Alarm noise reduction method and device, electronic equipment and storage medium

By performing timing and non-time sequence segmentation, clustering and merging of alarm information, the problem of unutilized timing data characteristics in the existing technology is solved, the accuracy and efficiency of alarm management are improved, the alarm rule system is optimized, and the operation and operation of the alarm is facilitated. The Wei team deals with high real-time alarm scenarios.

CN118940113APending Publication Date: 2024-11-12CHINA MOBILE (SUZHOU) SOFTWARE TECH CO LTD +1
View PDF 0 Cites 1 Cited by

Patent Information

Application Number
CN202410989356.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-23
Publication Date
2024-11-12

AI Technical Summary

Technical Problem

The existing alarm noise reduction strategy fails to fully utilize the timing data characteristics, resulting in poor alarm suppression effect. Especially in scenarios with high real-time requirements, it is difficult to accurately capture and deal with small deviations introduced by time factors, resulting in alarm management in operation and maintenance work. Insufficient efficiency and accuracy.

Method used

By dividing the alarm information into timing alarm information and non-timing alarm information, clustering is performed using clustering network model and preset clustering algorithm, and merging is performed, deduplication, priority evaluation and fault recovery, precisely capture timing Data characteristics improve the accuracy and efficiency of alarm suppression.

Benefits of technology

It significantly improves the accuracy and efficiency of alarm management, optimizes the alarm rule system, and can better deal with data scenarios with strict real-time requirements, helps the operation and maintenance team to deal with alarm storms more effectively, and improves the overall operation and maintenance level.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118940113A_ABST
    Figure CN118940113A_ABST
Patent Text Reader

Abstract

The invention provides an alarm noise reduction method and device, electronic equipment and a storage medium, and the method comprises the steps: obtaining the state information of a system component, analyzing the state information based on a preset alarm rule, and generating alarm information which comprises time sequence alarm information and non-time sequence alarm information; the time sequence alarm information is input into a clustering network model for clustering processing, and a time sequence alarm information classification result is obtained; performing clustering processing on the non-time-sequence alarm information by using a preset clustering algorithm to obtain a non-time-sequence alarm information classification result; and merging the time sequence alarm information classification result and the non-time sequence alarm information classification result to obtain an alarm information classification result. Compared with the prior art, according to the embodiment of the invention, the alarm information is divided into the time sequence alarm information and the non-time sequence alarm information, and the clustering processing is carried out respectively, so that the accuracy of alarm suppression can be improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present disclosure relates to the field of big data technology, and in particular to a method and device for alarm noise reduction, an electronic device, and a storage medium. Background Art

[0002] In daily operation and maintenance work, the alarm storm phenomenon is a major challenge faced by the operation and maintenance team. Although the operation and maintenance personnel have tried to curb the massive flow of alarm information by formulating various alarm rules, they still face two problems in actual operation: first, facing the vast amount of alarm data, it is difficult to accurately define and use a fixed model to effectively perform alarm noise reduction; second, with the continuous expansion of the alarm rule system, its management and maintenance complexity has increased exponentially, increasing the burden of operation and maintenance.

[0003] However, the current alarm noise reduction strategy has significant limitations, resulting in unsatisfactory suppression effects. Specifically, existing methods fail to fully consider and properly handle the large amount of time series data characteristics contained in alarm information. This omission directly leads to the absence of noise reduction algorithms for time series alarm data, which greatly reduces the effectiveness of alarm suppression. Especially in data scenarios with strict real-time requirements, due to the slight deviations introduced by time factors, the noise reduction processing effect of alarm information is even weaker, further weakening the actual effectiveness of alarm suppression. The current alarm suppression effect is not obvious, and it is urgent to develop a noise reduction algorithm that can accurately capture and effectively utilize the characteristics of time series data to improve the accuracy and efficiency of alarm management. Summary of the invention

[0004] The present disclosure provides a method and device for alarm noise reduction, an electronic device and a storage medium. The main purpose is to solve the problem that the effect of alarm suppression in operation and maintenance work is not obvious and accurate due to the inability to accurately reduce alarm noise. The main purpose is to improve the accuracy and efficiency of alarm management.

