Fault root cause positioning method and device, computer readable storage medium, and equipment

Abstract

This disclosure relates to the field of artificial intelligence technology, and provides a fault root cause localization method, a fault root cause localization device, a computer-readable storage medium, and an electronic device. The fault root cause localization method includes: extracting fault features from fault description information provided by a user regarding a target fault scenario, and identifying user intent from the fault description information; selecting key fault analysis nodes that are associated with the target fault scenario from candidate fault analysis nodes based on the fault features and the user intent; executing fault analysis tasks associated with the key fault analysis nodes to perform fault root cause localization processing on the target fault scenario, and obtaining a root cause localization result for the target fault scenario. This disclosure enables more intelligent and efficient root cause localization processing for complex fault scenarios.

Description

Technical Field

[0001] This disclosure relates to the field of artificial intelligence technology, and in particular to a fault root cause localization method, a fault root cause localization device, a computer-readable storage medium, and an electronic device. Background Technology

[0002] Root cause analysis is a systematic approach used to identify and understand the underlying causes of a problem or event. Its aim is to find the root cause of the problem through in-depth analysis, rather than simply addressing surface symptoms, thereby enabling effective measures to prevent similar problems from recurring.

[0003] In related technologies, root cause localization of faults usually relies on manual sorting and analysis, which requires a lot of time and resources and places high demands on personnel.

[0004] Therefore, there is an urgent need in this field to develop a new method and apparatus for locating the root cause of a fault.

[0005] It should be noted that the information disclosed in the background section above is only used to enhance the understanding of the background of this disclosure. Summary of the Invention

[0006] The purpose of this disclosure is to provide a method, apparatus, computer-readable storage medium, and electronic device for locating the root cause of a fault, thereby overcoming, to at least some extent, the technical problems of time-consuming, labor-intensive, and highly demanding personnel requirements caused by limitations in related technologies.

[0007] Other features and advantages of this disclosure will become apparent from the following detailed description, or may be learned in part from practice of this disclosure.

[0008] According to a first aspect of this disclosure, a fault root cause localization method is provided, applied to a trained fault root cause localization model, the method comprising:

[0009] Extract fault features from the fault description information of the user for the target fault scenario, and identify the user intent from the fault description information;

[0010] Based on the fault characteristics and the user intent, select key fault analysis nodes that are related to the target fault scenario from the candidate fault analysis nodes;

[0011] Execute the fault analysis task associated with the key fault analysis node to perform root cause localization processing on the target fault scenario and obtain the root cause localization result for the target fault scenario.

[0012] In an exemplary embodiment of this disclosure, the step of extracting fault features from the fault description information provided by the user for the target fault scenario includes:

[0013] The fault description information is subjected to information standardization processing to obtain standardized fault description information; the information standardization processing includes at least one of the following: synonym replacement, misspelling replacement, and omission of modal particles;

[0014] The fault features are extracted from the standardized fault description information.

[0015] In an exemplary embodiment of this disclosure, the step of selecting key fault analysis nodes that are related to the target fault scenario from candidate fault analysis nodes based on the fault characteristics and the user intent includes:

[0016] Based on the first correlation between the fault characteristics and the task information of the fault analysis tasks associated with the candidate fault analysis nodes, an initial fault analysis node is selected from the candidate fault analysis nodes.

[0017] The key fault analysis nodes are selected from the initial fault analysis nodes based on the second relevance between the user intent and the task information of the fault analysis task associated with the initial fault analysis node.

[0018] In an exemplary embodiment of this disclosure, the step of selecting key fault analysis nodes that are related to the target fault scenario from the candidate fault analysis nodes based on the fault characteristics and the user intent further includes:

[0019] Extract task keywords from the task information of the fault analysis task associated with each candidate fault analysis node;

[0020] Based on the first relevance between the fault characteristics and the task keywords associated with the candidate fault analysis nodes, an initial fault analysis node is selected from the candidate fault analysis nodes;

[0021] The key fault analysis nodes are selected from the initial fault analysis nodes based on the second relevance between the user intent and the task keywords associated with the initial fault analysis nodes.

[0022] In an exemplary embodiment of this disclosure, when there are multiple critical fault analysis nodes, executing the fault analysis task associated with the critical fault analysis nodes to perform root cause localization processing on the target fault scenario and obtain root cause localization results for the target fault scenario includes:

[0023] The multiple key fault analysis nodes are sorted according to the second relevance to obtain a sorting sequence;

[0024] According to the sorting order in the sorting sequence, the fault analysis tasks associated with each of the key fault analysis nodes are executed sequentially to perform fault root cause localization processing on the target fault scenario and obtain the root cause localization result for the target fault scenario.

[0025] In an exemplary embodiment of this disclosure, the step of sequentially executing the fault analysis task associated with each critical fault analysis node according to the sorting order in the sorting sequence to perform fault root cause localization processing on the target fault scenario and obtain the root cause localization result for the target fault scenario includes:

[0026] According to the sorting order in the sorting sequence, the fault analysis tasks associated with each of the key fault analysis nodes are executed sequentially to perform fault root cause localization processing on the target fault scenario and obtain the analysis results corresponding to each of the key fault analysis nodes.

[0027] Extract fault elements from the fault description information; the fault elements include the fault entity, fault time, and fault manifestation.

[0028] Based on the analysis results corresponding to each of the key fault analysis nodes and the fault elements, the root cause localization result for the target fault scenario is determined.

[0029] In an exemplary embodiment of this disclosure, the trained fault root cause localization model is obtained in the following manner:

[0030] Obtain a training sample set; the training sample set contains multiple training samples, different training samples correspond to different fault scenario samples, each training sample contains a fault description information sample for each fault scenario sample and a root cause localization result label corresponding to each fault scenario sample; each fault scenario sample is associated with a pre-constructed output guide word;

[0031] Each training sample is input into the initial fault root cause localization model, and the output guide word guides the initial fault root cause localization model to output the initial root cause localization result for each fault scenario sample;

[0032] The output guide word is adjusted based on the degree of difference between the initial root cause localization result and the root cause localization result label;

[0033] Each training sample is updated using the adjusted output guide word, and the initial fault root cause localization model is fine-tuned based on each updated training sample to obtain the trained fault root cause localization model.

[0034] In an exemplary embodiment of this disclosure, the output guide words associated with each fault scenario sample include node filtering guide words and result generation guide words;

[0035] The step of guiding the initial fault root cause localization model to output initial root cause localization results for each fault scenario sample through the output guide word includes:

[0036] The node filtering guides the initial fault root cause localization model to select target fault analysis nodes that are associated with each fault scenario sample from the candidate fault analysis nodes.

[0037] The initial fault root cause localization model is used to execute the fault analysis task associated with the target fault analysis node to perform fault root cause localization processing on each fault scenario sample, and obtain the analysis results corresponding to the target fault analysis node.

[0038] The results generate guiding words to guide the initial fault root cause localization model to output initial root cause localization results for each fault scenario sample based on the analysis results corresponding to the target fault analysis node.

