Fault analysis method and device, storage medium and processor

Abstract

The embodiment of the invention provides a fault analysis method and device, a storage medium and a processor, and the method comprises the steps: obtaining operation and maintenance data and real-time alarm data, carrying out the data processing and cleaning of the operation and maintenance data, and obtaining target operation and maintenance data, the target operation and maintenance data comprises configuration management data, an emergency plan, index data and historical fault data; setting a cue word template according to user requirements, inputting the real-time alarm data into the cue word template to obtain an updated cue word template, and filling the updated cue word template with the target operation and maintenance data to obtain a target problem; constructing a large language model and a standard problem according to the historical fault data; and substituting the target problem and the standard problem into the large language model to obtain a fault analysis result. According to the method, efficient and reliable fault analysis is realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of computer technology and fault analysis, and in particular to a fault analysis method and device, a storage medium and a processor. BACKGROUND

[0002] Currently, with the wide application of new technologies such as virtualization and cloud computing, the scale of IT infrastructure in enterprise data centers has grown exponentially, and the scale of computer hardware and software has also expanded, resulting in frequent computer failures. Frontline maintenance personnel urgently need more professional and powerful maintenance tools.

[0003] In daily maintenance work in a data center, it is generally necessary to write emergency plan schemes for maintenance software and hardware, that is, to write operation steps that can handle possible faults of maintenance software and hardware in advance. When a fault occurs, maintenance personnel handle the fault according to the scheme description in the emergency plan. In actual maintenance work, an alarm is a warning means for exceeding the monitoring threshold of various software and hardware, and the alarm is usually the most direct and fastest way to reflect a fault.

[0004] Existing methods for analyzing faults mainly rely on manual or manually preset rules, and the analysis process is mainly carried out by manpower, which is costly and the experience is not easy to transfer, and new faults cannot be analyzed or reasonable suggestions cannot be given in time. SUMMARY

[0005] The purpose of the embodiments of the present application is to provide a fault analysis method, device, storage medium and processor. The method realizes efficient and reliable fault analysis.

[0006] To achieve the above purpose, the first aspect of the present application provides a fault analysis method, which comprises: Obtaining maintenance data and real-time alarm data, performing data processing and cleaning on the maintenance data to obtain target maintenance data, and the target maintenance data comprising configuration management data, emergency plans, index data and historical fault data; According to the user demand, a prompt word template is set, the real-time alarm data is input into the prompt word template to obtain an updated prompt word template, and the target maintenance data is filled into the updated prompt word template to obtain a target problem; According to the historical fault data, a large language model and a standard problem are constructed; The target problem and the standard problem are input into the large language model to obtain a fault analysis result.

[0007] Optionally, the data processing comprises cold and hot data separation and / or index addition on the maintenance data; The cleaning includes processing missing values, removing duplicate data, correcting data errors, unifying data formats, and cleaning noise data.

[0008] Optionally, the bringing the target question and the standard question into the large language model to obtain the fault analysis result comprises: acquiring a length of the target question, and determining a questioning manner to the large language model according to the length; obtaining the fault analysis result according to the questioning manner, the target question, the standard question and the large language model.

[0009] Optionally, the determining the questioning manner to the large language model according to the length comprises: if the length does not exceed a question length threshold of the large language model, obtaining the fault analysis result by questioning the large language model with the target question; if the length exceeds the question length threshold of the large language model, obtaining a modified question by modifying the target question with the standard question, and obtaining the fault analysis result by questioning the large language model with the modified question.

[0010] Optionally, the large language model is a matrix, and the method further comprises: setting a bypass parameter system, dimensions of the bypass parameter system are and respectively; transforming the target question and the standard question into an input tensor X, and bringing the input tensor X into the large language model to obtain a tensor H1; training the bypass parameter system according to the historical fault data, and bringing the target question and the standard question into the trained bypass parameter system to obtain a tensor H2; determining the fault analysis result according to the tensor H1 and the tensor H2.

