Alarm processing method and device of network equipment, storage medium and electronic equipment

Abstract

The invention discloses an alarm processing method and device for network equipment, a storage medium and electronic equipment. The method comprises the following steps: receiving alarm information of network equipment, and judging whether the alarm information is matched with a translation rule in an alarm translation library or not; under the condition that the alarm information is matched with the target translation rule, target translation content is extracted from the target translation rule, and the target translation content is displayed on a front-end page; under the condition that the translation rule matched with the alarm information does not exist in the alarm translation library, the alarm information is input into a target model, predicted translation content is obtained, the target model is obtained through training of multiple sets of training samples, and each set of training samples comprise historical alarm information and historical translation content; and displaying the predicted translation content on a front-end page. Through the method and the device, the problem of low operation and maintenance efficiency of network equipment due to the fact that partial alarm information cannot be matched with the translation content in related technologies is solved.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence, and more specifically, to an alarm processing method, apparatus, storage medium, and electronic device for network devices. Background Technology

[0002] In related technologies, various alarms from network devices are first translated using multiple translation components before being displayed to frontline operations and maintenance personnel. This ensures a clearer and more concise presentation of the alarms, guaranteeing rapid response and intelligent operation and maintenance on the platform. With the rapid development of enterprises and the explosive growth in the number and types of network devices, alarm management for network devices has become increasingly important.

[0003] However, alarm information from some network devices cannot be matched with alarm translations. Due to the lack of preliminary processing of unknown alarm information, there is a situation where some alarms that do not need attention are over-focused, or alarms that need attention are ignored because they cannot be matched with alarm translations, which increases the pressure on operation and maintenance.

[0004] There is currently no effective solution to the problem of low network device maintenance efficiency caused by the inability to match some alarm information with translated content in related technologies. Summary of the Invention

[0005] The main objective of this application is to provide a method, apparatus, storage medium, and electronic device for alarm processing of network devices, so as to solve the problem of low operation and maintenance efficiency of network devices due to the inability to match some alarm information with the translation content in the related technology.

[0006] To achieve the above objectives, according to one aspect of this application, an alarm processing method for a network device is provided. The method includes: receiving alarm information from the network device and determining whether the alarm information matches a translation rule in an alarm translation library, wherein the alarm translation library contains multiple translation rules, each containing an alarm message and its translation content; if the alarm information matches a target translation rule, extracting the target translation content from the target translation rule and displaying the target translation content on a front-end page; if no matching translation rule exists in the alarm translation library, inputting the alarm information into a target model to obtain predicted translation content, wherein the target model is trained using multiple sets of training samples, each set of training samples including historical alarm information and historical translation content; and displaying the predicted translation content on the front-end page.

[0007] Optionally, the target model is trained as follows: obtain raw alarm information from various network devices, extract multiple historical alarm messages and the historical translation content of each historical alarm message from the raw alarm information, wherein the network devices include at least one of the following: router, switch, server, and firewall; determine a set of training samples based on each historical alarm message and the historical translation content of the historical alarm message to obtain multiple sets of training samples; train an initial natural language processing model based on the multiple sets of training samples to obtain the target model.

[0008] Optionally, before training the initial natural language processing model based on multiple sets of training samples, the method further includes: preprocessing the original alarm information to obtain preprocessed alarm data, wherein the preprocessing includes at least one of the following: deduplication, noise reduction, text cleaning, and word segmentation encoding; converting the preprocessed alarm data into a word vector sequence using a decoder, and calculating the attention weight of each word vector in the word vector sequence using an attention mechanism; weighting the corresponding word vectors in the word vector sequence based on the attention weight of each word vector to obtain an updated word vector sequence; and determining the initial natural language processing model based on the updated word vector sequence.

[0009] Optionally, after displaying the predicted translation content on the front-end page, the method further includes: receiving user feedback on the predicted translation content, wherein the feedback includes whether the prediction is accurate or inaccurate; if the feedback is that the prediction is accurate, adding the alarm information and the predicted translation content to the alarm translation library; if the feedback is that the prediction is inaccurate, repeatedly executing the step of inputting the alarm information into the target model to obtain the predicted translation content until the number of executions is greater than or equal to the execution count threshold, or the feedback is that the prediction is accurate; if the number of executions is greater than or equal to the execution count threshold, and the feedback result of the predicted translation content after each execution is that the prediction is inaccurate, issuing a prompt message, wherein the prompt message is used to prompt the operation and maintenance personnel to handle the alarm information.

[0010] Optionally, the method further includes: collecting feedback results of all predicted translation content output by the target model at preset intervals; determining the total number of feedback results of all predicted translation content, and determining the target number of accurate predicted translation content based on the feedback results; calculating the ratio of the target number to the total number, and determining whether the ratio is greater than or equal to a preset accuracy threshold; if the ratio is less than the accuracy threshold, determining the feedback result as an inaccurate prediction alarm message, and obtaining the actual translation content of the alarm message; updating the target model based on the inaccurate prediction alarm message and the actual translation content.

[0011] Optionally, updating the target model based on inaccurate prediction alert information and actual translated content includes: identifying inaccurate prediction alert information and actual translated content as new training samples; adding the new training samples to multiple sets of training samples to obtain updated multiple sets of training samples; and training the initial natural language processing model based on the updated multiple sets of training samples to obtain the updated target model.

[0012] Optionally, before determining whether the alarm information matches the translation rules in the alarm translation library, the method further includes: obtaining device information for each network device, extracting alarm information and translated content of the alarm information from the device information; if the device information does not provide alarm information or translated content of the alarm information, obtaining historical alarm processing records of the network device, extracting alarm information and translated content of the alarm information from the historical alarm processing records; and constructing an alarm translation library based on the alarm information and translated content of the alarm information for each network device.

[0013] To achieve the above objectives, according to another aspect of this application, an alarm processing apparatus for a network device is provided. The apparatus includes: a receiving unit, configured to receive alarm information from the network device and determine whether the alarm information matches a translation rule in an alarm translation library, wherein the alarm translation library contains multiple translation rules, each translation rule containing an alarm information and its translation content; an extraction unit, configured to extract target translation content from the target translation rule when the alarm information matches the target translation rule, and display the target translation content on a front-end page; an input unit, configured to input the alarm information into a target model to obtain predicted translation content when no matching translation rule exists in the alarm translation library, wherein the target model is trained from multiple sets of training samples, each set of training samples including historical alarm information and historical translation content; and a display unit, configured to display the predicted translation content on the front-end page.