[0039] In an exemplary embodiment of this disclosure, guiding the initial fault root cause localization model to select target fault analysis nodes that are correlated with each fault scenario sample from the candidate fault analysis nodes using the node selection guide words includes:

[0040] The initial fault root cause localization model extracts fault feature samples from fault description information samples for each fault scenario sample, and identifies user intent samples from the fault description information samples.

[0041] The node selection guides the initial fault root cause localization model to select target fault analysis nodes that are associated with each fault scenario sample from the candidate fault analysis nodes based on the fault feature samples and the user intent samples.

[0042] In an exemplary embodiment of this disclosure, after guiding the initial fault root cause localization model to output initial root cause localization results for each fault scenario sample via the output guide word, the method further includes:

[0043] Obtain the historical root cause localization results for each fault scenario sample;

[0044] The initial root cause localization results are expanded and optimized using the historical root cause localization results.

[0045] According to a second aspect of this disclosure, a fault root cause localization apparatus is provided, the apparatus being equipped with a trained fault root cause localization model, the apparatus comprising:

[0046] The information processing module is used to extract fault features from the fault description information of the user for the target fault scenario, and to identify the user's intent from the fault description information;

[0047] The node filtering module is used to filter key fault analysis nodes that are related to the target fault scenario from the candidate fault analysis nodes based on the fault characteristics and the user intent.

[0048] The fault root cause localization module is used to execute the fault analysis task associated with the key fault analysis node to perform fault root cause localization processing on the target fault scenario and obtain the root cause localization result for the target fault scenario.

[0049] According to a third aspect of this disclosure, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the fault root cause localization method described in the first aspect above.

[0050] According to a fourth aspect of this disclosure, an electronic device is provided, comprising: a processor; and a memory for storing executable instructions of the processor; wherein the processor is configured to perform the fault root cause localization method described in the first aspect by executing the executable instructions.

[0051] As can be seen from the above technical solutions, the fault root cause localization method, fault root cause localization device, computer-readable storage medium, and electronic device in the exemplary embodiments of this disclosure have at least the following advantages and positive effects:

[0052] In the technical solutions provided by some embodiments of this disclosure, on the one hand, by simultaneously considering fault characteristics and user intent, a more comprehensive understanding of the fault scenario can be achieved, thereby selecting the most relevant fault analysis node and ensuring that the finally selected analysis node is highly matched with the scenario; furthermore, more targeted key fault analysis nodes can be matched for different fault scenarios, thereby adapting to diverse fault scenarios and providing personalized analysis results; on the other hand, by executing fault analysis tasks associated with key fault analysis nodes to perform root cause localization processing on the target fault scenario and obtaining root cause localization results for the target fault scenario, resources can be concentrated on the key nodes most likely to solve the fault problem, reducing redundant analysis and improving the speed and efficiency of root cause localization, thereby generating more accurate and reliable root cause localization results for the target fault scenario, and performing root cause localization for complex fault scenarios more intelligently and efficiently.

[0053] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0054] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure. It is obvious that the drawings described below are merely some embodiments of this disclosure, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort.

[0055] Figure 1 A flowchart illustrating the fault root cause localization method in an embodiment of this disclosure is shown.

[0056] Figure 2 This diagram illustrates the process of training the aforementioned fault root cause localization model in an embodiment of the present disclosure.

[0057] Figure 3 This is a flowchart illustrating how, in an embodiment of this disclosure, the initial fault root cause localization model is guided to output initial root cause localization results for each fault scenario sample by outputting guiding words.

[0058] Figure 4 This illustration shows another flowchart in an embodiment of the present disclosure on how to guide the initial fault root cause localization model to select target fault analysis nodes that are associated with each fault scenario sample from candidate fault analysis nodes by using node filtering guide words.

[0059] Figure 5 This illustration shows a flowchart of a first embodiment of the present disclosure, which shows how to select key fault analysis nodes that are related to the target fault scenario from candidate fault analysis nodes based on fault characteristics and user intent.

[0060] Figure 6 This illustration shows a flowchart of a second method in this disclosure for selecting key fault analysis nodes that are related to the target fault scenario from candidate fault analysis nodes based on fault characteristics and user intent.

[0061] Figure 7 This is a flowchart illustrating how, in an embodiment of the present disclosure, the fault analysis tasks associated with each key fault analysis node are executed sequentially according to the sorting order in the sorting sequence to perform fault root cause localization processing on the target fault scenario and obtain the root cause localization result for the target fault scenario.

[0062] Figure 8 This diagram illustrates the overall flow of the fault root cause localization method in an embodiment of this disclosure.

[0063] Figure 9 This diagram illustrates the structure of a fault root cause localization device in an exemplary embodiment of this disclosure.

[0064] Figure 10 A schematic diagram of the structure of an electronic device in an exemplary embodiment of this disclosure is shown. Detailed Implementation

[0065] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided to make this disclosure more comprehensive and complete, and to fully convey the concept of the example embodiments to those skilled in the art. The described features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. In the following description, numerous specific details are provided to give a full understanding of embodiments of this disclosure. However, those skilled in the art will recognize that the technical solutions of this disclosure can be practiced with one or more of the specific details omitted, or other methods, components, apparatus, steps, etc., can be employed. In other instances, well-known technical solutions are not shown or described in detail to avoid obscuring various aspects of this disclosure.

[0066] The terms “a,” “an,” “the,” and “the” are used in this specification to indicate the presence of one or more elements / components / etc.; the terms “including” and “having” are used to indicate an open-ended inclusion and to mean that there may be other elements / components / etc. in addition to the listed elements / components / etc.; the terms “first” and “second” are used only as markings and are not a limitation on the number of objects.

[0067] Furthermore, the accompanying drawings are merely illustrative of this disclosure and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities.

[0068] In related technologies, fault scenario analysis typically relies on manual processes to compile and organize analytical checklists. These checklists may include collecting data from multiple sources, performing complex queries and calculations, and interpreting and inferring results based on experience and expertise.

[0069] However, the above solution has at least the following drawbacks:

[0070] First, the limitations of manual processing: relying on manual sorting, arranging, and analyzing checklists requires a significant amount of time and resources. Manual processing is also prone to errors, especially when dealing with complex systems and diverse fault scenarios.

[0071] Second, the challenge of updating and expanding: As the system becomes increasingly complex and new failure scenarios emerge, the analysis scheme needs to be constantly updated and expanded. This is a huge challenge for manual processing, because it requires experts not only to have in-depth system knowledge, but also to have enough time and energy to keep up with the changes.

[0072] Third, the professional knowledge threshold: only professional personnel such as system administrators have the ability to perform these operations, which makes it impossible for ordinary users to perform fault diagnosis and analysis independently. This limits knowledge sharing and collaboration within the organization and may delay the resolution of problems.