[0011] Optionally, the determining the fault analysis result according to the tensor H1 and the tensor H2 comprises: acquiring a total tensor H of the tensor H1 and the tensor H2: H = H1 + H2; predicting an output text according to the total tensor H, and determining the fault analysis result according to the output text.

[0012] Optionally, the method further comprises: when an alarm occurs, determining a system to which the alarm belongs, a deployment unit to which the alarm belongs and an alarm type of the alarm according to attributes of the alarm, and screening the system to which the alarm belongs, the deployment unit to which the alarm belongs and the alarm type of the alarm to obtain real-time alarm data.

[0013] The second aspect of the application provides a device for fault analysis, and the device comprises: The acquisition module is configured to acquire operation and maintenance data and real-time alarm data, perform data processing and cleaning on the operation and maintenance data to obtain target operation and maintenance data, and the target operation and maintenance data includes configuration management data, emergency plans, index data, and historical fault data. The first processing module is configured to set a prompt word template according to user requirements, input the real-time alarm data into the prompt word template to obtain an updated prompt word template, and fill the target operation and maintenance data into the updated prompt word template to obtain a target problem. The second processing module is configured to construct a large language model and a standard question according to the historical fault data. The third processing module is configured to input the target problem and the standard question into the large language model to obtain a fault analysis result.

[0014] The third aspect of the present application provides a processor configured to execute the above-mentioned fault analysis method.

[0015] The fourth aspect of the present application provides a machine-readable storage medium having instructions stored thereon, which, when executed by a processor, cause the processor to be configured to execute the above-mentioned fault analysis method.

[0016] The fifth aspect of the present application provides a computer program product comprising a computer program which, when executed by a processor, implements the above-mentioned fault analysis method.

[0017] Through the above technical solution, the present application establishes a flexible and controllable prompt word engineering, which can modify the prompt word template according to the actual data of the operation and maintenance big data platform at any time to increase the answer quality of the large language model; suitable question sentences are pre-processed by vectorization and stored in a database, and the large language model is used to analyze real-time faults to give an analysis scheme to locate the root cause of the fault, assisting operation and maintenance personnel to quickly locate and handle faults.

[0018] Other features and advantages of the embodiments of the present application will be described in detail in the following specific embodiments. BRIEF DESCRIPTION OF DRAWINGS

[0019] The accompanying drawings are used to provide a further understanding of the embodiments of the present application, and constitute a part of the specification, and are used to explain the embodiments of the present application together with the following specific embodiments, but do not constitute a limitation on the embodiments of the present application. In the drawings: Figure 1 The flowchart of the method for fault analysis according to the embodiments of the present application is schematically shown; Figure 2 The structure diagram of the prompt word engineering according to the embodiments of the present application is schematically shown; Figure 3A schematic diagram of automatic fault analysis is shown according to the embodiment of the application. Figure 4 A structure diagram of large language model fine-tuning is shown according to the embodiment of the application. Figure 5 A schematic diagram of a device for fault analysis is shown according to the embodiment of the application. Figure 6 A schematic diagram of the internal structure of a computer device is shown according to the embodiment of the application.

[0020] Explanation of reference signs 100 - device for fault analysis; 200 - acquisition module; 300 - first processing module; 400 - second processing module; 500 - third processing module. DETAILED DESCRIPTION

[0021] In order to make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application. It should be understood that the specific embodiments described herein are only used to illustrate and explain the embodiments of the present application, and are not used to limit the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without making creative efforts fall within the scope of protection of the present application.

[0022] It should be noted that if the embodiments of the present application involve directional indications (such as up, down, left, right, front, back, etc.), the directional indications are only used to explain the relative positional relationship, movement condition, etc. between components in a certain posture (as shown in the drawings), and if the certain posture changes, the directional indications also change accordingly.