[0014] In this embodiment, alarm information from network devices is received, and it is determined whether the alarm information matches a translation rule in an alarm translation library. The alarm translation library contains multiple translation rules, each containing a type of alarm information and its translation content. If the alarm information matches a target translation rule, the target translation content is extracted from the target translation rule and displayed on the front-end page. If no matching translation rule exists in the alarm translation library, the alarm information is input into a target model to obtain a predicted translation content. The target model is trained using multiple sets of training samples, each set including historical alarm information and historical translation content. By displaying the predicted translation content on the front-end page, the alarm translation library is used to match the translation content of the alarm information. Furthermore, if no matching translation content is found in the alarm translation library, the target model outputs the predicted translation content for the alarm information. This achieves the goal of timely parsing of alarm information and providing translation content, thereby improving the operational efficiency of network devices and solving the technical problem of low operational efficiency due to the inability to match translation content for some alarm information. Attached Figure Description

[0015] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:

[0016] Figure 1 A hardware structure block diagram of a computer terminal (or mobile device) for implementing an alarm processing method for network devices is shown.

[0017] Figure 2 This is a flowchart of an alarm processing method for a network device according to an embodiment of this application;

[0018] Figure 3 This is a schematic diagram illustrating the construction process of the target model provided in the embodiments of this application;

[0019] Figure 4 This is a schematic diagram of the preprocessing flow of alarm information provided according to the embodiments of this application;

[0020] Figure 5 This is a schematic diagram of the processing flow of feedback results provided in the embodiments of this application;

[0021] Figure 6 This is a schematic diagram of an alarm processing system for a network device provided according to an embodiment of this application;

[0022] Figure 7 This is a schematic diagram of an alarm processing device for a network device according to an embodiment of this application;

[0023] Figure 8 This is a structural block diagram of an electronic device according to an embodiment of this application. Detailed Implementation

[0024] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0025] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0026] It should be noted that the information collected in this application (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for display, data used for analysis, etc.) are information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, storage, use, processing, transmission, provision, disclosure, and application of this data all comply with relevant laws, regulations, and standards, necessary confidentiality measures have been taken, and they do not violate public order and good morals. Corresponding access points are provided for users to choose to authorize or refuse. For example, interfaces are set up between this system and relevant users or organizations, providing users with corresponding access points to choose to agree to or refuse automated decision-making results; if the user chooses to refuse, the process proceeds to the expert decision-making stage.

[0027] Example 1

[0028] According to an embodiment of this application, an embodiment of a method for alarm processing of a network device is also provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0029] The method embodiment provided in Embodiment 1 of this application can be executed on a mobile terminal, computer terminal, or similar computing device. Figure 1 A hardware block diagram of a computer terminal (or mobile device) for implementing an alarm processing method for network devices is shown. Figure 1 As shown, the computer terminal 10 (or mobile device) may include one or more processors 102 (shown as 102a, 102b, ..., 102n in the figure) 102 (processor 102 may include, but is not limited to, processing devices such as MCU (Microcontroller Unit) or FPGA (Field-Programmable Gate Array), memory 104 for storing data, and transmission device 106 for communication functions. In addition, it may also include: a display, input / output interfaces (I / O interfaces), a USB (Universal Serial Bus) port (which may be included as one of the ports of a BUS (Business Bus), a network interface, a power supply, and / or a camera. Those skilled in the art will understand that... Figure 1 The structure shown is for illustrative purposes only and does not limit the structure of the aforementioned electronic device. For example, computer terminal 10 may also include... Figure 1 The more or fewer components shown, or having the same Figure 1 The different configurations shown.

[0030] It should be noted that the aforementioned one or more processors 102 and / or other data processing circuits are generally referred to herein as "data processing circuits". These data processing circuits may be embodied, in whole or in part, in software, hardware, firmware, or any other combination thereof. Furthermore, the data processing circuits may be a single, independent processing module, or may be integrated, in whole or in part, into any other element within the computer terminal 10 (or mobile device). As involved in the embodiments of this application, the data processing circuits serve as a processor control mechanism (e.g., selection of a variable resistor termination path connected to an interface).

[0031] The memory 104 can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the alarm processing method of the network device in this embodiment of the application. The processor 102 executes various functional applications and data processing by running the software programs and modules stored in the memory 104, thereby realizing the alarm processing method of the network device described above. The memory 104 may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include memory remotely located relative to the processor 102, and these remote memories can be connected to the computer terminal 10 via a network. Examples of the above-mentioned networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0032] The transmission device 106 is used to receive or send data via a network. Specific examples of the network described above may include a wireless network provided by the communication provider of the computer terminal 10. In one example, the transmission device 106 includes a Network Interface Controller (NIC), which can connect to other network devices via a base station to communicate with the Internet. In another example, the transmission device 106 may be a Radio Frequency (RF) module, used for wireless communication with the Internet.

[0033] The display may be, for example, a touchscreen LCD display that allows the user to interact with the user interface of the computer terminal 10 (or mobile device).

[0034] Under the above operating environment, this application provides a method for alarm processing of network devices. Figure 2 This is a flowchart of an alarm processing method for a network device according to an embodiment of this application, such as... Figure 2 As shown, the method includes:

[0035] Step S201: Receive alarm information from network devices and determine whether the alarm information matches the translation rules in the alarm translation library. The alarm translation library contains multiple translation rules, and each translation rule contains an alarm message and its translation content.

[0036] In step S201, the network device can be a router, switch, server, etc. When an abnormal state of the network device is detected, the network device will generate alarm information. The alarm information may contain key data such as device status, error code, and timestamp. After receiving the alarm information, it is first necessary to parse the alarm information and transform it from the original format (such as JSON (JavaScript Object Notation, a lightweight data interchange format), XML (Extensible Markup Language, a markup language), or a specific text format) into analyzable structured data.

[0037] The system applying the alarm processing method provided in this embodiment can maintain an alarm translation library, which stores various translation rules. Each rule can include two parts: a trigger mode (i.e., characteristics of the alarm information, such as a specific error code or keyword) and translated content (i.e., a clearer and easier-to-understand alarm description). The system matches the parsed alarm information with the rules in the alarm translation library, checking whether the alarm information contains any trigger mode from the translation rules. When the alarm information matches a translation rule, the system outputs the translated content from that rule as the alarm explanation, providing it to the operations and maintenance personnel.