[0073] Fourth, inefficiency: Due to the limitations of manual processing, the speed and efficiency of fault scenario analysis are often very low. Even the most experienced experts need to spend a lot of time collecting, analyzing, and interpreting data, which may lead to the blockage or delay of important business processes;

[0074] Fifth, poor scalability: As business grows and systems expand, manual processing methods become increasingly impractical. They struggle to handle the demands of large-scale fault scenario analysis and fail to meet the requirements of modern enterprises for rapid, accurate, and reliable fault diagnosis and analysis.

[0075] To address the aforementioned issues, this disclosure proposes a fault root cause localization method. By introducing a small intent recognition model, constructing an output prompt, extracting fault elements, and employing multi-round prompt interaction, a fault root cause localization model is trained. This model can automatically filter key fault analysis nodes that are correlated with the target fault scenario from a variety of candidate fault analysis nodes, thereby contributing to the generation of more accurate and reliable root cause localization results for the target fault scenario. Furthermore, this disclosure can also interactively supplement existing root cause localization results with the output of a large model, further improving the coverage of online fault scenario analysis, significantly lowering the threshold for fault scenario analysis, and improving efficiency and accuracy.

[0076] In the embodiments of this disclosure, a fault root cause localization method is first provided, which at least to some extent overcomes the shortcomings of related technologies that are time-consuming, labor-intensive, and have high requirements for personnel.

[0077] Figure 1 The flowchart of the fault root cause localization method in the present disclosure is shown. The execution subject of the fault root cause localization method can be a server that performs root cause analysis on the fault.

[0078] refer to Figure 1 A fault root cause localization method according to an embodiment of the present disclosure includes the following steps:

[0079] Step S110: Extract fault features from the fault description information of the user for the target fault scenario, and identify the user intent from the fault description information;

[0080] Step S120: Based on the fault characteristics and user intent, select key fault analysis nodes that are related to the target fault scenario from the candidate fault analysis nodes.

[0081] Step S130: Execute the fault analysis task associated with the key fault analysis node to perform root cause localization processing on the target fault scenario and obtain the root cause localization result for the target fault scenario.

[0082] exist Figure 1 In the technical solution provided by the illustrated embodiment, on the one hand, by simultaneously considering fault characteristics and user intent, a more comprehensive understanding of the fault scenario can be achieved, thereby selecting the most relevant fault analysis node and ensuring that the final selected analysis node is highly matched with the scenario; furthermore, more targeted key fault analysis nodes can be matched for different fault scenarios, thereby adapting to diverse fault scenarios and providing personalized analysis results; on the other hand, by executing fault analysis tasks associated with key fault analysis nodes to perform root cause localization processing on the target fault scenario and obtaining root cause localization results for the target fault scenario, resources can be concentrated on the key nodes most likely to solve the fault problem, reducing redundant analysis and improving the speed and efficiency of root cause localization, thereby generating more accurate and reliable root cause localization results for the target fault scenario, and performing root cause localization for complex fault scenarios more intelligently and efficiently.

[0083] The following are Figure 1 The specific implementation process of each step in the process will be explained in detail:

[0084] Before step S110, it should be noted that the present disclosure can pre-train a fault root cause localization model. The function of the fault root cause localization model is: after receiving the fault description information of the user for the target fault scenario, it selects key fault analysis nodes that are related to the target fault scenario from the candidate fault analysis nodes, executes the fault analysis tasks associated with the key fault analysis nodes to perform fault root cause localization processing on the target fault scenario, and outputs the root cause localization result for the target fault scenario.

[0085] The training process of the fault root cause localization model in this disclosure will be explained below. (Reference) Figure 2 , Figure 2 This diagram illustrates the process of training the aforementioned fault root cause localization model in this embodiment of the present disclosure, including steps S201-S204:

[0086] In step S201, the training sample set is obtained.

[0087] In this step, a training sample set can be obtained. The training sample set can contain multiple training samples. Different training samples correspond to different fault scenario samples. That is, each training sample corresponds to a specific fault situation or scenario. Different training samples represent different fault scenarios, ensuring that the model can learn multiple types of faults and their handling methods.

[0088] Each training sample can contain fault description information for each fault scenario sample and root cause localization result labels for each fault scenario sample.

[0089] The above-mentioned fault description information sample is a detailed description of a specific fault situation, which may include the following:

[0090] Faulty entity: The specific device, component, or system involved (such as a server, network connection, etc.);

[0091] Failure time: The point in time or period during which the failure occurred;

[0092] Fault manifestation: Specific symptoms or abnormal behavior of the fault (such as "The server cannot respond to HTTP requests");

[0093] Environmental conditions: Background information on the occurrence of the fault (such as operating system version, network configuration, etc.);

[0094] Scope of impact: This describes which systems, services, or users were affected by the failure.

[0095] The root cause localization result labels mentioned above can be the correct root cause localization results pre-labeled for each fault scenario sample, and typically include the following:

[0096] Root cause: The most fundamental factor that causes the failure (such as hardware failure, software bug, configuration error, etc.);

[0097] Solution: Specific measures or suggestions for fixing the fault (such as replacing hardware, updating software, adjusting configuration, etc.).

[0098] In addition, each fault scenario sample can be associated with a pre-built output prompt, which helps the model to understand the fault scenario information more accurately and guides the model to generate answers with the expected format and content.

[0099] In step S202, each training sample is input into the initial fault root cause localization model, and the initial fault root cause localization model is guided by outputting a guiding word to output the initial root cause localization result for each fault scenario sample.

[0100] In this step, each training sample can be input into the initial fault root cause localization model, and the initial fault root cause localization model can be guided to output the initial root cause localization result for each fault scenario sample by outputting guiding words.

[0101] The aforementioned initial fault root cause localization model is the fault root cause localization model to be trained, and the aforementioned initial root cause localization result refers to the preliminary analysis result of the model on the root cause of the fault generated based on the input fault description information.

[0102] Specifically, the output guidance words associated with each of the above fault scenario samples can include node selection guidance words and result generation guidance words. The role of node selection guidance words is to guide the model to select appropriate fault analysis nodes for more in-depth analysis. It helps the model to select the most relevant initial or key fault analysis nodes from the candidate fault analysis nodes based on fault description information and user intent. The role of result generation guidance words is to guide the model to generate the root cause localization result with the highest accuracy that is closest to the labeled label.

[0103] refer to Figure 3 , Figure 3 This embodiment of the present disclosure illustrates how the initial fault root cause localization model is guided to output initial root cause localization results for each fault scenario sample by outputting guiding words, including steps S301-S303:

[0104] In step S301, the initial fault root cause localization model is guided by node filtering guide words to select target fault analysis nodes that are related to each fault scenario sample from the candidate fault analysis nodes.