[0023] In addition, if the embodiments of the present application involve descriptions such as "first", "second", etc., the descriptions of "first", "second", etc. are only for description purposes, and cannot be understood as indicating or implying the relative importance of the indicated technical features or implicitly indicating the number of the indicated technical features. Therefore, the features limited by "first", "second" can explicitly or implicitly include at least one of the features. In addition, the technical solutions of each embodiment can be combined with each other, but it must be based on the realization of those of ordinary skill in the art, and when the combination of technical solutions contradicts each other or cannot be realized, it should be considered that the combination of technical solutions does not exist and is not within the scope of protection claimed by the present application.

[0024] The acquisition, transmission, storage, use, processing and the like of data in the technical solutions of the present application comply with relevant provisions of national laws and regulations. In addition, it should be noted that in the embodiments of the present application, some industry existing solutions such as software, components, models and the like may be mentioned, which should be considered as exemplary, and the purpose is only to illustrate the feasibility of the implementation of the technical solutions of the present application, but it does not mean that the applicant has or will necessarily use the solutions.

[0025] Figure 1 The flowchart of the method for fault analysis according to the embodiments of the present application is schematically shown. As shown in Figure 1 and Figure 2 In an embodiment of the present application, a method for fault analysis is provided, which comprises: Step S101 is to acquire operation and maintenance data and real-time alarm data, to perform data processing and cleaning on the operation and maintenance data to obtain target operation and maintenance data, the target operation and maintenance data including configuration management data, emergency plan, index data and historical fault data.

[0026] According to a specific implementation, the data processing includes cold and hot data separation and / or index addition on the operation and maintenance data; and the cleaning includes missing value processing, duplicate data removal, data error correction, data format unification and noise data cleaning.

[0027] The method further comprises: when an alarm occurs, determining the system to which the alarm belongs, the deployment unit to which the alarm belongs and the alarm type through the attributes of the alarm, and screening the system to which the alarm belongs, the deployment unit to which the alarm belongs and the alarm type to obtain real-time alarm data.

[0028] Specifically, the operation and maintenance data and real-time alarm data relied on by the present application are relatively large, and need to be supported by an operation and maintenance big data platform. At the same time, some basic algorithm models are needed to provide example support. Generally, open source big data components (such as hdfs, kudu, hive, elasticsearch, etc.) are used to store massive operation and maintenance related data (such as configuration management data, knowledge base data, alarm data, log data, index data and emergency plan data, etc.). Because the data is relatively large, cold and hot data separation and index addition may be considered; if the data is messy and missing, data cleaning may be needed. At the same time, the operation and maintenance big data platform provides a method for cleaning data.

[0029] The selection of the historical fault data includes the corresponding alarm, surrounding data (alarms, indexes, work orders, etc.) and corresponding emergency plans (manually screened) when the historical fault occurs. The fault pointer (data center) in the present application is a software and hardware error on a computer device.

[0030] Step S102 is to set a prompt word template according to user needs, input the real-time alarm data into the prompt word template to obtain an updated prompt word template, and fill the target operation and maintenance data into the updated prompt word template to obtain a target problem.

[0031] According to a specific embodiment, when asking a large language model, a prompt word language needs to be constructed to obtain better feedback results. The present application can expand the prompt word through rich data on the operation and maintenance big data platform. The more detailed the prompt word is, the better the result will be. The system can also automatically obtain the required data according to the user-defined prompt word template and fill it into the prompt word template.

[0032] As shown in Figure 3 , a single real-time alarm is input into the prompt word template. According to the user's pre-selected prompt word template, the data of the operation and maintenance big data platform is called as needed to fill in the elements, and finally a complete question is spliced out and submitted to the large language model.

[0033] For example, the prompt word template can be: assuming that the user is a senior operation and maintenance expert of the ${Application} system and is very familiar with the business logic of the deployment unit ${Deploy}. When the user receives the following ${AlarmType} alarm: ${AlarmContent}, and the emergency plan associated with the alarm is ${EmergencyPlan}, and there are some related data: ticket ${ticket}, CPU index ${}, the user analyzes the fault caused by the alarm and finds the root cause.