[0038] If the alarm message does not match any rule in the translation library, the system can mark the alarm message as "unknown" or "untranslated" and then proceed to the next step of processing.

[0039] Step S202: If the alarm information matches the target translation rule, extract the target translation content from the target translation rule and display the target translation content on the front-end page.

[0040] In step S202, after determining the matching target translation rule from the alarm translation library, the target translation content is extracted from the target translation rule. The target translation content can describe the meaning of the alarm information in clearer and more intuitive language, such as including the type of problem, possible causes, and handling suggestions. The target translation content is sent to the front-end interface to ensure that operations and maintenance personnel can immediately see the translated content when viewing alarm information.

[0041] Step S203: If there is no translation rule matching the alarm information in the alarm translation library, input the alarm information into the target model to obtain the predicted translation content. The target model is trained by multiple sets of training samples, and each set of training samples includes historical alarm information and historical translation content.

[0042] In step S203, if the alarm information does not match any rule in the alarm translation library, the unknown alarm information needs to be decrypted. The target model can learn the relevant semantic relationships in the alarm information by training a natural language processing model, thereby predicting the translation content of the unknown alarm information.

[0043] Before deploying the target model, a large amount of historical alarm information and its corresponding historical translations need to be collected to form multiple sets of training samples. The historical alarm information in the training samples can be the original alarm text generated by network devices, while the historical translations are alarm descriptions that are translated manually or according to existing rules to make them easier to understand. Based on this data, a natural language processing model is trained to learn the mapping relationship between alarm information and translations.

[0044] The prediction results from the target model, i.e., the predicted translation content, will be converted into a format that operations and maintenance personnel can understand and displayed on the front-end page. The predicted translation content can include the type of alarm, possible causes, scope of impact, and preliminary handling suggestions, providing operations and maintenance personnel with immediate information interpretation.

[0045] Step S204: Display the predicted translation on the front-end page.

[0046] In step S204, after receiving the returned predicted translation content, the returned data is parsed and converted into a format understandable by the front end. This includes steps such as JSON parsing and data format conversion to ensure the predicted translation content can be correctly displayed by the front-end component. The predicted translation content is then displayed through a pop-up alarm details panel or an additional information bar in the alarm list.

[0047] In the alarm information display area of ​​the front-end page, the original alarm information is displayed in its original format, maintaining its integrity and original format. In the predicted translation content display area, the predicted translation content is displayed in a more readable format, which can be distinguished using different colors, fonts, or icons, highlighting key translation results such as fault type, possible causes, and suggested actions.

[0048] The alarm processing method for network devices provided in this application receives alarm information from the network device and determines whether the alarm information matches a translation rule in an alarm translation library. The alarm translation library contains multiple translation rules, each containing a type of alarm information and its translation content. If the alarm information matches a target translation rule, the target translation content is extracted from the target translation rule and displayed on a front-end page. If no matching translation rule exists in the alarm translation library, the alarm information is input into a target model to obtain a predicted translation content. The target model is trained using multiple sets of training samples, each set including historical alarm information and historical translation content. The predicted translation content is displayed on the front-end page. By matching the alarm information with the translation content in the alarm translation library, and if no matching translation content is found in the library, the target model outputs the predicted translation content for the alarm information. This achieves the goal of timely parsing of alarm information and providing translation content, thereby improving the operational efficiency of network devices and solving the technical problem of low operational efficiency due to the inability to match translation content for some alarm information.

[0049] To translate unknown alarm information, a target model needs to be trained. Optionally, in the alarm processing method for network devices provided in this application embodiment, the target model is trained in the following manner: obtaining the original alarm information of various network devices; extracting multiple historical alarm messages and the historical translation content of each historical alarm message from the original alarm information; wherein the network devices include at least one of the following: router, switch, server, and firewall; determining a set of training samples based on each historical alarm message and the historical translation content of the historical alarm messages to obtain multiple sets of training samples; and training an initial natural language processing model based on the multiple sets of training samples to obtain the target model.

[0050] In some examples, a large amount of raw alarm information is collected from monitoring logs of various network devices (such as routers, switches, servers, firewalls, etc.). This information may contain key fields such as device status, error codes, and timestamps. For each collected historical alarm message, its corresponding translation in the alarm translation library is obtained, i.e., the historical translation content. The historical alarm information is preprocessed, including but not limited to word segmentation, stop word removal, and stemming, to extract its features. At the same time, the historical translation content is also preprocessed accordingly to ensure data consistency and cleanliness. The preprocessed historical alarm information and historical translation content are converted into a format that the model can understand, such as word vectors or sequence data.

[0051] Each preprocessed historical alert message is paired with its corresponding historical translation to form a set of training samples. Each set of samples contains two parts: input (i.e., the feature representation of the alert message) and output (i.e., the feature representation of the translated content). A natural language processing model is selected, its parameters are initialized, and the model is trained using the constructed sets of training samples. During training, the model learns the mapping relationship between historical alert messages and their translated content through forward and backward propagation. The goal is to minimize the difference between the predicted translated content and the actual historical translated content, and cross-entropy loss can be used as the optimization objective.

[0052] Evaluate the performance of the trained model on an independent validation set to check its accuracy and generalization ability. Based on the performance on the validation set, adjust the model's hyperparameters, such as learning rate, batch size, and number of network layers, to optimize model performance. Repeat the training, evaluation, and tuning process until the model converges or reaches the predetermined performance metrics. Once the model achieves satisfactory performance, save the model's weights and architecture, i.e., the target model. Deploy the trained target model to the production environment to meet the translation needs of unknown alarm information in real-time or batch processing.

[0053] For example, Figure 3 This is a schematic diagram illustrating the construction process of the target model provided in the embodiments of this application, such as... Figure 3 As shown, the process includes: Step S301, defining the input variables of the model, using the preprocessed data information as the input variables of the deep learning model; Step S302, training the deep learning model, constructing and training the deep learning model until the model converges; Step S303, outputting the model results. Through this model, alarm information can be classified based on the vectorization of words or sentences, continuously learning language features, mastering higher-level and more abstract language features, and classifying alarm information.

[0054] The target model applies deep learning models, utilizing an attention mechanism to capture the relationships between all other words and generate a new representation for each word. It analyzes and models the original alarm information, extracting the relationships between all words without considering word position, and performs language classification by learning from the word vectors of the alarm information. This enables pre-classification and analysis when unknown alarms occur, reducing operational burden.