[0105] In this step, in one optional implementation, after inputting each training sample into the initial fault root cause localization model, the model can be guided by node filtering prompts to select target fault analysis nodes that are associated with each fault scenario sample from the candidate fault analysis nodes based on the fault description information sample of each fault scenario sample. Specifically, since each candidate fault analysis node is associated with a fault analysis task, and each fault analysis task is associated with fault analysis task information, which describes the task responsibilities corresponding to the fault analysis task, the initial fault root cause localization model can be guided by node filtering prompts to calculate the correlation between the fault description information sample of each fault scenario sample and the fault analysis task information corresponding to each candidate fault analysis node, and select target fault analysis nodes that are associated with each fault scenario sample from the candidate fault analysis nodes based on the correlation.

[0106] In another alternative implementation, reference can be made to Figure 4 , Figure 4This illustration shows another flowchart in this embodiment of the process of guiding the initial fault root cause localization model to select target fault analysis nodes that are associated with each fault scenario sample from the candidate fault analysis nodes by using node selection guide words, including steps S401-S402:

[0107] In step S401, fault feature samples are extracted from the fault description information samples for each fault scenario sample through the initial fault root cause localization model, and user intent samples are identified from the fault description information samples.

[0108] In this step, fault feature samples can be extracted from the fault description information samples for each fault scenario sample using an initial fault root cause localization model, and user intent samples can be identified from the fault description information samples based on an intent recognition model. Fault feature samples refer to specific, quantifiable features or attributes extracted from the fault description information. These features help the model identify and understand the nature of the fault. For example, they may include the fault entity, fault time, fault manifestation, environmental conditions, etc., and can be set according to actual conditions; this disclosure does not impose any special limitations on this. User intent samples refer to the user's needs, purposes, or expectations identified from the fault description information, to help the model generate responses that better meet user needs.

[0109] In step S402, the initial fault root cause localization model is guided by node filtering guide words to select target fault analysis nodes that are related to each fault scenario sample from the candidate fault analysis nodes based on fault feature samples and user intent samples.

[0110] In this step, after obtaining fault feature samples and user intent samples, task keywords can be extracted from the task information of the fault analysis tasks associated with each candidate fault analysis node. Then, the initial fault root cause localization model can first calculate the first relevance between the fault feature samples and the task keywords associated with each candidate fault analysis node. Using node selection guide words, the initial fault root cause localization model can select initial fault analysis nodes from the candidate fault analysis nodes based on the aforementioned first relevance. Next, the initial fault root cause localization model can calculate the second relevance between the user intent samples and the task keywords associated with each candidate fault analysis node. Using node selection guide words, the initial fault root cause localization model can select target fault analysis nodes that are associated with each fault scenario sample from the candidate fault analysis nodes based on the aforementioned second relevance.

[0111] Optionally, when there are multiple target fault analysis nodes selected, after the initial fault root cause localization model selects the target fault analysis nodes according to the second relevance mentioned above, it can also sort the selected target fault analysis nodes in descending order of the second relevance to form a sorting sequence, so that the most relevant fault analysis task can be executed first when the fault analysis task is executed in the future, thereby helping to find the root cause of the fault more quickly and provide an effective solution.

[0112] Next, refer to Figure 3 In step S302, the fault analysis task associated with the target fault analysis node is executed through the initial fault root cause localization model to perform fault root cause localization processing on each fault scenario sample and obtain the analysis results corresponding to the target fault analysis node.

[0113] In this step, the initial fault root cause localization model can execute the fault analysis tasks associated with the aforementioned target fault analysis nodes to perform fault root cause localization processing on each fault scenario sample, thereby obtaining the analysis results corresponding to the target fault analysis nodes. Alternatively, it can execute the fault analysis tasks associated with each target fault analysis node in the order listed above to perform fault root cause localization processing on each fault scenario sample, thereby obtaining the analysis results corresponding to the target fault analysis nodes.

[0114] In step S303, the initial fault root cause localization model is guided by the result generation guide words to output the initial root cause localization result for each fault scenario sample based on the analysis result corresponding to the target fault analysis node.

[0115] In this step, in one optional implementation, the initial fault root cause localization model can be guided by the result generation guide word to output the initial root cause localization result for each fault scenario sample based on the analysis result corresponding to the target fault analysis node.

[0116] In another optional implementation, the initial fault root cause localization model can also extract fault elements from the fault description information sample corresponding to each fault scenario sample, such as fault entity, fault time, fault manifestation, etc. Then, the initial fault root cause localization model can be guided by the result generation guide words to output the initial root cause localization result for each fault scenario sample based on the analysis results corresponding to the target fault analysis node and the fault elements of each fault scenario sample.

[0117] Optionally, after obtaining the initial root cause localization results for each fault scenario sample, the historical root cause localization results for each fault scenario sample can also be obtained. The historical root cause localization results can be the root cause localization results designed by domain experts after conducting manual fault analysis for each fault scenario sample. Thus, the historical root cause localization results can be used to expand and optimize the above-mentioned initial root cause localization results, so as to realize the interactive supplementation of the model output results by using the existing root cause analysis scheme, and further improve the accuracy of the model's results for fault scenario analysis.

[0118] Next, refer to Figure 2 In step S203, the output guide word is adjusted according to the degree of difference between the initial root cause localization result and the root cause localization result label.

[0119] In this step, after obtaining the initial root cause localization result or the expanded and optimized initial root cause localization result for each fault scenario sample, the output guidance words can be adjusted according to the degree of difference between the initial root cause localization result and the root cause localization result label. That is, the node filtering guidance words and result generation guidance words associated with each fault scenario sample are adjusted so that the designed output guidance words can help the model better understand the fault scenario.

[0120] For example, taking the above-mentioned result generation guide words as an example, for a fault scenario sample involving network service latency issues, the initially constructed result generation guide words could be "Please analyze the root cause of the following network service latency issues: <fault description>", and the adjusted result generation guide words could be "Please consider the impact of geographical location and analyze the root cause of the following network service latency issues: <fault description>". Thus, by continuously optimizing the guide words, the model can better understand and handle complex operation and maintenance scenarios, ultimately enabling the model to generate accurate root cause location results by combining different fault phenomena, impact ranges, and root causes.

[0121] In step S204, each training sample is updated using the adjusted output guide word, and the initial fault root cause localization model is fine-tuned based on each updated training sample to obtain a trained fault root cause localization model.

[0122] In this step, each training sample can be updated using the adjusted output guide words, and the initial fault root cause localization model can be fine-tuned based on each updated training sample. After multiple rounds of interactive fine-tuning, if the initial root cause localization results output by the initial fault root cause localization model can cover the root cause localization result labels well, that is, the initial root cause localization results output by the model have high accuracy, then the training of the initial fault root cause localization model can be terminated, and a trained fault root cause localization model can be obtained.

[0123] After obtaining the trained root cause localization model, the model application stage can be entered, that is, the trained root cause localization model is deployed to the production environment to perform the relevant processes of steps S110-S130 through the trained model. The following explains steps S110-S130 in this disclosure.

[0124] In step S110, fault features are extracted from the fault description information of the user for the target fault scenario, and user intent is identified from the fault description information.

[0125] In this step, the system can receive fault description information from the user regarding the target fault scenario, extract fault features from the fault description information, and use an intent recognition model to identify the user's intent from the fault description information.