[0034] After filling in the elements of the above template through the operation and maintenance big data platform, the question becomes: assuming that the user is a senior operation and maintenance expert of the ${Application} system and is very familiar with the business logic of the deployment unit ${Deploy}. When the user receives the following ${AlarmType} alarm: ${AlarmContent}, and the emergency plan associated with the alarm is ${EmergencyPlan}, and there are some related data: ticket ${ticket}, CPU index ${}, the user analyzes the fault caused by the alarm and finds the root cause.

[0035] Step S103 is to construct a large language model and a standard question according to the historical fault data.

[0036] According to a specific embodiment, the large language model is a matrix, and the method further comprises: setting a bypass parameter system, the dimensions of the bypass parameter system are and ; converting the target question and the standard question into an input tensor X, and bringing the input tensor X into the large language model to obtain a tensor H1; training the bypass parameter system according to the historical fault data, and bringing the target question and the standard question into the trained bypass parameter system to obtain a tensor H2; and determining a fault analysis result according to the tensor H1 and the tensor H2.

[0037] Specifically, the determining of the fault analysis result according to the tensor H1 and the tensor H2 comprises: obtaining a total tensor H of the tensor H1 and the tensor H2: H = H1 + H2; predicting an output text according to the total tensor H; and determining the fault analysis result according to the output text.

[0038] The large language model (Large Language Model (LLM)) is a large parameter model trained by a transformer model, and can be efficiently used to solve logical reasoning, text generation and other tasks in natural language processing. The transformer model is a new type of machine learning model, which can better utilize the foregoing data.

[0039] When asking a large language model a question and obtaining an answer, if the large language model can be fine-tuned with positive sample data provided by the operation and maintenance vertical field, the answer will be more in line with expectations. However, the large language model is called large because the parameter quantity of the model can reach hundreds of billions or even trillions, and on this basis, fine-tuning the model parameters requires huge hardware resources, which is not very realistic. Therefore, a bypass parameter system with a smaller parameter quantity is designed, and after the bypass parameter system is trained, the output result of the bypass parameter system is superimposed on the result of the large language model to generate a result adapted to the positive sample data in the operation and maintenance field.

[0040] As shown in Figure 4 , the tensor X is a vector converted from input text, and the dimension is m; the tensor H is an output result, which is converted by a softmax function and can correspond to specific text; m and n are very large and can reach tens of thousands or even hundreds of thousands, which are consistent with the dimensions of the large language model on the left; Figure 4 ; Figure 4 The right side is the bypass parameter system, which contains two matrices A and B, and the dimensions are and , m and n are consistent with the foregoing, r is a number, which can be set to 2 or 3, and is usually not more than 10, so that the matrices A and B are two parameter models with small quantities (very wide and short matrices). The tensor is a dimensional space, and a certain object is mathematically represented as a set of numbers or a one-dimensional matrix.

[0041] The specific use steps are as follows: a plurality of positive sample data, i.e. historical fault related data, are manually screened; the model is trained based on the positive sample data, which is actually to train the bypass parameter system on the right; after the bypass parameter system is trained, when a question enters, it is first converted into an input tensor X; the tensor X enters the original large language model for calculation to obtain a tensor H1; the tensor X enters the bypass parameter system and is calculated with the parameter matrix B and the parameter matrix A respectively to obtain a tensor H2; the tensor H = H1 + H2 is calculated, and the tensor H is used to calculate the predicted output text to obtain the returned result. The historical fault positive sample data and the fine-tuned large language model make the answer result more effective. The positive sample is data that has a positive effect on model training, and the present application refers to historical faults and related peripheral data.

[0042] Step S104 is to bring the target question and the standard question into the large language model to obtain a fault analysis result.