[0005] According to a first aspect of the present disclosure, a method for alarm noise reduction is provided, comprising:

[0006] Acquire status information of system components, and analyze the status information based on preset alarm rules to generate alarm information, wherein the alarm information includes sequential alarm information and non-sequential alarm information;

[0007] Inputting the time series alarm information into a clustering network model for clustering processing to obtain a classification result of the time series alarm information;

[0008] Using a preset clustering algorithm to perform clustering processing on the non-time series alarm information to obtain a classification result of the non-time series alarm information;

[0009] The time series alarm information classification result and the non-time series alarm information classification result are combined to obtain an alarm information classification result.

[0010] In some embodiments, after merging the time series alarm information classification result and the non-time series alarm information classification result, the method further includes:

[0011] Based on the alarm information classification result, deduplication processing is performed on the alarm information in different alarm classifications;

[0012] Prioritize the warning information after deduplication and mark the corresponding priority;

[0013] The alarm recovery rule corresponding to the alarm information is called, and fault recovery processing is performed on the alarm information in the alarm classification in sequence according to the priority order.

[0014] In some embodiments, the inputting the time series alarm information into a clustering network model for clustering processing to obtain a classification result of the time series alarm information includes:

[0015] Performing sampling and dimensionality reduction processing on the time series alarm information to obtain low-dimensional time series alarm information;

[0016] Inputting the low-dimensional time series alarm information into the clustering network model, performing clustering processing on the low-dimensional time series alarm information, and obtaining a model classification result;

[0017] According to the model classification result, the unclassified time series alarm information is classified into the cluster category in the model classification result to obtain the time series alarm information classification result.

[0018] In some embodiments, the sampling and dimensionality reduction processing is performed on the time series alarm information to obtain low-dimensional time series alarm information, including:

[0019] According to a preset number of sampling times and a preset sampling ratio, the timing alarm information is randomly sampled to obtain a sampling set;

[0020] Divide each time series alarm information in the sampling set into a first number of data frames, and calculate the arithmetic mean of each of the data frames;

[0021] The arithmetic mean of each of the time series alarm information is arranged in chronological order to generate the low-dimensional time series alarm information.

[0022] In some embodiments, the inputting the low-dimensional time series alarm information into the clustering network model, performing clustering processing on the low-dimensional time series alarm information, and obtaining a model classification result includes:

[0023] Calculating a cluster center coefficient between the low-dimensional time series alarm information and a first initial cluster center, and determining the first initial cluster center based on the cluster center coefficient;

[0024] According to the first initial cluster center, the clustering network model is called to perform clustering processing on the low-dimensional time series alarm information to obtain the model classification result.

[0025] In some embodiments, before calling the clustering network model to perform clustering processing on the low-dimensional time series alarm information according to the first initial cluster center, the method further includes:

[0026] Based on a preset similarity algorithm, a first similarity between the training warning information and the training cluster core is calculated, and based on the first similarity, the training warning information is classified into a training cluster corresponding to the training cluster core;

[0027] The training cluster centers are iteratively trained until the change value of the training cluster centers between two adjacent iterations is less than a first threshold, thereby obtaining a trained clustering network model.

[0028] In some embodiments, the step of classifying the unclassified time series alarm information into the cluster category in the model classification result according to the model classification result to obtain the time series alarm information classification result includes:

[0029] Calculating a second similarity between the unclassified time series alarm information and the cluster center of each first cluster in the model classification result;

[0030] Based on the clustering radius of each of the first clusters and the second similarity, the unclassified time series alarm information is classified to obtain the time series alarm information classification result.

[0031] In some embodiments, clustering the non-time-series alarm information using a preset clustering algorithm to obtain a classification result of the non-time-series alarm information includes:

[0032] Calculating the attribute distance between different non-time series alarm information, wherein the attribute distance is the characteristic distance of the same attribute information in different non-time series alarm information;

[0033] Based on the attribute distance, clustering processing is performed on the non-time series alarm information to obtain a classification result of the non-time series alarm information.