[0055] This embodiment trains a target model, enabling the model to learn and predict the translation of alarm information from different network devices. Even if there is no directly matching rule in the alarm translation library, it can provide a preliminary and predictive translation, thereby reducing the workload of operation and maintenance personnel and improving processing efficiency.

[0056] Before training a natural language processing model based on training samples, the natural language processing model needs to be pre-trained. Optionally, in the alarm processing method for network devices provided in this application embodiment, before training the initial natural language processing model based on multiple sets of training samples, the method further includes: preprocessing the original alarm information to obtain preprocessed alarm data, wherein the preprocessing includes at least one of the following: deduplication, noise reduction, text cleaning, and word segmentation encoding; converting the preprocessed alarm data into a word vector sequence through a decoder, and calculating the attention weight of each word vector in the word vector sequence through an attention mechanism; weighting the corresponding word vectors in the word vector sequence based on the attention weight of each word vector to obtain an updated word vector sequence; and determining the initial natural language processing model based on the updated word vector sequence.

[0057] In some examples, deduplication removes duplicate alert messages to prevent the model from learning unnecessary redundant data, thus improving training efficiency and the model's generalization ability. Noise reduction removes or corrects irrelevant or erroneous information in alert messages, such as incorrect encodings or unusual symbols, ensuring the model can learn accurate alert features. Text cleaning removes special characters, stop words, and numbers from the alert text, making the text more consistent with natural language processing requirements and reducing interference. Word segmentation and encoding use Chinese word segmentation tools to divide the original alert information into word sequences, and then use word embedding to convert the words into word vectors, providing numerical input to the model.

[0058] The decoder takes the preprocessed alarm data (word vector sequence) as input and begins generating predicted translations or alarm classifications. The attention mechanism assigns different attention weights by calculating the correlation between the word vectors at each time step in the decoder and all word vectors in the encoder. This ensures that the target model focuses more on the key parts of the alarm information when generating the predicted output. Based on the attention weights calculated for each word vector by the attention mechanism, the word vectors in the word vector sequence are weighted and summed to form an updated word vector sequence. Increasing the weight of key information allows the model to focus on the most relevant words. Using the updated word vector sequence as input, the initial natural language processing model is determined through the model training process. During model training, parameters are continuously adjusted to minimize the error between the predicted output and the actual translated content or alarm classification.

[0059] For example, Figure 4 This is a schematic diagram of the preprocessing flow of alarm information provided according to the embodiments of this application, such as... Figure 4As shown, the process includes: Step S401, defining input variables and selecting original alarm information from the database as input; Step S402, defining attention mechanism variables and introducing a representation related to the task, called the query vector; Step S403, calculating the attention distribution using the given input variables and query vector; Step S404, calculating the weighted average of the input information, using the attention distribution as the weight of the input vector, and performing a weighted average calculation on the input information to complete the information extraction of data preprocessing.

[0060] This embodiment transforms the raw alarm information into a more structured and easily analyzable data format through preprocessing and optimization by a deep learning model, thereby helping the target model to understand the meaning and priority of the alarm information more quickly and accurately.

[0061] To ensure the accuracy of the translation content predicted by the target model, it is necessary to adjust the predicted translation content in a timely manner based on user feedback. Optionally, in the alarm handling method for network devices provided in this application embodiment, after displaying the predicted translation content on the front-end page, the method further includes: receiving user feedback on the predicted translation content, wherein the feedback result includes accurate prediction and inaccurate prediction; if the feedback result is accurate prediction, adding the alarm information and the predicted translation content to the alarm translation library; if the feedback result is inaccurate prediction, repeatedly executing the step of inputting the alarm information into the target model to obtain the predicted translation content until the number of executions is greater than or equal to the execution count threshold, or the feedback is accurate prediction; if the number of executions is greater than or equal to the execution count threshold, and the feedback result of the predicted translation content after each execution is inaccurate prediction, issuing a prompt message, wherein the prompt message is used to prompt maintenance personnel to handle the alarm information.

[0062] In some examples, the front-end page can be designed with mechanisms to facilitate user feedback on the accuracy of predicted translations, such as clicking "Accurate" or "Inaccurate" buttons, or providing a rating system. When a user confirms the accuracy of the translation and receives feedback indicating "Prediction Accurate," the alert information and its corresponding predicted translation are added to the alert translation library as new translation rules. This allows for the direct use of existing rules for translation when encountering similar alerts in the future, improving efficiency.

[0063] If the feedback result is "prediction inaccurate," the translation process will be re-executed, meaning the alarm information will be input into the target model again to obtain new predicted translation content. Each translation result and its feedback will be recorded until the number of translation attempts reaches a pre-set execution threshold, or the target model's predicted translation content receives a "prediction accurate" feedback. Before the execution threshold is reached, if the target model's continuously predicted translation content is consistently reported as "prediction inaccurate" by the user, a prompt message will be issued, reminding operations and maintenance personnel to manually handle the alarm information.

[0064] Based on their practical experience and knowledge, operations and maintenance personnel manually translate the alarm message and add the translation results to the alarm translation library as a translation rule for similar alarms in the future. Simultaneously, operations and maintenance personnel can adjust or add training data to help the target model learn this type of alarm translation, further improving the model's translation accuracy.

[0065] For example, Figure 5 This is a schematic diagram of the processing flow of feedback results provided in the embodiments of this application, such as... Figure 5 As shown, the process includes: Step S501, analyzing the output of the target model and comparing it with data in the database to ensure that the accuracy of the translation predicted by the target model reaches 80%; Step S502, sending the output results to the operation and maintenance front-end, adding pre-classification results after the unknown alarms displayed in the centralized monitoring system through a visualization script, serving as an auxiliary suggestion for front-line operation and maintenance personnel to confirm the alarm content and improve operation and maintenance efficiency; Step S503, iterative optimization based on feedback results, adding a "Feedback" button to the visualization page. After the operation and maintenance personnel have handled the alarm issue, clicking the button confirms whether the pre-classification results are correct, and the page will provide real-time feedback to the target model. Through continuous iterative optimization, the accuracy of the target model is improved. This achieves rapid response and automatic learning when unknown alarms occur in the centralized monitoring system, reducing the pressure on manual operation and maintenance.