[0126] Optionally, after receiving the fault description information from the user for the target fault scenario, the fault description information can be first processed to obtain standardized fault description information, and then fault features can be extracted from the standardized fault description information.

[0127] The aforementioned information standardization process may include at least one of the following: synonym replacement, misspelling replacement, and omission of modal particles. Synonym replacement refers to replacing certain words in the original text with semantically similar or identical words to enhance the diversity and accuracy of expression; misspelling replacement refers to identifying and correcting spelling errors in the text to ensure that every word is correct; and omission of modal particles refers to removing modal particles, redundant words, or other unnecessary expressions from the text to make it more concise and clear.

[0128] By standardizing the fault description information before extracting fault features, noise in the text can be eliminated, making key information more prominent, thereby improving the accuracy of feature extraction and enhancing the model's understanding ability.

[0129] The above are specific, quantifiable features extracted from fault description information, which help the model identify and understand the nature of the fault. The user intent mentioned above is the user's needs, purposes, or expectations identified from the fault description information, which helps the model generate responses that better meet the user's needs.

[0130] In step S120, based on the fault characteristics and user intent, key fault analysis nodes that are related to the target fault scenario are selected from the candidate fault analysis nodes.

[0131] In this step, the trained root cause analysis model can select key fault analysis nodes that are related to the target fault scenario from the candidate fault analysis nodes based on fault characteristics and user intent.

[0132] In one alternative implementation, refer to Figure 5 , Figure 5 This illustration shows a flowchart of the first method in this disclosure for selecting key fault analysis nodes that are related to the target fault scenario from candidate fault analysis nodes based on fault characteristics and user intent, including steps S501-S502:

[0133] In step S501, an initial fault analysis node is selected from the candidate fault analysis nodes based on the first correlation between the fault characteristics and the task information of the fault analysis task associated with the candidate fault analysis node.

[0134] In this step, the first correlation between the fault characteristics and the task information of the fault analysis task associated with the candidate fault analysis nodes can be calculated, and the initial fault analysis nodes can be coarsely screened out from the candidate fault analysis nodes based on the first correlation.

[0135] In step S502, key fault analysis nodes are selected from the initial fault analysis nodes based on the second correlation between the user intent and the task information of the fault analysis task associated with the initial fault analysis node.

[0136] In this step, a second correlation can be calculated between the user's intent and the task information of the fault analysis task associated with the initial fault analysis node. Based on the second correlation, key fault analysis nodes can be further filtered out from the initial fault analysis nodes.

[0137] In another alternative implementation, refer to Figure 6 , Figure 6 This illustration shows a flowchart of a second method in this disclosure for selecting key fault analysis nodes that are related to the target fault scenario from candidate fault analysis nodes based on fault characteristics and user intent, including steps S601-S603:

[0138] In step S601, task keywords are extracted from the task information of the fault analysis task associated with each candidate fault analysis node.

[0139] In this step, since each candidate fault analysis node is associated with a fault analysis task, and each fault analysis task is associated with task information, we can first extract task keywords from each task information. Task keywords refer to words or phrases extracted from the task information of each fault analysis task that describe the core content and key operations of the task. These keywords can help the model to more accurately understand and select the most relevant fault analysis node, thereby improving the efficiency and accuracy of subsequent fault diagnosis.

[0140] In step S602, an initial fault analysis node is selected from the candidate fault analysis nodes based on the first relevance between the fault characteristics and the task keywords associated with the candidate fault analysis nodes.

[0141] In this step, the initial fault analysis node can be roughly screened out from the candidate fault analysis nodes based on the first relevance between the fault characteristics of the target fault scenario and the task keywords associated with the candidate fault analysis nodes.

[0142] In step S603, key fault analysis nodes are selected from the initial fault analysis nodes based on the second relevance between the user intent and the task keywords associated with the initial fault analysis nodes.

[0143] In this step, key fault analysis nodes can be further filtered from the initial fault analysis nodes based on the second relevance between the user intent and the task keywords associated with the initial fault analysis nodes.

[0144] Thus, this disclosure can ensure that the selected nodes are highly matched with the fault characteristics and user intent through accurate task keyword extraction and dual relevance evaluation, thus solving the illusion problem that may occur when large models face multiple fault analysis nodes, i.e., not knowing which fault analysis nodes to select or selecting the wrong ones.

[0145] After obtaining the critical fault analysis node based on step S502 or step S603, this disclosure can automatically create a fault root cause analysis task to automatically execute the fault analysis task associated with each critical fault analysis node, thereby reducing manual intervention.

[0146] Next, refer to Figure 1 In step S130, the fault analysis task associated with the key fault analysis node is executed to perform root cause localization processing on the target fault scenario and obtain the root cause localization result for the target fault scenario.

[0147] In this step, the trained fault root cause localization model can execute fault analysis tasks associated with the above-mentioned key fault analysis nodes to perform fault root cause localization processing on the above-mentioned target fault scenario and obtain the root cause localization results for the target fault scenario.

[0148] Specifically, when there are multiple key fault analysis nodes, after obtaining the key fault analysis nodes based on step S502 or step S603, the multiple key fault analysis nodes can be sorted according to the second relevance to obtain a sorting sequence. Then, the fault analysis tasks associated with each key fault analysis node can be executed sequentially according to the sorting order in the sorting sequence to perform fault root cause localization processing on the target fault scenario and obtain the root cause localization result for the target fault scenario.

[0149] refer to Figure 7 , Figure 7 This embodiment of the present disclosure illustrates how, in accordance with the sorting order in the sorting sequence, the fault analysis tasks associated with each key fault analysis node are executed sequentially to perform fault root cause localization processing on the target fault scenario, and to obtain the root cause localization result for the target fault scenario. The flowchart includes steps S701-S703:

[0150] In step S701, the fault analysis tasks associated with each key fault analysis node are executed sequentially according to the sorting order in the sorting sequence to perform fault root cause localization processing on the target fault scenario and obtain the analysis results corresponding to each fault analysis node.

[0151] In this step, the trained root cause localization model can sequentially execute the fault analysis tasks associated with each key fault analysis node according to the sorting order in the sorting sequence to perform root cause localization processing on the target fault scenario and obtain the analysis results corresponding to each fault analysis node. For example, when the fault analysis task associated with a certain key fault analysis node is "check memory usage", the analysis result obtained after executing the fault analysis task associated with that key fault analysis node could be: a memory leak problem was found, and it is recommended to fix the application code with the memory leak and restart the server; when the fault analysis task associated with a certain key fault analysis node is "check network connectivity status", the analysis result obtained after executing the fault analysis task associated with that key fault analysis node could be: the network connection is confirmed to be normal, with no obvious problems.

[0152] In step S702, fault elements are extracted from the fault description information.