[0034] In some embodiments, clustering the non-time series alarm information based on the attribute distance to obtain the classification result of the non-time series alarm information includes:

[0035] Randomly selecting a second number of second initial cluster centers, and calculating a third similarity between the non-time series alarm information and the second initial cluster centers;

[0036] Based on the third similarity and the attribute distance, classifying the non-time series alarm information into a second cluster corresponding to the second initial cluster center;

[0037] The second initial cluster center is iteratively updated until the change value of the second initial cluster center in two adjacent iterations is less than a second threshold, and the non-time series alarm information is reclassified into a new second cluster to obtain the classification result of the non-time series alarm information.

[0038] In some embodiments, the merging of the time series alarm information classification result and the non-time series alarm information classification result to obtain the alarm information classification result includes:

[0039] Calculating the similarity between the cluster center of each first cluster in the time series alarm information classification result and the cluster center of each second cluster in the non-time series alarm information classification result to obtain a fourth similarity;

[0040] Merging the first cluster and the second cluster whose fourth similarity is greater than a third threshold to obtain a third cluster;

[0041] The unmerged first cluster, the unmerged second cluster, and the third cluster are integrated to generate the alarm information classification result.

[0042] According to a second aspect of the present disclosure, a device for alarm noise reduction is provided, comprising:

[0043] An acquisition unit, used to acquire status information of system components, and analyze the status information based on preset alarm rules to generate alarm information, wherein the alarm information includes sequential alarm information and non-sequential alarm information;

[0044] A first clustering unit, used for inputting the time series alarm information into a clustering network model for clustering processing to obtain a classification result of the time series alarm information;

[0045] A second clustering unit is used to perform clustering processing on the non-time series alarm information by using a preset clustering algorithm to obtain a classification result of the non-time series alarm information;

[0046] The merging unit is used to merge the time series alarm information classification result and the non-time series alarm information classification result to obtain the alarm information classification result.

[0047] In some embodiments, the apparatus further comprises:

[0048] a deduplication unit, configured to perform deduplication processing on alarm information in different alarm classifications based on the alarm information classification result after the merging unit merges the time series alarm information classification result and the non-time series alarm information classification result;

[0049] An evaluation unit, configured to perform priority evaluation on the deduplicated alarm information classification and mark the corresponding priority;

[0050] The recovery unit is used to call the alarm recovery rule corresponding to the alarm information, and perform fault recovery processing on the alarm information in the alarm classification in order of priority.

[0051] In some embodiments, the first clustering unit comprises:

[0052] A dimensionality reduction module, used for performing sampling and dimensionality reduction processing on the time series alarm information to obtain low-dimensional time series alarm information;

[0053] A first clustering module, used for inputting the low-dimensional time series alarm information into the clustering network model, performing clustering processing on the low-dimensional time series alarm information, and obtaining a model classification result;

[0054] A classification module is used to classify unclassified time series alarm information into cluster categories in the model classification result according to the model classification result to obtain the time series alarm information classification result.

[0055] In some embodiments, the dimensionality reduction module is further used to:

[0056] According to a preset number of sampling times and a preset sampling ratio, the timing alarm information is randomly sampled to obtain a sampling set;

[0057] Divide each time series alarm information in the sampling set into a first number of data frames, and calculate the arithmetic mean of each of the data frames;

[0058] The arithmetic mean of each of the time series alarm information is arranged in chronological order to generate the low-dimensional time series alarm information.

[0059] In some embodiments, the first clustering module is further configured to:

[0060] Calculating a cluster center coefficient between the low-dimensional time series alarm information and a first initial cluster center, and determining the first initial cluster center based on the cluster center coefficient;

[0061] According to the first initial cluster center, the clustering network model is called to perform clustering processing on the low-dimensional time series alarm information to obtain the model classification result.

[0062] In some embodiments, the apparatus further comprises:

[0063] A first training unit is configured to calculate a first similarity between the training alarm information and the training cluster center based on a preset similarity algorithm before the first clustering unit calls the clustering network model to perform clustering processing on the low-dimensional time series alarm information according to the first initial cluster center, and classify the training alarm information into a training cluster cluster corresponding to the training cluster center based on the first similarity;

[0064] The second training unit is used to iteratively train the training cluster centers until the change value of two consecutive iterations of the training cluster centers is less than a first threshold, thereby obtaining a trained clustering network model.

