sequence set; In response to the existence of the target document sequence set, interpret the target document sequence set according to the subject quality evaluation value corresponding to the alarm handling user, and display it on the target device; In response to the non-existence of the target document sequence set, generate a target Q&A instruction according to the target alarm item, determine the target information corresponding to the target alarm item according to the target Q&A instruction and the preset large language model, and display the target information on the target device.

2. The method according to claim 1, characterized in that, The interpreting the target document sequence set according to the subject quality evaluation value corresponding to the alarm handling user includes: Determine the ability level corresponding to the alarm handling user according to the subject quality evaluation value corresponding to the alarm handling user; According to the ability level corresponding to the alarm handling user, determine each difficult information in the target document sequence set, and determine the interpretation information corresponding to each difficult information according to each difficult information and the preset large language model; Interpret the target document sequence set according to the interpretation information corresponding to each difficult information.

3. The method according to claim 1, wherein The determining whether there is a target document sequence set corresponding to the target alarm item in each candidate document sequence set according to the target alarm item includes: According to the target alarm item, determine whether there is at least one candidate document sequence set bound to the target alarm item; In response to the existence of a candidate document sequence set bound to the target alarm item, use the candidate document sequence set bound to the target alarm item as the target document sequence set; In response to the non-existence of a candidate document sequence set bound to the target alarm item, determine the candidate matching degrees corresponding to the target alarm item and each candidate document sequence set respectively according to the alarm trigger condition corresponding to the target alarm item and the alarm trigger conditions corresponding to each candidate document sequence set; In response to the existence of at least one candidate matching degree greater than or equal to the matching degree threshold, use the candidate document sequence sets corresponding to the candidate matching degrees greater than or equal to the matching degree threshold as the reference document sequence sets, and determine the target document sequence set corresponding to the target alarm item according to the document quality evaluation values corresponding to each reference document sequence set and the preset reference quantity; In response to the non-existence of at least one candidate matching degree greater than or equal to the matching degree threshold, determine that there is no target document sequence set corresponding to the target alarm item in each candidate document sequence set.

4. The method according to claim 3, characterized in that, The determining the target document sequence set corresponding to the target alarm item according to the document quality evaluation values corresponding to each reference document sequence set and the preset reference quantity includes: In response to the number of sets of reference document sequences being less than or equal to a preset reference number, each set of reference document sequences is used as a target document sequence set corresponding to the target alarm item, and for each target document sequence set, they are sorted in descending order according to the corresponding document quality evaluation value; In response to the number of sets of reference document sequences being greater than the preset reference number, each set of reference document sequences is sorted in descending order according to the corresponding document quality evaluation value. If there are at least two sets of reference document sequences with the same corresponding document quality evaluation value, then the at least two sets of reference document sequences are sorted in descending order according to the corresponding subject quality evaluation value; according to the preset reference number, the sorted sets of reference document sequences are selected to determine the target document sequence set corresponding to the target alarm item; wherein, the subject quality evaluation value is determined according to the document quality evaluation values of each candidate document sequence set uploaded by the target user corresponding to the set of reference document sequences.

5. The method according to claim 1, characterized in that, It further includes: Receiving the initial document sequence set uploaded by the target user and the initial alarm item corresponding to the initial document sequence set, performing structured processing on the initial document sequence set, and updating the initial document sequence set; Determining the dimension evaluation value of the initial document sequence set on each quality evaluation dimension according to the evaluation rules corresponding to each quality evaluation dimension; Determining the document quality evaluation value corresponding to the initial document sequence set according to each dimension evaluation value; In response to the document quality evaluation value corresponding to the initial document sequence set being less than the preset evaluation value, generating document modification information, and feeding back the document modification information and the initial document sequence set to the target user, and when receiving the updated initial document sequence set from the target user, returning to execute the step of determining the dimension evaluation value of the initial document sequence set on each quality evaluation dimension according to the evaluation rules corresponding to each quality evaluation dimension; In response to the document quality evaluation value corresponding to the initial document sequence set being greater than or equal to the preset evaluation value, using the initial document sequence set as a candidate document sequence set, and using the initial alarm item as the candidate alarm item corresponding to the candidate document sequence set; For each target user, determining the subject quality evaluation value corresponding to the target user according to the document quality evaluation values corresponding to each candidate document sequence set uploaded by the target user.

6. The method according to claim 5, wherein After the target document sequence set is displayed on the target device according to the document sequence, it further includes: In response to the candidate alarm item bound to the target document sequence set being the same as the target alarm item, receiving the processing result corresponding to the target document sequence set; In response to the processing result being processing success, updating the document quality evaluation value corresponding to the target document sequence set according to the first preset incentive value; In response to the processing result being a processing failure, update the document quality evaluation value corresponding to the target document sequence set according to the second preset incentive value, and delete the target document sequence set if the updated document quality evaluation value is less than the preset evaluation value.

7. The method according to claim 1, wherein It further includes: Obtain the basic information of each basic device, and construct a device topology network according to the basic information; For each device node in the device topology network, determine the upstream node of the device node, and use the upstream node and each downstream node corresponding to the upstream node as the impact service chain corresponding to the device node; Correspondingly, the determining the target alarm item corresponding to the current alarm item according to the pre-constructed device topology network includes: Determine the alarm service chain according to the current alarm item; According to the alarm service chain and the impact service chains corresponding to each device node in the pre-constructed device topology network, determine at least one target service chain, and use the device nodes corresponding to the target service chain as target nodes; Use the current alarm item corresponding to the target node as the target alarm item.

8. The method according to claim 1, wherein After determining the target alarm item corresponding to the current alarm item, it further includes: In response to the existence of a target self-healing script corresponding to the target alarm item, execute the self-healing script and determine whether the target alarm item has been processed; In response to the target alarm item being processed, cancel the display of the target document sequence set or the target information on the target device.

9. An electronic device, characterized in that, The electronic device includes: A processor and a memory; The processor is configured to execute the steps of the intelligent operation and maintenance method according to any one of claims 1 to 8 by calling the program or instruction stored in the memory.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a program or instruction, and the program or instruction causes a computer to execute the steps of the intelligent operation and maintenance method according to any one of claims 1 to 8.

Images