[0153] In this step, the trained root cause localization model can extract fault elements from the fault description information. For example, fault elements can include the fault entity, fault time, and fault manifestation. For instance, when the user inputs the fault description information as "Around 3 pm yesterday, my computer suddenly crashed with a blue screen and could not start normally after restarting," the extracted fault elements could be: fault entity (the user's computer), fault time (around 3 pm yesterday), and fault manifestation (the computer crashed with a blue screen and could not start normally).

[0154] In step S703, the root cause localization result for the target fault scenario is determined based on the analysis results corresponding to each fault analysis node and the fault elements.

[0155] In this step, the trained root cause localization model can integrate the fault elements of the target fault scenario and the analysis results corresponding to each key fault analysis node to generate root cause localization results for the target fault scenario.

[0156] It should be noted that after the trained root cause localization model is deployed to the production environment, this disclosure can also monitor the model in real time, evaluate the model performance regularly, and continuously optimize the model based on the performance evaluation results.

[0157] refer to Figure 8 , Figure 8 This diagram illustrates the overall flow chart of the fault root cause localization method in this embodiment, including steps S801-S806:

[0158] In step S801, a training sample set is obtained. Specifically, fault description information samples for each fault scenario sample can be collected, and output guidance words corresponding to each fault scenario sample can be designed.

[0159] In step S802, a small model for intent recognition is constructed;

[0160] In step S803, user intent is identified based on the intent recognition mini-model, and key fault analysis nodes are selected based on the correlation between the user intent and the task information of the fault analysis tasks associated with the candidate fault analysis nodes.

[0161] In step S804, fault elements are extracted from the fault description information sample;

[0162] In step S805, an initial root cause localization result is generated based on the analysis results obtained after performing the fault analysis task associated with the key fault analysis node and the above-mentioned fault elements. Based on the degree of difference between the initial root cause localization result and the root cause localization result label, the output prompt is adjusted and the training samples are updated to optimize the model.

[0163] In step S806, after the model is trained, it is saved, deployed, and continuously optimized.

[0164] Based on the above technical solutions, this disclosure has at least the following technical effects:

[0165] First, by adding a small intent recognition model to identify user intent, key fault analysis nodes can be filtered based on fault characteristics and user intent. By simultaneously considering fault characteristics (such as fault entity, time, and manifestation) and user intent (such as understanding the cause and seeking solutions), the key fault analysis nodes most relevant to the target fault scenario can be filtered more accurately. This avoids the problem of large model illusion, not knowing which nodes to select or selecting the wrong ones, thereby significantly reducing the resource consumption for generating fault analysis solutions, facilitating the rapid output of fault analysis solutions, and obtaining root cause localization results with high accuracy.

[0166] Second, by adjusting the output prompts to update the training samples and fine-tuning the large model through multiple rounds of prompt interaction, the large model can better understand the root cause analysis operation and maintenance scenario.

[0167] Third, by interacting and supplementing the initial root cause localization results output by the large model with the manually designed historical root cause localization results, the analytical coverage of the large model for online fault scenarios can be further improved.

[0168] Therefore, this disclosure can be applied to at least the following scenarios:

[0169] ① Log template alerts: can monitor the logs generated by the system or application to locate the root cause of system or application failures;

[0170] ② Unified Alarm Center (a centralized platform for collecting and managing alarm information from multiple sources): It can locate the root cause of various alarms issued by the unified alarm center;

[0171] ③ Robot interaction: It can help users generate fault scenario analysis solutions and generate fault root cause location results through chat.

[0172] This disclosure also provides a fault root cause localization device, which is equipped with a trained fault root cause localization model. Figure 9 This diagram illustrates the structure of a fault root cause localization device in an exemplary embodiment of this disclosure; as shown below. Figure 9 As shown, the fault root cause localization device 900 may include an information processing module 910, a node filtering module 920, a fault root cause localization module 930, and a model training module 940.

[0173] in:

[0174] The information processing module 910 is used to extract fault features from the fault description information of the user for the target fault scenario, and to identify the user's intent from the fault description information.

[0175] The node filtering module 920 is used to filter key fault analysis nodes that are related to the target fault scenario from the candidate fault analysis nodes based on the fault characteristics and the user intent.

[0176] The fault root cause localization module 930 is used to execute the fault analysis task associated with the key fault analysis node to perform fault root cause localization processing on the target fault scenario and obtain the root cause localization result for the target fault scenario.

[0177] In an exemplary embodiment of this disclosure, the information processing module 910 extracts fault features from the user's fault description information for the target fault scenario, including:

[0178] The fault description information is subjected to information standardization processing to obtain standardized fault description information; the information standardization processing includes at least one of the following: synonym replacement, misspelling replacement, and omission of modal particles;

[0179] The fault features are extracted from the standardized fault description information.

[0180] In an exemplary embodiment of this disclosure, the node filtering module 920 filters key fault analysis nodes that are related to the target fault scenario from candidate fault analysis nodes based on the fault characteristics and the user intent, including:

[0181] Based on the first correlation between the fault characteristics and the task information of the fault analysis tasks associated with the candidate fault analysis nodes, an initial fault analysis node is selected from the candidate fault analysis nodes.

[0182] The key fault analysis nodes are selected from the initial fault analysis nodes based on the second relevance between the user intent and the task information of the fault analysis task associated with the initial fault analysis node.

[0183] In an exemplary embodiment of this disclosure, the node filtering module 920, based on the fault characteristics and the user intent, filters key fault analysis nodes from candidate fault analysis nodes that are related to the target fault scenario, and further includes:

[0184] Extract task keywords from the task information of the fault analysis task associated with each candidate fault analysis node;

[0185] Based on the first relevance between the fault characteristics and the task keywords associated with the candidate fault analysis nodes, an initial fault analysis node is selected from the candidate fault analysis nodes;

[0186] The key fault analysis nodes are selected from the initial fault analysis nodes based on the second relevance between the user intent and the task keywords associated with the initial fault analysis nodes.

[0187] In an exemplary embodiment of this disclosure, when the number of key fault analysis nodes is multiple, the fault root cause localization module 930 executes the fault analysis task associated with the key fault analysis nodes to perform fault root cause localization processing on the target fault scenario, and obtains root cause localization results for the target fault scenario, including:

[0188] The multiple key fault analysis nodes are sorted according to the second relevance to obtain a sorting sequence;

[0189] According to the sorting order in the sorting sequence, the fault analysis tasks associated with each of the key fault analysis nodes are executed sequentially to perform fault root cause localization processing on the target fault scenario and obtain the root cause localization result for the target fault scenario.

[0190] In an exemplary embodiment of this disclosure, the fault root cause localization module 930 executes the fault analysis task associated with each key fault analysis node sequentially according to the sorting order in the sorting sequence to perform fault root cause localization processing on the target fault scenario, and obtains the root cause localization result for the target fault scenario, including:

[0191] According to the sorting order in the sorting sequence, the fault analysis tasks associated with each of the key fault analysis nodes are executed sequentially to perform fault root cause localization processing on the target fault scenario and obtain the analysis results corresponding to each of the key fault analysis nodes.

[0192] Extract fault elements from the fault description information; the fault elements include the fault entity, fault time, and fault manifestation.