[0065] In some embodiments, the classification module is further configured to:

[0066] Calculating a second similarity between the unclassified time series alarm information and the cluster center of each first cluster in the model classification result;

[0067] Based on the clustering radius of each of the first clusters and the second similarity, the unclassified time series alarm information is classified to obtain the time series alarm information classification result.

[0068] In some embodiments, the second clustering unit comprises:

[0069] A first calculation module, configured to calculate an attribute distance between different non-time series alarm information, wherein the attribute distance is a characteristic distance of the same attribute information in different non-time series alarm information;

[0070] The second clustering module is used to perform clustering processing on the non-time series alarm information based on the attribute distance to obtain the classification result of the non-time series alarm information.

[0071] In some embodiments, the second clustering module is further configured to:

[0072] Randomly selecting a second number of second initial cluster centers, and calculating a third similarity between the non-time series alarm information and the second initial cluster centers;

[0073] Based on the third similarity and the attribute distance, classifying the non-time series alarm information into a second cluster corresponding to the second initial cluster center;

[0074] The second initial cluster center is iteratively updated until the change value of the second initial cluster center in two adjacent iterations is less than a second threshold, and the non-time series alarm information is reclassified into a new second cluster to obtain the classification result of the non-time series alarm information.

[0075] In some embodiments, the merging unit comprises:

[0076] A second calculation module, configured to calculate a similarity between a cluster center of each first cluster in the time series alarm information classification result and a cluster center of each second cluster in the non-time series alarm information classification result to obtain a fourth similarity;

[0077] a merging module, configured to merge the first cluster and the second cluster whose fourth similarity is greater than a third threshold, to obtain a third cluster;

[0078] The generating module is used to integrate the unmerged first cluster, the unmerged second cluster and the third cluster to generate the alarm information classification result.

[0079] According to a third aspect of the present disclosure, there is provided an electronic device, including:

[0080] at least one processor; and

[0081] a memory communicatively connected to the at least one processor; wherein,

[0082] The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method described in the first aspect.

[0083] According to a fourth aspect of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to enable the computer to execute the method described in the first aspect.

[0084] According to a fifth aspect of the present disclosure, a computer program product is provided, comprising a computer program, wherein when the computer program is executed by a processor, the computer program implements the method as described in the first aspect above.

[0085] The present disclosure provides a method and device for alarm noise reduction, an electronic device and a storage medium, which obtain the status information of the system components, and analyze the status information based on the preset alarm rules to generate alarm information, wherein the alarm information includes time series alarm information and non-time series alarm information; the time series alarm information is input into the clustering network model for clustering processing to obtain the classification result of the time series alarm information; the non-time series alarm information is clustered using a preset clustering algorithm to obtain the classification result of the non-time series alarm information; the time series alarm information classification result and the non-time series alarm information classification result are merged to obtain the alarm information classification result. Compared with the related art, the embodiment of the present disclosure can more accurately capture the different characteristics in the alarm data by distinguishing the alarm information into time series alarm information and non-time series alarm information, and performing clustering processing respectively. In particular, for time series alarm data, the use of clustering network models for processing helps to fully extract and utilize the characteristics of time series data, thereby improving the accuracy of alarm suppression; by accurately capturing the characteristics of time series alarm data, it can more effectively handle the slight deviations introduced by time factors and improve the actual effectiveness of alarm suppression. By classifying and processing alarm information, it can significantly improve the accuracy and efficiency of alarm management, optimize the alarm rule system, and better cope with data scenarios with strict real-time requirements. This will help the operation and maintenance team to more effectively deal with the alarm storm phenomenon and improve the overall operation and maintenance level.

[0086] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present application, nor is it intended to limit the scope of the present application. Other features of the present application will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0087] The accompanying drawings are used to better understand the present solution and do not constitute a limitation of the present disclosure.