[0066] This embodiment dynamically optimizes the target model based on user feedback and enriches the alarm translation library to achieve effective management and translation of unknown or difficult-to-translate alarm information, thereby improving the efficiency and accuracy of network operation and maintenance.

[0067] To ensure the accuracy of the translations predicted by the target model, the target model needs to be updated in a timely manner. Optionally, in the alarm processing method for network devices provided in this application embodiment, the method further includes: collecting feedback results of all predicted translations output by the target model at preset intervals; determining the total number of feedback results of all predicted translations, and determining the target number of accurate predicted translations based on the feedback results; calculating the ratio of the target number to the total number, and determining whether the ratio is greater than or equal to a preset accuracy threshold; if the ratio is less than the accuracy threshold, determining the feedback result as an alarm message of inaccurate prediction, and obtaining the actual translation content of the alarm message; and updating the target model based on the alarm message of inaccurate prediction and the actual translation content.

[0068] In some examples, user feedback on all predicted translations output by the model is automatically collected by setting a preset period, such as hourly, daily, or weekly. The total number of feedback results for all predicted translations within this preset period is counted, and the target number of those feedbacks marked "accurate prediction" is determined. The ratio of the target number to the total number is calculated, representing the prediction accuracy of the target model within this preset period. The calculated accuracy is compared to a preset accuracy threshold. If the accuracy is below the threshold, it indicates that the performance of the target model may have degraded, requiring further optimization and adjustment. In cases where the accuracy is below the threshold, the feedback results are further analyzed to identify those alarm messages marked as "inaccurate prediction," as well as the actual translation content provided by maintenance personnel. This data is used for iterative training of the model to improve translation accuracy.

[0069] The collected inaccurate prediction alerts and their actual translations are used as new training data to retrain or fine-tune the target model. After updating the model, it is evaluated using an independent validation set to ensure performance improvement. If the updated target model still does not perform well on the validation set, further model optimization is needed, such as adjusting the architecture, adding data, and improving preprocessing steps.

[0070] This embodiment regularly updates the target model to ensure its translation accuracy remains at a high level. Simultaneously, it continuously improves model performance based on user feedback, ultimately enhancing operational efficiency and effectiveness. It adapts to dynamic changes in the network environment, maintaining the timeliness and accuracy of translations.

[0071] Optionally, in the alarm processing method for network devices provided in this application embodiment, updating the target model based on inaccurately predicted alarm information and actual translated content includes: determining the inaccurately predicted alarm information and actual translated content as new training samples; adding the new training samples to multiple sets of training samples to obtain updated multiple sets of training samples; and training the initial natural language processing model based on the updated multiple sets of training samples to obtain the updated target model.

[0072] In some examples, all alert messages marked as "inaccurate prediction" and their actual translations are filtered from user feedback. These are used as new training samples to help the target model learn more accurate translation rules. These newly identified training samples are added to existing sets of training samples and merged with the original dataset to form updated sets of training samples. Based on these updated sets of training samples, the initial natural language processing model is retrained. During training, the target model learns how to more accurately translate alert messages in the new samples and how to better apply attention mechanisms to focus on key translation information. Hyperparameters of the target model, such as learning rate and batch size, can be adjusted during training to optimize model performance. After training, the updated target model is validated on an independent test set to ensure improved translation accuracy while maintaining high generalization ability.

[0073] This embodiment improves the accuracy and reliability of translation by continuously iterating and optimizing the target model with new training data. It ensures that the model continuously improves with the emergence of new data, adapting to the ever-changing needs of alarm information translation.

[0074] To improve the efficiency of alarm information processing, an alarm translation library can be constructed. Optionally, in the alarm processing method for network devices provided in this application embodiment, before determining whether the alarm information matches the translation rules in the alarm translation library, the method further includes: obtaining device information for each network device, extracting alarm information and translated content of the alarm information from the device information; if the device information does not provide alarm information or translated content of the alarm information, obtaining historical alarm processing records of the network device, extracting alarm information and translated content of the alarm information from the historical alarm processing records; and constructing an alarm translation library based on the alarm information and translated content of the alarm information for each network device.

[0075] In some examples, for each network device, such as a router, switch, server, firewall, etc., the first step is to try to obtain the alarm information and its recommended translation directly from the official documentation, manuals, or configuration files of the network device. If the device information explicitly provides the alarm information and its corresponding translation, this information is extracted directly. For example, from the router's documentation, the standard translation for the alarm information "Interface GigabitEthernet0 / 0 input queue dropped" can be found as: "Gigabit Ethernet interface 0 / 0 input queue lost packets".

[0076] When alarm information or translations are not provided in the device information, a translation library can be built using the network device's historical alarm handling records. Historical alarm handling records can contain the actual handling procedures and experience of operations and maintenance personnel when dealing with various alarms, including specific descriptions of the alarms and human-translated explanations. Alarm information and the operations and maintenance personnel's translations can be extracted from the historical records. This content does not need to be standardized, but it provides examples of how alarms are understood and handled. For example, from the switch's history, we can see that the operations and maintenance personnel interpreted the "storm control exceeded" alarm as: "The storm control threshold has been exceeded."

[0077] All alarm messages and their translations obtained through the two methods described above are organized and categorized to construct an alarm translation library. Each record in the library contains the original alarm message, a recommended translation, and possible translation variations. For the same alarm message, if multiple translations exist, all variations can be retained in the library to support broader understanding and translation adaptability. Furthermore, the library can include alarm message categories (such as hardware failure, software failure, configuration error, etc.) and associated device types, facilitating quick translation searches based on device and alarm type.

[0078] Building an alarm translation library is an ongoing process that involves regularly updating its content. This includes adding new alarm information, updating translations, and verifying the accuracy of the translations. This can be achieved by establishing a feedback mechanism to collect feedback from frontline operations personnel, or by analyzing the translation accuracy of the model.

[0079] This embodiment constructs an alarm translation library containing alarm information from various network devices and their translations, providing timely preliminary classification and translation suggestions for alarm information, reducing the burden on maintenance personnel and improving maintenance response speed.

[0080] According to another embodiment of this application, an alarm processing system for a network device is also provided. Figure 6This is a schematic diagram of an alarm processing system for a network device provided according to an embodiment of this application, such as... Figure 6 As shown, the system includes: a data preprocessing device 601, a deep learning modeling device 602, and a model result visualization device 603.