[0043] According to a specific embodiment, the step of bringing the target question and the standard question into the large language model to obtain a fault analysis result comprises: acquiring the length of the target question, determining the questioning manner to the large language model according to the length; and obtaining a fault analysis result according to the questioning manner, the target question, the standard question and the large language model.

[0044] Specifically, the step of determining the questioning manner to the large language model according to the length comprises: if the length does not exceed the question length threshold of the large language model, asking the large language model the target question to obtain a fault analysis result; and if the length exceeds the question length threshold of the large language model, modifying the target question into a modified question by using the standard question, and asking the large language model the modified question to obtain a fault analysis result. The fault analysis result is sent to a front-line operation and maintenance personnel to assist them to quickly solve the fault.

[0045] The present application establishes a flexible and controllable prompt word engineering, which can modify the prompt word template according to the actual data of the operation and maintenance big data platform at any time to increase the answer quality of the large language model; appropriate question sentences are pre-processed into vectors and stored in a database, and the large language model is used to analyze real-time faults to give an analysis scheme to locate the root cause of the fault, and assist operation and maintenance personnel to quickly locate and handle faults.

[0046] In one embodiment, as Figure 5As shown, a fault analysis device 100 is provided, which comprises: an acquisition module 200, configured to acquire operation and maintenance data and real-time alarm data, perform data processing and cleaning on the operation and maintenance data to obtain target operation and maintenance data, and the target operation and maintenance data comprises configuration management data, emergency plan, index data and historical fault data; a first processing module 300, configured to set a prompt word template according to user demand, input the real-time alarm data into the prompt word template to obtain an updated prompt word template, and fill the updated prompt word template with the target operation and maintenance data to obtain a target problem; a second processing module 400, configured to construct a large language model and a standard question according to the historical fault data; and a third processing module 500, configured to input the target problem and the standard question into the large language model to obtain a fault analysis result. The device realizes efficient and reliable fault analysis.

[0047] The fault analysis method of the present application comprises: acquiring operation and maintenance data and real-time alarm data, performing data processing and cleaning on the operation and maintenance data to obtain target operation and maintenance data, and the target operation and maintenance data comprises configuration management data, emergency plan, index data and historical fault data; setting a prompt word template according to user demand, inputting the real-time alarm data into the prompt word template to obtain an updated prompt word template, and filling the updated prompt word template with the target operation and maintenance data to obtain a target problem; constructing a large language model and a standard question according to the historical fault data; and inputting the target problem and the standard question into the large language model to obtain a fault analysis result. The present application establishes a flexible and controllable prompt word engineering, modifies the prompt word template according to the actual data of the operation and maintenance big data platform at any time to increase the answer quality of the large language model, pre-processes suitable question sentences into vectors, stores them in a database, uses the large language model to analyze real-time faults, gives an analysis scheme to locate fault root causes, and assists operation and maintenance personnel to quickly locate faults and handle faults.

[0048] The fault analysis device 100 comprises a processor and a memory, and the above acquisition module 200, first processing module 300, second processing module 400 and third processing module 500 are stored in the memory as program units, and the corresponding functions are realized by the processor executing the above program modules stored in the memory.

[0049] The processor contains a core, and the core calls the corresponding program units from the memory. The core can be set to one or more, and the core parameters are adjusted to realize the fault analysis method.

[0050] The memory can include a non-permanent memory in a computer readable medium, a random access memory (RAM) and / or a non-volatile memory such as a read-only memory (ROM) or a flash memory (flash RAM), and the memory comprises at least one memory chip.

[0051] This application provides a storage medium storing a program that, when executed by a processor, implements the aforementioned fault analysis method.

[0052] This application provides a processor for running a program, wherein the program executes the above-described fault analysis method during runtime.