[0088] Figure 1 A flowchart of a method for reducing alarm noise provided by an embodiment of the present disclosure;

[0089] Figure 2 A flowchart of a method for clustering time series alarm information provided by an embodiment of the present disclosure;

[0090] Figure 3 A flowchart of a method for clustering non-time-series alarm information provided by an embodiment of the present disclosure;

[0091] Figure 4 A flowchart of another method for alarm noise reduction provided by an embodiment of the present disclosure;

[0092] Figure 5 A schematic diagram of the structure of a device for alarm noise reduction provided by an embodiment of the present disclosure;

[0093] Figure 6 A schematic diagram of the structure of another device for alarm noise reduction provided by an embodiment of the present disclosure;

[0094] Figure 7 A schematic block diagram of an exemplary electronic device provided for an embodiment of the present disclosure. DETAILED DESCRIPTION

[0095] The following is a description of exemplary embodiments of the present disclosure in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0096] The following describes the alarm noise reduction method and device, electronic device, and storage medium according to the embodiments of the present disclosure with reference to the accompanying drawings.

[0097] Figure 1 A flowchart of a method for alarm noise reduction provided by an embodiment of the present disclosure is provided.

[0098] like Figure 1 As shown, the method comprises the following steps:

[0099] Step 101, obtaining status information of system components, and analyzing the status information based on preset alarm rules to generate alarm information, wherein the alarm information includes sequential alarm information and non-sequential alarm information.

[0100] In the embodiments of the present disclosure, obtaining the status information of system components is the core link of operation and maintenance work, which involves real-time monitoring of various key components such as servers, network devices, and application services. Through the mobile cloud monitoring platform, various status data of these components are collected in real time, including operating status, performance indicators, resource usage, etc. These data can not only reflect the current status of the system, but also provide an important basis for subsequent alarm analysis.

[0101] After obtaining the status information, it is necessary to analyze this data based on the preset alarm rules. Alarm rules are formulated based on historical experience, business requirements, and system characteristics to determine whether the status information reaches the threshold for triggering an alarm. Through in-depth analysis and comparison of status information, abnormal situations in the system can be accurately identified and corresponding alarm information can be generated.

[0102] Alarm information includes two categories: time series alarm information and non-time series alarm information. Time series alarm information is data closely related to time series, and they usually reflect the continuous changes in system performance or resource usage. For example, key indicators such as CPU utilization, memory usage, and network bandwidth will fluctuate over time. By analyzing these time series data, potential performance bottlenecks, resource bottlenecks, or abnormal trends can be discovered, so that timely warnings can be issued and corresponding processing measures can be taken. In contrast, non-time series alarm information focuses more on information describing specific events or states. This information is usually not related to time series, but focuses on specific events that occur at a certain point in time or within a period of time. For example, system crashes, application errors, security incidents, etc., are all sudden and non-continuous. By analyzing these non-time series alarm information, the source of the problem can be quickly located and targeted solutions can be taken.

[0103] Therefore, in the mobile cloud monitoring platform, alarm information is divided into two categories: time-series alarm information and non-time-series alarm information, so as to more accurately analyze and process different types of alarm data. This not only helps to improve the efficiency and accuracy of alarm management, but also helps to reduce operation and maintenance costs and improve the stability and reliability of the overall system.

[0104] Step 102: input the time series alarm information into a clustering network model for clustering processing to obtain a classification result of the time series alarm information.

[0105] In an embodiment of the present disclosure, after obtaining the time series alarm information, in order to further explore the potential patterns and correlations therein, the embodiment of the present disclosure performs clustering processing on it through a clustering network model. The clustering network model is a powerful machine learning method that can automatically divide data into multiple groups or categories through unsupervised learning of the data. In practical applications, the time series alarm information is used as input data and sent to a pre-trained clustering network model. This model is built based on deep learning technology and has powerful feature extraction and classification capabilities. By learning and analyzing the features of the time series alarm information, the model can automatically identify the similarities and differences in the data and divide it into different categories.

[0106] During the clustering process, the model considers multiple dimensions of time series alarm information, including the timestamp, frequency, duration, and associated components or performance indicators of the alarm. These factors together constitute the feature space of time series alarm information, and the model will find the best clustering division method in this space.