[0081] The data preprocessing unit 601 is responsible for preprocessing alarm information, including collecting raw alarm information, defining input variables, and calculating the attention distribution and weighted average of the input information; the deep learning modeling unit 602 is responsible for building a deep learning model, training the model, and outputting the results; and the model result visualization unit 603 is responsible for analyzing whether the output results have practical utility, sending the analysis results to the centralized monitoring system, and continuously iterating and optimizing the model based on feedback results. This optimizes the entire process of managing unknown alarms and improves the work efficiency of frontline maintenance personnel.

[0082] This embodiment improves the efficiency and management capabilities of unknown alarm governance through a network device alarm processing system. By preprocessing the model, alarm categories can be quickly analyzed, relevant suggestions can be proposed, and the response efficiency of frontline operations and maintenance personnel can be improved. It effectively enhances the overall analytical capabilities of alarm governance; through deep learning model analysis and modeling, the entire process of unknown alarm governance can be continuously iterated and optimized, improving intelligent operations and maintenance capabilities.

[0083] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0084] Example 2

[0085] This application also provides an alarm processing apparatus for a network device. It should be noted that the alarm processing apparatus for a network device in this application can be used to execute the alarm processing method for a network device provided in this application. The alarm processing apparatus for a network device provided in this application is described below.

[0086] According to an embodiment of this application, an apparatus for implementing the alarm processing method of the above-described network device is also provided. Figure 7 This is a schematic diagram of an alarm processing device for a network device according to an embodiment of this application, such as... Figure 7 As shown, the device includes:

[0087] The receiving unit 701 is used to receive alarm information from network devices and determine whether the alarm information matches the translation rules in the alarm translation library. The alarm translation library contains multiple translation rules, and each translation rule contains an alarm message and its translation content.

[0088] Extraction unit 702 is used to extract the target translation content from the target translation rule when the alarm information matches the target translation rule, and to display the target translation content on the front-end page;

[0089] The input unit 703 is used to input the alarm information into the target model to obtain the predicted translation content when there is no translation rule matching the alarm information in the alarm translation library. The target model is trained by multiple sets of training samples, and each set of training samples includes historical alarm information and historical translation content.

[0090] Display unit 704 is used to display the predicted translation content on the front-end page.

[0091] The alarm processing device for network devices provided in this application embodiment receives alarm information from the network device through a receiving unit 701 and determines whether the alarm information matches a translation rule in an alarm translation library. The alarm translation library contains multiple translation rules, each containing a type of alarm information and its translation content. An extraction unit 702 extracts the target translation content from the target translation rule when the alarm information matches, and displays the target translation content on a front-end page. An input unit 703 inputs the alarm information into a target model when no matching translation rule exists in the alarm translation library. The system predicts the translation content. The target model is trained from multiple sets of training samples, each set including historical alarm information and historical translation content. The display unit 704 displays the predicted translation content on the front-end page. The system matches the translation content of the alarm information with the alarm translation library. If no translation content can be matched in the alarm translation library, the system outputs the predicted translation content of the alarm information based on the target model. This achieves the goal of timely parsing of alarm information and providing translation content, thereby improving the technical effect of network device operation and maintenance efficiency. It also solves the technical problem of low network device operation and maintenance efficiency due to the inability to match translation content for some alarm information.

[0092] Optionally, in the alarm processing apparatus for network devices provided in this application embodiment, the apparatus further includes: a first acquisition unit, configured to acquire original alarm information of various network devices, and extract multiple historical alarm messages and historical translation content of each historical alarm message from the original alarm information, wherein the network devices include at least one of the following: router, switch, server, and firewall; a first determination unit, configured to determine a set of training samples based on each historical alarm message and the historical translation content of the historical alarm message, thereby obtaining multiple sets of training samples; and a training unit, configured to train an initial natural language processing model based on the multiple sets of training samples, thereby obtaining a target model.

[0093] Optionally, in the alarm processing apparatus for network devices provided in this application embodiment, the apparatus further includes: a preprocessing unit, used to preprocess the original alarm information to obtain preprocessed alarm data, wherein the preprocessing includes at least one of the following: deduplication, noise reduction, text cleaning, and word segmentation encoding; a conversion unit, used to convert the preprocessed alarm data into a word vector sequence through a decoder, and to calculate the attention weight of each word vector in the word vector sequence through an attention mechanism; a weighting unit, used to perform weighting processing on the corresponding word vectors in the word vector sequence based on the attention weight of each word vector to obtain an updated word vector sequence; and a second determination unit, used to determine an initial natural language processing model based on the updated word vector sequence.

[0094] Optionally, in the alarm processing apparatus for network devices provided in this application embodiment, the apparatus further includes: a feedback unit, configured to receive feedback results from users on the predicted translation content, wherein the feedback results include prediction accuracy and prediction inaccuracy; an adding unit, configured to add the alarm information and the predicted translation content to the alarm translation library when the feedback result is prediction accuracy; an execution unit, configured to repeatedly execute the step of inputting the alarm information into the target model to obtain the predicted translation content when the feedback result is prediction inaccuracy, until the number of executions is greater than or equal to the number of executions threshold, or the feedback result is prediction accuracy; and a prompting unit, configured to issue a prompting message when the number of executions is greater than or equal to the number of executions threshold, and the feedback result of the predicted translation content after each execution is prediction inaccuracy, wherein the prompting message is used to prompt maintenance personnel to handle the alarm information.

[0095] Optionally, in the alarm processing device for network devices provided in this application embodiment, the device further includes: a collection unit, used to collect feedback results of all predicted translation content output by the target model at preset intervals; a third determination unit, used to determine the total number of feedback results of all predicted translation content, and determine the target number of feedback results that are accurately predicted translation content; a calculation unit, used to calculate the ratio of the target number to the total number, and determine whether the ratio is greater than or equal to a preset accuracy threshold; a fourth determination unit, used to determine that the feedback result is an alarm message of inaccurate prediction when the ratio is less than the accuracy threshold, and obtain the actual translation content of the alarm message; and an update unit, used to update the target model based on the alarm message of inaccurate prediction and the actual translation content.

[0096] Optionally, in the alarm processing apparatus for network devices provided in this application embodiment, the updating unit includes: a first determining module, used to determine the alarm information with inaccurate prediction and the actual translated content as new training samples; an adding module, used to add the new training samples to multiple sets of training samples to obtain updated multiple sets of training samples; and a training module, used to train an initial natural language processing model based on the updated multiple sets of training samples to obtain an updated target model.