[0193] Based on the analysis results corresponding to each of the key fault analysis nodes and the fault elements, the root cause localization result for the target fault scenario is determined.

[0194] In an exemplary embodiment of this disclosure, the apparatus includes a model training module 940, which is configured to obtain a trained fault root cause localization model based on the following:

[0195] Obtain a training sample set; the training sample set contains multiple training samples, different training samples correspond to different fault scenario samples, each training sample contains a fault description information sample for each fault scenario sample and a root cause localization result label corresponding to each fault scenario sample; each fault scenario sample is associated with a pre-constructed output guide word;

[0196] Each training sample is input into the initial fault root cause localization model, and the output guide word guides the initial fault root cause localization model to output the initial root cause localization result for each fault scenario sample;

[0197] The output guide word is adjusted based on the degree of difference between the initial root cause localization result and the root cause localization result label;

[0198] Each training sample is updated using the adjusted output guide word, and the initial fault root cause localization model is fine-tuned based on each updated training sample to obtain the trained fault root cause localization model.

[0199] In an exemplary embodiment of this disclosure, the output guide words associated with each fault scenario sample include node filtering guide words and result generation guide words;

[0200] The model training module 940 guides the initial fault root cause localization model to output initial root cause localization results for each fault scenario sample through the output guide words, including:

[0201] The node filtering guides the initial fault root cause localization model to select target fault analysis nodes that are associated with each fault scenario sample from the candidate fault analysis nodes.

[0202] The initial fault root cause localization model is used to execute the fault analysis task associated with the target fault analysis node to perform fault root cause localization processing on each fault scenario sample, and obtain the analysis results corresponding to the target fault analysis node.

[0203] The results generate guiding words to guide the initial fault root cause localization model to output initial root cause localization results for each fault scenario sample based on the analysis results corresponding to the target fault analysis node.

[0204] In an exemplary embodiment of this disclosure, the model training module 940 guides the initial fault root cause localization model to select target fault analysis nodes that are correlated with each fault scenario sample from the candidate fault analysis nodes by using the node selection guide words, including:

[0205] The initial fault root cause localization model extracts fault feature samples from fault description information samples for each fault scenario sample, and identifies user intent samples from the fault description information samples.

[0206] The node selection guides the initial fault root cause localization model to select target fault analysis nodes that are associated with each fault scenario sample from the candidate fault analysis nodes based on the fault feature samples and the user intent samples.

[0207] In an exemplary embodiment of this disclosure, after the initial fault root cause localization model is guided by the output guide word to output the initial root cause localization result for each fault scenario sample, the model training module 940 is configured to:

[0208] Obtain the historical root cause localization results for each fault scenario sample;

[0209] The initial root cause localization results are expanded and optimized using the historical root cause localization results.

[0210] The specific details of each module in the above-mentioned fault root cause localization device have been described in detail in the corresponding fault root cause localization method, so they will not be repeated here.

[0211] It should be noted that although several modules or units for the device used to perform actions have been mentioned in the detailed description above, this division is not mandatory. In fact, according to embodiments of this disclosure, the features and functions of two or more modules or units described above can be embodied in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided and embodied by multiple modules or units.

[0212] Furthermore, although the steps of the method in this disclosure are described in a specific order in the accompanying drawings, this does not require or imply that the steps must be performed in that specific order, or that all the steps shown must be performed to achieve the desired result. Additional or alternative steps may be omitted, multiple steps may be combined into one step, and / or a step may be broken down into multiple steps.

[0213] From the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein can be implemented by software or by combining software with necessary hardware. Therefore, the technical solutions according to the embodiments of this disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, external hard drive, etc.) or on a network, including several instructions to cause a computing device (such as a personal computer, server, mobile terminal, or network device, etc.) to execute the methods according to the embodiments of this disclosure.

[0214] This disclosure also provides a computer-readable storage medium, which may be included in the electronic device described in the above embodiments; or it may exist independently and not assembled into the electronic device.

[0215] Computer-readable storage media can be, for example—but not limited to—electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatuses, or devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this disclosure, a computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0216] A computer-readable storage medium can be sent, propagated, or transmitted for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable storage medium can be transmitted using any suitable medium, including but not limited to: wireless, wireline, optical fiber, RF, etc., or any suitable combination thereof.

[0217] A computer-readable storage medium carries one or more programs that, when executed by an electronic device, cause the electronic device to perform the methods described in the above embodiments.

[0218] Furthermore, this disclosure also provides an electronic device capable of implementing the above-described method.

[0219] Those skilled in the art will understand that various aspects of this disclosure can be implemented as a system, method, or program product. Therefore, various aspects of this disclosure can be specifically implemented in the following forms: a completely hardware implementation, a completely software implementation (including firmware, microcode, etc.), or a combination of hardware and software aspects, collectively referred to herein as a "circuit," "module," or "system."

[0220] The following reference Figure 10 To describe an electronic device 1000 according to such an embodiment of the present disclosure. Figure 10 The electronic device 1000 shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments disclosed herein.

[0221] like Figure 10 As shown, the electronic device 1000 is presented in the form of a general-purpose computing device. The components of the electronic device 1000 may include, but are not limited to: at least one processor 1010, at least one memory 1020, a bus 1030 connecting different system components (including memory 1020 and processor 1010), and a display 1040.

[0222] The memory stores program code that can be executed by the processor 1010, causing the processor 1010 to perform the steps described in the "Exemplary Methods" section of this specification according to various exemplary embodiments of this disclosure. For example, the processor 1010 can perform actions such as... Figure 1 As shown: Step S110, extract fault features from the fault description information of the user for the target fault scenario, and identify the user intent from the fault description information; Step S120, based on the fault features and the user intent, select key fault analysis nodes that are related to the target fault scenario from the candidate fault analysis nodes; Step S130, execute the fault analysis task associated with the key fault analysis nodes to perform fault root cause localization processing on the target fault scenario, and obtain the root cause localization result for the target fault scenario.

[0223] The memory 1020 may include a readable medium in the form of volatile storage, such as random access memory (RAM) 10201 and / or cache memory 10202, and may further include read-only memory (ROM) 10203.

[0224] The memory 1020 may also include a program / utility 10204 having a set (at least one) of program modules 10205, such program modules 10205 including but not limited to: an operating system, one or more application programs, other program modules, and program data, each or some combination of these examples may include an implementation of a network environment.

[0225] Bus 1030 can represent one or more of several types of bus structures, including a memory bus or memory controller, peripheral bus, graphics acceleration port, processor, or a local bus using any of the various bus structures.