[0107] After processing by the clustering network model, the classification results of the time series alarm information can be obtained. These classification results integrate the originally scattered alarm data into orderly and structured groups, each of which represents a specific alarm mode or trend. The classification results of the time series alarm information can more clearly reflect the operating status and potential risks of the system, and also provide strong support for subsequent alarm merging and noise reduction. By further analyzing these classification results, more accurate alarm strategies can be formulated to improve the efficiency and accuracy of alarm management.

[0108] Step 103: cluster the non-time series alarm information using a preset clustering algorithm to obtain a classification result of the non-time series alarm information.

[0109] In the embodiments of the present disclosure, non-time series alarm information generally includes event data, log data, etc., which are not closely related to the time series like time series alarm information, but are more focused on describing the changes of specific events or states. In order to effectively cluster these non-time series alarm information, the embodiments of the present disclosure are configured with clustering algorithms suitable for such data. These algorithms can fully consider the characteristics of non-time series alarm information, such as the type of event, the context of occurrence, the keywords of the log, etc. In the clustering process, the non-time series alarm information is first converted into feature vectors according to a preset feature extraction method, and these feature vectors can fully reflect the core information of the alarm information. Then, these feature vectors are input into the clustering algorithm, and the algorithm will iteratively calculate according to the similarity or distance between the vectors, and gradually divide the data into different categories. Through the processing of the clustering algorithm, the classification results of the non-time series alarm information can be obtained. These classification results integrate the originally chaotic non-time series alarm information into orderly groups, each of which represents a specific event type or state mode. Such classification results not only reflect the various events and state changes in the system, but also provide strong support for subsequent alarm merging, noise reduction and cause analysis.

[0110] Step 104: Combine the time series alarm information classification result and the non-time series alarm information classification result to obtain an alarm information classification result.

[0111] In the embodiment of the present disclosure, after completing the classification processing of the time series alarm information and the non-time series alarm information, it is necessary to further merge the classification results of the two types of alarm information. The merging process is intended to integrate the two different types of alarm information into a unified alarm information classification result, so as to understand the overall alarm status of the system more comprehensively and systematically.

[0112] During the merging process, the classification results of the time-series alarm information and the classification results of the non-time-series alarm information are first carefully compared and analyzed. By comparing this information, the potential correlation and mutual influence between the two types of alarm information can be analyzed and discovered. By adopting an appropriate merging strategy, the classification results of the two types of alarm information are merged. This may involve merging similar or related alarm information, deduplicating duplicate or redundant alarm information, and sorting out and integrating the correlations between different alarm information. Through the merging process, a more concise, clear and meaningful alarm information classification result can be obtained.

[0113] The present disclosure provides a method for alarm noise reduction, which obtains the status information of system components, and analyzes the status information based on preset alarm rules to generate alarm information, wherein the alarm information includes sequential alarm information and non-sequential alarm information; the sequential alarm information is input into a clustering network model for clustering processing to obtain a sequential alarm information classification result; the non-sequential alarm information is clustered using a preset clustering algorithm to obtain a non-sequential alarm information classification result; the sequential alarm information classification result and the non-sequential alarm information classification result are merged to obtain an alarm information classification result. Compared with the related art, the embodiment of the present disclosure can more accurately capture different characteristics in the alarm data by distinguishing the alarm information into sequential alarm information and non-sequential alarm information, and performing clustering processing respectively. In particular, for time series alarm data, the use of clustering network models for processing helps to fully extract and utilize the characteristics of time series data, thereby improving the accuracy of alarm suppression; by accurately capturing the characteristics of time series alarm data, it can more effectively handle the slight deviations introduced by time factors and improve the actual effectiveness of alarm suppression. By classifying and processing alarm information, it can significantly improve the accuracy and efficiency of alarm management, optimize the alarm rule system, and better cope with data scenarios with strict real-time requirements. This will help the operation and maintenance team to more effectively deal with the alarm storm phenomenon and improve the overall operation and maintenance level.

[0114] When faced with time series alarm information with a large number of instances and high dimensions in the monitoring platform, traditional clustering methods may not be able to quickly perform cluster analysis. At the same time, the presence of phase disturbances and random noise in the time series data makes it difficult to characterize similarities, and inappropriate similarity measures will reduce the accuracy of clustering. Therefore, the disclosed embodiment proposes a new type of time series alarm information (data) clustering algorithm, which can be simply summarized into three steps: data dimensionality reduction, time series data clustering, and allocation strategy to achieve fast and accurate clustering of large-scale time series data.