[0097] Optionally, in the alarm processing apparatus for network devices provided in this application embodiment, the apparatus further includes: a second acquisition unit, configured to acquire device information for each type of network device, and extract alarm information and translated content of the alarm information from the device information; a third acquisition unit, configured to acquire historical alarm processing records of the network device when the device information does not provide alarm information or translated content of the alarm information, and extract alarm information and translated content of the alarm information from the historical alarm processing records; and a construction unit, configured to construct an alarm translation library based on the alarm information and translated content of the alarm information for each type of network device.

[0098] It should be noted that the receiving unit 701, extraction unit 702, input unit 703, and display unit 704 mentioned above correspond to steps S201 to S204 in Embodiment 1. The two modules and the corresponding steps implement the same instances and application scenarios, but are not limited to the content disclosed in Embodiment 1. It should be noted that the above modules or units can be hardware or software components stored in memory (e.g., memory 104) and processed by one or more processors (e.g., processors 102a, 102b, ..., 102n). The above modules can also be part of the device and run in the computer terminal 10 provided in Embodiment 1.

[0099] Example 3

[0100] Embodiments of this application may provide an electronic device. Figure 8 This is a structural block diagram of an electronic device according to an embodiment of this application. Figure 8 As shown, the electronic device may include: one or more ( Figure 8 (Only one is shown) processor 802, memory 804, memory controller, and peripheral interface, wherein the peripheral interface is connected to the radio frequency module, audio module and display.

[0101] The memory can be used to store software programs and modules, such as the program instructions / modules corresponding to the methods and apparatus in the embodiments of this application. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory, thereby implementing the above-described methods. The memory may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory may further include memory remotely located relative to the processor, and these remote memories can be connected to the terminal via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0102] The processor can access information and applications stored in memory via a transmission device to perform the following steps: receiving alarm information from network devices and determining whether the alarm information matches a translation rule in an alarm translation library, wherein the alarm translation library contains multiple translation rules, each containing a type of alarm information and its translation content; if the alarm information matches a target translation rule, extracting the target translation content from the target translation rule and displaying the target translation content on the front-end page; if no matching translation rule exists in the alarm translation library, inputting the alarm information into a target model to obtain predicted translation content, wherein the target model is trained using multiple sets of training samples, each set of training samples including historical alarm information and historical translation content; and displaying the predicted translation content on the front-end page.

[0103] The processor can also access information and applications stored in the memory via a transmission device to perform the following steps: acquiring raw alarm information from various network devices; extracting multiple historical alarm messages and their historical translations from the raw alarm information; wherein the network devices include at least one of the following: routers, switches, servers, and firewalls; determining a set of training samples based on each historical alarm message and its historical translations to obtain multiple sets of training samples; and training an initial natural language processing model based on the multiple sets of training samples to obtain the target model.

[0104] The processor can also invoke information and applications stored in the memory via a transmission device to perform the following steps: preprocessing the original alarm information to obtain preprocessed alarm data, wherein the preprocessing includes at least one of the following: deduplication, noise reduction, text cleaning, and word segmentation encoding; converting the preprocessed alarm data into a word vector sequence through a decoder, and calculating the attention weight of each word vector in the word vector sequence through an attention mechanism; weighting the corresponding word vectors in the word vector sequence based on the attention weight of each word vector to obtain an updated word vector sequence; and determining the initial natural language processing model based on the updated word vector sequence.

[0105] The processor can also access information and applications stored in the memory via a transmission device to perform the following steps: receiving user feedback on the predicted translation content, where the feedback includes whether the prediction is accurate or inaccurate; if the feedback is that the prediction is accurate, adding the alarm information and the predicted translation content to the alarm translation library; if the feedback is that the prediction is inaccurate, repeatedly executing the step of inputting the alarm information into the target model to obtain the predicted translation content until the number of executions is greater than or equal to the execution count threshold, or the feedback is that the prediction is accurate; if the number of executions is greater than or equal to the execution count threshold, and the feedback result of the predicted translation content after each execution is that the prediction is inaccurate, issuing a prompt message, whereby the prompt message is used to prompt maintenance personnel to handle the alarm information.

[0106] The processor can also access information and applications stored in the memory via a transmission device to perform the following steps: collecting feedback results of all predicted translations output by the target model at preset intervals; determining the total number of feedback results of all predicted translations and identifying the target number of accurately predicted translations; calculating the ratio of the target number to the total number and determining whether the ratio is greater than or equal to a preset accuracy threshold; if the ratio is less than the accuracy threshold, identifying the feedback result as an inaccurate prediction alarm and obtaining the actual translation content of the alarm; and updating the target model based on the inaccurate prediction alarm and the actual translation content.

[0107] The processor can also call the information and application stored in the memory through the transmission device to perform the following steps: identify the alarm information with inaccurate prediction and the actual translated content as new training samples; add the new training samples to multiple sets of training samples to obtain updated multiple sets of training samples; train the initial natural language processing model based on the updated multiple sets of training samples to obtain the updated target model.

[0108] The processor can also access information and applications stored in the memory via the transmission device to perform the following steps: obtain device information for each network device, extract alarm information and translated content of the alarm information from the device information; if the device information does not provide alarm information or translated content of the alarm information, obtain historical alarm processing records of the network device, extract alarm information and translated content of the alarm information from the historical alarm processing records; and construct an alarm translation library based on the alarm information and translated content of the alarm information for each network device.

[0109] This application provides a method for receiving alarm information from network devices and determining whether the alarm information matches translation rules in an alarm translation library. The alarm translation library contains multiple translation rules, each containing a type of alarm information and its translation content. If the alarm information matches a target translation rule, the target translation content is extracted from the target translation rule and displayed on a front-end page. If no matching translation rule exists in the alarm translation library, the alarm information is input into a target model to obtain a predicted translation content. The target model is trained using multiple sets of training samples, each set including historical alarm information and historical translation content. The predicted translation content is then displayed on the front-end page. By matching the alarm information's translation content with the alarm translation library, and outputting a predicted translation content based on the target model when no matching translation content is found in the library, this method achieves the goal of timely parsing of alarm information and providing translation content. This improves the operational efficiency of network devices and solves the technical problem of low operational efficiency due to the inability to match translation content for some alarm information.