[0053] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 6 As shown. The computer device includes a processor A01, a network interface A02, memory (not shown), and a database (not shown) connected via a system bus. The processor A01 provides computing and control capabilities. The memory includes internal memory A03 and a non-volatile storage medium A04. The non-volatile storage medium A04 stores an operating system B01, a computer program B02, and a database (not shown). The internal memory A03 provides an environment for the operation of the operating system B01 and the computer program B02 stored in the non-volatile storage medium A04. The database stores fault analysis data. The network interface A02 communicates with external terminals via a network connection. When the computer program B02 is executed by the processor A01, it implements a fault analysis method.

[0054] Those skilled in the art will understand that the structure shown in Figure Y is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or may combine certain components, or may have different component arrangements.

[0055] In one embodiment, the fault analysis apparatus provided in this application can be implemented as a computer program, which can be implemented in, for example... Figure 6 The fault analysis apparatus runs on the computer device shown. The computer device's memory can store various program modules that constitute the fault analysis device. The computer program, composed of the various program modules, causes the processor to execute the steps of the fault analysis methods described in the various embodiments of this application.

[0056] The embodiment of the present application provides a device, the device comprises a processor, a memory and a program stored on the memory and executable on the processor, and the processor implements the following steps when executing the program: obtaining operation and maintenance data and real-time alarm data, performing data processing and cleaning on the operation and maintenance data to obtain target operation and maintenance data, the target operation and maintenance data comprising configuration management data, emergency plan, index data and historical fault data; setting a prompt word template according to user demand, inputting the real-time alarm data into the prompt word template to obtain an updated prompt word template, and filling the updated prompt word template with the target operation and maintenance data to obtain a target problem; constructing a large language model and a standard problem according to the historical fault data; and bringing the target problem and the standard problem into the large language model to obtain a fault analysis result.

[0057] The present application also provides a computer program product adapted to execute a program that initializes the following method steps when executed on a data processing device: obtaining operation and maintenance data and real-time alarm data, performing data processing and cleaning on the operation and maintenance data to obtain target operation and maintenance data, the target operation and maintenance data comprising configuration management data, emergency plan, index data and historical fault data; setting a prompt word template according to user demand, inputting the real-time alarm data into the prompt word template to obtain an updated prompt word template, and filling the updated prompt word template with the target operation and maintenance data to obtain a target problem; constructing a large language model and a standard problem according to the historical fault data; and bringing the target problem and the standard problem into the large language model to obtain a fault analysis result.

[0058] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can adopt a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can adopt a computer program product embodied on one or more computer usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer usable program code.

[0059] The present application is described with reference to flowcharts and / or block diagrams according to the embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of flows and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to produce a machine, so that the instructions executed by the computer or other programmable data processing devices produce a device that implements the functions specified in the flowcharts and / or block diagrams. Figure 1 The function of one flow or multiple flows and / or blocks Figure 1 The function of one flow or multiple flows and / or blocks

[0060] These computer program instructions can also be stored in a computer readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer readable memory produce an article of manufacture including instructions which implement the flow Figure 1 one or more flow or block Figure 1 one or more blocks or steps of the flow.

[0061] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions that are executed on the computer or other programmable apparatus provide steps for implementing the flow Figure 1 one or more flow or block Figure 1 one or more blocks or steps of the flow.

[0062] In one typical configuration, the computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.

[0063] The memory can include non-persistent memory and / or volatile memory, such as random access memory (RAM) and / or cache memory. The memory can also include non-volatile memory, such as read only memory (ROM), EPROM, and / or flash memory. The memory can be volatile, non-volatile, or a combination of volatile and non-volatile memory. The memory can also be another form of computer-readable media.

[0064] Computer-readable media includes permanent and non-permanent, removable and non-removable media, which can be implemented by any method or technology for storage of information. The information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read only memory (ROM), electrically erasable programmable read only memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassette, magnetic disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information accessible to a computing device. According to the definition herein, computer-readable media does not include transitory media, such as modulated data signals and carrier waves.

[0065] It should also be noted that the terms "comprising", "comprises" or other variations thereof are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can also include other elements not expressly listed or inherent to such process, method, article, or apparatus.