[0226] Electronic device 1000 can also communicate with one or more external devices 1100 (e.g., keyboard, pointing device, Bluetooth device, etc.), one or more devices that enable a user to interact with electronic device 1000, and / or any device that enables electronic device 1000 to communicate with one or more other computing devices (e.g., router, modem, etc.). This communication can be performed via input / output (I / O) interface 1050. Furthermore, electronic device 1000 can also communicate with one or more networks (e.g., local area network (LAN), wide area network (WAN), and / or public networks, such as the Internet) via network adapter 1060. As shown, network adapter 1060 communicates with other modules of electronic device 1000 via bus 1030. It should be understood that, although not shown in the figures, other hardware and / or software modules can be used in conjunction with electronic device 1000, including but not limited to: microcode, device drivers, redundant processors, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0227] Other embodiments of this disclosure will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This disclosure is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not disclosed herein. The specification and embodiments are to be considered exemplary only, and the true scope and spirit of this disclosure are indicated by the claims.

Claims

1. A method for locating the root cause of a fault, characterized in that, The method, applied to a trained fault root cause localization model, includes: Extract fault features from the fault description information of the user for the target fault scenario, and identify the user intent from the fault description information; Based on the fault characteristics and the user intent, select key fault analysis nodes that are related to the target fault scenario from the candidate fault analysis nodes; Execute the fault analysis task associated with the key fault analysis node to perform root cause localization processing on the target fault scenario and obtain the root cause localization result for the target fault scenario.

2. The method according to claim 1, characterized in that, The step of extracting fault features from the user's fault description information for the target fault scenario includes: The fault description information is subjected to information standardization processing to obtain standardized fault description information; the information standardization processing includes at least one of the following: synonym replacement, misspelling replacement, and omission of modal particles; The fault features are extracted from the standardized fault description information.

3. The method according to claim 2, characterized in that, The step of selecting key fault analysis nodes that are related to the target fault scenario from the candidate fault analysis nodes based on the fault characteristics and the user intent includes: Based on the first correlation between the fault characteristics and the task information of the fault analysis tasks associated with the candidate fault analysis nodes, an initial fault analysis node is selected from the candidate fault analysis nodes. The key fault analysis nodes are selected from the initial fault analysis nodes based on the second relevance between the user intent and the task information of the fault analysis task associated with the initial fault analysis node.

4. The method according to claim 2, characterized in that, The step of selecting key fault analysis nodes that are related to the target fault scenario from the candidate fault analysis nodes based on the fault characteristics and the user intent further includes: Extract task keywords from the task information of the fault analysis task associated with each candidate fault analysis node; Based on the first relevance between the fault characteristics and the task keywords associated with the candidate fault analysis nodes, an initial fault analysis node is selected from the candidate fault analysis nodes; The key fault analysis nodes are selected from the initial fault analysis nodes based on the second relevance between the user intent and the task keywords associated with the initial fault analysis nodes.

5. The method according to claim 3 or 4, characterized in that, When there are multiple key fault analysis nodes, the execution of the fault analysis task associated with the key fault analysis nodes to perform root cause localization processing on the target fault scenario and obtain the root cause localization result for the target fault scenario includes: The multiple key fault analysis nodes are sorted according to the second relevance to obtain a sorting sequence; According to the sorting order in the sorting sequence, the fault analysis tasks associated with each of the key fault analysis nodes are executed sequentially to perform fault root cause localization processing on the target fault scenario and obtain the root cause localization result for the target fault scenario.

6. The method according to claim 5, characterized in that, The step of sequentially executing the fault analysis tasks associated with each key fault analysis node according to the sorting order in the sorting sequence to perform fault root cause localization processing on the target fault scenario and obtain the root cause localization result for the target fault scenario includes: According to the sorting order in the sorting sequence, the fault analysis tasks associated with each of the key fault analysis nodes are executed sequentially to perform fault root cause localization processing on the target fault scenario and obtain the analysis results corresponding to each of the key fault analysis nodes. Extract fault elements from the fault description information; the fault elements include the fault entity, fault time, and fault manifestation. Based on the analysis results corresponding to each of the key fault analysis nodes and the fault elements, the root cause localization result for the target fault scenario is determined.

7. The method according to claim 1, characterized in that, The trained fault cause localization model is obtained through the following method: Obtain a training sample set; the training sample set contains multiple training samples, different training samples correspond to different fault scenario samples, each training sample contains a fault description information sample for each fault scenario sample and a root cause localization result label corresponding to each fault scenario sample; each fault scenario sample is associated with a pre-constructed output guide word; Each training sample is input into the initial fault root cause localization model, and the output guide word guides the initial fault root cause localization model to output the initial root cause localization result for each fault scenario sample; The output guide word is adjusted based on the degree of difference between the initial root cause localization result and the root cause localization result label; Each training sample is updated using the adjusted output guide word, and the initial fault root cause localization model is fine-tuned based on each updated training sample to obtain the trained fault root cause localization model.

8. The method according to claim 7, characterized in that, The output guidance words associated with each fault scenario sample include node filtering guidance words and result generation guidance words; The step of guiding the initial fault root cause localization model to output initial root cause localization results for each fault scenario sample through the output guide word includes: The node filtering guides the initial fault root cause localization model to select target fault analysis nodes that are associated with each fault scenario sample from the candidate fault analysis nodes. The initial fault root cause localization model is used to execute the fault analysis task associated with the target fault analysis node to perform fault root cause localization processing on each fault scenario sample, and obtain the analysis results corresponding to the target fault analysis node. The results generate guiding words to guide the initial fault root cause localization model to output initial root cause localization results for each fault scenario sample based on the analysis results corresponding to the target fault analysis node.

9. The method according to claim 8, characterized in that, The step of guiding the initial fault root cause localization model to select target fault analysis nodes that are correlated with each fault scenario sample from the candidate fault analysis nodes by using the node selection guide words includes: The initial fault root cause localization model extracts fault feature samples from fault description information samples for each fault scenario sample, and identifies user intent samples from the fault description information samples. The node selection guides the initial fault root cause localization model to select target fault analysis nodes that are associated with each fault scenario sample from the candidate fault analysis nodes based on the fault feature samples and the user intent samples.

10. The method according to any one of claims 7 to 9, characterized in that, After guiding the initial fault root cause localization model to output initial root cause localization results for each fault scenario sample via the output guide word, the method further includes: Obtain the historical root cause localization results for each fault scenario sample; The initial root cause localization results are expanded and optimized using the historical root cause localization results.

11. A fault root cause location device, characterized in that, The device is equipped with a trained fault root cause localization model, and the device includes: The information processing module is used to extract fault features from the fault description information of the user for the target fault scenario, and to identify the user's intent from the fault description information; The node filtering module is used to filter key fault analysis nodes that are related to the target fault scenario from the candidate fault analysis nodes based on the fault characteristics and the user intent. The fault root cause localization module is used to execute the fault analysis task associated with the key fault analysis node to perform fault root cause localization processing on the target fault scenario and obtain the root cause localization result for the target fault scenario.

12. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the fault root cause localization method according to any one of claims 1 to 10.

13. An electronic device, characterized in that, include: processor; as well as Memory for storing the executable instructions of the processor; The processor is configured to execute the fault root cause localization method according to any one of claims 1 to 10 by executing the executable instructions.

Images