[0110] Those skilled in the art will understand that Figure 8 The structure shown is for illustrative purposes only. Electronic devices can also be smartphones, tablets, handheld computers, mobile internet devices (MIDs), PADs, and other terminal devices. Figure 8 This does not limit the structure of the aforementioned electronic device. For example, electronic devices may also include components that are more... Figure 8 The more or fewer components shown (such as network interfaces, display devices, etc.), or having the same Figure 8 The different configurations shown.

[0111] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing the hardware related to the terminal device. The program can be stored in a computer-readable storage medium, which may include: flash drive, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.

[0112] Example 4

[0113] Embodiments of this application also provide a storage medium. Optionally, in this embodiment, the storage medium can be used to store the program code executed by the alarm processing method of the network device provided in Embodiment 1.

[0114] Optionally, in this embodiment, the storage medium may be located in any computer terminal in a group of computer terminals in a computer network, or in any mobile terminal in a group of mobile terminals.

[0115] This application also provides a computer program product, which, when executed on a data processing device, is suitable for performing alarm processing method steps of a network device.

[0116] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0117] In the above embodiments of this application, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0118] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection may be through some interfaces; the indirect coupling or communication connection between units or modules may be electrical or other forms.

[0119] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0120] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0121] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard drive, magnetic disk, or optical disk.

[0122] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A method for handling alarms in a network device, characterized in that, include: Receive alarm information from network devices and determine whether the alarm information matches the translation rules in the alarm translation library, wherein the alarm translation library contains multiple translation rules, and each translation rule contains a translation of an alarm information and the alarm information. If the alarm information matches the target translation rule, the target translation content is extracted from the target translation rule and displayed on the front-end page; If no translation rule matching the alarm information exists in the alarm translation library, the alarm information is input into the target model to obtain the predicted translation content. The target model is trained by multiple sets of training samples, and each set of training samples includes historical alarm information and historical translation content. The predicted translation content is displayed on the front-end page.

2. The method according to claim 1, characterized in that, The target model is trained in the following manner: Obtain raw alarm information from various network devices, extract multiple historical alarm messages and historical translations of each historical alarm message from the raw alarm information, wherein the network devices include at least one of the following: router, switch, server, and firewall; Based on each historical alarm message and the historical translation content of the historical alarm message, a set of training samples is determined, resulting in multiple sets of training samples; The initial natural language processing model is trained based on the multiple sets of training samples to obtain the target model.

3. The method according to claim 2, characterized in that, Before training an initial natural language processing model based on the multiple sets of training samples, the method further includes: The original alarm information is preprocessed to obtain preprocessed alarm data, wherein the preprocessing includes at least one of the following: deduplication, noise reduction, text cleaning, and word segmentation encoding; The preprocessed alarm data is converted into a word vector sequence by a decoder, and the attention weight of each word vector in the word vector sequence is calculated by an attention mechanism. The word vectors in the word vector sequence are weighted based on the attention weight of each word vector to obtain the updated word vector sequence. The initial natural language processing model is determined based on the updated word vector sequence.

4. The method according to claim 1, characterized in that, After displaying the predicted translation on the front-end page, the method further includes: Receive user feedback on the predicted translation content, wherein the feedback includes whether the prediction is accurate or inaccurate; If the feedback result indicates that the prediction is accurate, the alarm information and the predicted translation content are added to the alarm translation library; If the feedback result indicates that the prediction is inaccurate, the step of inputting the alarm information into the target model to obtain the predicted translation content is repeated until the number of executions is greater than or equal to the execution number threshold, or the feedback indicates that the prediction is accurate. If the number of executions is greater than or equal to the execution threshold, and the feedback result of the predicted translation content after each execution is that the prediction is inaccurate, a prompt message is issued, wherein the prompt message is used to prompt the operation and maintenance personnel to handle the alarm message.

5. The method according to claim 4, characterized in that, The method further includes: Feedback results of all predicted translations output by the target model are collected at preset intervals; Determine the total number of feedback results for all predicted translations, and determine the target number of feedback results that accurately predict the translations. Calculate the ratio of the target quantity to the total quantity, and determine whether the ratio is greater than or equal to a preset accuracy threshold; If the ratio is less than the accuracy threshold, the feedback result is determined to be an alarm message indicating inaccurate prediction, and the actual translated content of the alarm message is obtained. The target model is updated based on the warning information regarding inaccurate predictions and the actual translated content.

6. The method according to claim 5, characterized in that, Updating the target model based on the inaccurate prediction alert information and the actual translated content includes: The alarm information indicating inaccurate predictions and the actual translated content are identified as new training samples. The newly added training sample is added to the multiple sets of training samples to obtain updated multiple sets of training samples; The initial natural language processing model is trained based on the updated multiple sets of training samples to obtain the updated target model.

7. The method according to claim 1, characterized in that, Before determining whether the alarm information matches the translation rules in the alarm translation library, the method further includes: Obtain device information for each network device, and extract alarm information and translated content of the alarm information from the device information; If the device information does not provide alarm information or translated alarm information, obtain the historical alarm processing records of the network device, and extract the alarm information and translated alarm information of the network device from the historical alarm processing records; The alarm translation library is constructed based on the alarm information and its translation for each network device.

8. An alarm processing device for a network device, characterized in that, include: A receiving unit is used to receive alarm information from network devices and determine whether the alarm information matches the translation rules in the alarm translation library. The alarm translation library contains multiple translation rules, and each translation rule contains a translation of an alarm information and the alarm information. The extraction unit is used to extract the target translation content from the target translation rule when the alarm information matches the target translation rule, and to display the target translation content on the front-end page; The input unit is used to input the alarm information into the target model to obtain the predicted translation content when there is no translation rule matching the alarm information in the alarm translation library. The target model is trained by multiple sets of training samples, and each set of training samples includes historical alarm information and historical translation content. The display unit is used to display the predicted translation content on the front-end page.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored executable program, wherein, when the executable program is executed, it controls the device where the computer-readable storage medium is located to perform the alarm processing method of the network device according to any one of claims 1 to 7.

10. An electronic device, characterized in that, include: Memory, which stores executable programs; A processor for running the program, wherein the program executes the alarm processing method of the network device according to any one of claims 1 to 7.

11. A computer program product comprising computer instructions, characterized in that, When the computer instructions are executed by the processor, they implement the steps of the alarm processing method for the network device according to any one of claims 1 to 7.

Images