[0066] The above embodiments are only used to illustrate the present application, but not to limit it. Instead of the above, various modifications and changes can be made to the application by those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the application shall fall into the scope of the claims of the application.

Claims

1. A method of failure analysis, characterized by, The method of the fault analysis comprises: Obtaining operation and maintenance data and real-time alarm data, performing data processing and cleaning on the operation and maintenance data to obtain target operation and maintenance data, the target operation and maintenance data comprising configuration management data, emergency plan, index data and historical fault data; Setting a prompt word template according to user demand, inputting the real-time alarm data into the prompt word template to obtain an updated prompt word template, and filling the updated prompt word template with the target operation and maintenance data to obtain a target question; Constructing a large language model and a standard question according to the historical fault data; Bringing the target question and the standard question into the large language model to obtain a fault analysis result.

2. The method of claim 1, wherein: The data processing comprises cold and hot data separation and / or index addition on the operation and maintenance data; The cleaning comprises processing missing values, removing duplicate data, correcting data errors, unifying data formats and removing noise data.

3. The method of claim 1, wherein, The bringing of the target question and the standard question into the large language model to obtain the fault analysis result comprises: Obtaining the length of the target question, and determining a questioning mode to the large language model according to the length; Obtaining the fault analysis result according to the questioning mode, the target question, the standard question and the large language model.

4. The method of claim 3, wherein, The determining of the questioning mode to the large language model according to the length comprises: If the length does not exceed a question length threshold of the large language model, asking the large language model a question through the target question to obtain the fault analysis result; If the length exceeds the question length threshold of the large language model, correcting the target question through the standard question to obtain a corrected question, and asking the large language model a question through the corrected question to obtain the fault analysis result.

5. The method of claim 1, wherein, The large language model is a matrix, and the method further comprises: A bypass parameter system is provided, the dimension of the bypass parameter system is and ; Converting the target question and the standard question into an input tensor X, and bringing the input tensor X into the large language model to obtain a tensor H1; Training the bypass parameter system according to the historical fault data, and bringing the target question and the standard question into the trained bypass parameter system to obtain a tensor H2; Determining the fault analysis result according to the tensor H1 and the tensor H2.

6. The method of claim 5, wherein, The determining of the fault analysis result according to the tensor H1 and the tensor H2 comprises: Obtaining a total tensor H of the tensor H1 and the tensor H2: H = H1 + H2; Predicting an output text according to the total tensor H, and determining the fault analysis result according to the output text.

7. The method of claim 1, wherein, The method further comprises: When an alarm occurs, determining the system, the deployment unit and the alarm type to which the alarm belongs according to the attributes of the alarm, and screening the system, the deployment unit and the alarm type to which the alarm belongs to obtain real-time alarm data.

8. An apparatus for failure analysis, characterized by, The apparatus comprises: An obtaining module, configured to obtain operation and maintenance data and real-time alarm data, perform data processing and cleaning on the operation and maintenance data to obtain target operation and maintenance data, the target operation and maintenance data comprising configuration management data, emergency plan, index data and historical fault data; A first processing module, configured to set a prompt word template according to user demand, input the real-time alarm data into the prompt word template to obtain an updated prompt word template, and fill the updated prompt word template with the target operation and maintenance data to obtain a target question; A second processing module configured to construct a large language model and standard questions based on the historical failure data; A third processing module configured to input the target question and the standard question into the large language model to obtain a failure analysis result.

9. A processor, comprising: A computer program product comprising computer executable instructions that, when executed by a processor, cause the processor to perform the method of failure analysis according to any one of claims 1 to 7.

10. A machine-readable storage medium having stored thereon instructions, the instructions being executable by a machine to cause the machine to: The instructions, when executed by a processor, cause the processor to be configured to perform the method of failure analysis according to any one of claims 1 to 7.

11. A computer program product comprising a computer program, characterized in that, The computer program, when executed by a processor, implements the method of failure analysis according to any one of claims 1 to 7.

Images