[0115] In order to clearly illustrate the embodiments of the present disclosure, this embodiment provides a flow chart of a method for clustering time series alarm information.

[0116] like Figure 2 As shown in the figure, the method includes the following steps:

[0117] Step 201: Randomly sample the time-series alarm information according to a preset number of sampling times and a preset sampling ratio to obtain a sampling set.

[0118] Step 202: Divide each time-series alarm information in the sampling set into a first number of data frames, and calculate the arithmetic mean of each data frame.

[0119] Step 203: Arrange the arithmetic means of each time-series alarm information in chronological order to generate the low-dimensional time-series alarm information.

[0120] Specifically, in steps 201 to 203, the time-series alarm information (data) has the characteristics of high dimension, a large number of random perturbations and noises. Therefore, for time-series alarm data, the embodiments of the present disclosure adopt a method of dimensionality reduction first and then clustering. First, for high-dimensional time-series data (time-series alarm information), it is transformed into a low-dimensional representation through dimensionality reduction technology to reduce the complexity of the data and the computational load.

[0121] The number of instances and dimensions are reduced by means of random sampling and dimensionality reduction. The sampled data set is used as the input of the clustering module to reduce the computational overhead. Since no assumptions are made about the distribution of the input data, random sampling is an effective means of reducing the number of data instances. Two principles need to be followed during the sampling process: (1) Set the minimum number of sampling times (i.e., the preset number of sampling times), and the data of each category is not less than the set minimum number of sampling times in the sampling set. (2) The proportion of data of each category in the sampling set (i.e., the preset sampling ratio) deviates from the proportion in the original data set by no more than a preset threshold. The original alarm data is randomly sampled based on the above principles.

[0122] For each time-series data (time-series alarm information) in the sampling set, the method of aggregating the average value segment by segment is used for dimensionality reduction. Specifically, for a time-series data with a length of D, it is divided into d frames (d < D), and then each frame is represented by the mean value of the data points on the frame, so as to reduce the length of the time-series data from D to d, achieving the purpose of dimensionality reduction. Finally, a set of time-series data feature vectors Z = {z 1 , z 2 ,..., z N} (low-dimensional time-series alarm information) is obtained, and N represents the total number of time-series data (time-series alarm information).

[0123] After data sampling and data dimension reduction operations, a time series alarm sampling data set of size s*d (s≤N,d≤D) can be obtained from a time series alarm data set of size N*D, and the sampling set keeps the distribution of the original data set unchanged. Using the sampling set as the input of the clustering module greatly reduces the computational overhead.

[0124] Step 204: Calculate the cluster center coefficient between the low-dimensional time series alarm information and the first initial cluster center, and determine the first initial cluster center based on the cluster center coefficient.

[0125] Step 205: Based on the first initial cluster centers, the clustering network model is called to perform clustering processing on the low-dimensional time series alarm information to obtain the model classification result.

[0126] Specifically, in step 204 to step 205, considering that the time series data (time series alarm information) is large in scale and has more noise, the data features after dimensionality reduction are clustered to achieve large-scale clustering of alarm vectors (time series alarm information).

[0127] The selection of cluster centers is crucial to the accuracy of clustering. The embodiment of the present disclosure defines formula (1) to obtain the coefficients of the feature vector and the cluster center.

[0128]

[0129] In formula (1), x represents the xth eigenvector, y represents the yth eigenvector, and z x and z y All of them belong to the eigenvectors after dimensionality reduction. u represents the expected value of the vector.

[0130] In order to initialize the cluster centers, the disclosed embodiment proposes a similarity measurement method for high-dimensional time series data (time series alarm information), which effectively solves the problem that commonly used similarity measurement algorithms such as Euclidean distance and cosine similarity algorithm cannot find features in time series data with strong continuity, as shown in formula (2).

[0131]

[0132] Among them, x x,y is the eigenvector z and k cluster centers w j After obtaining the similarity s, the feature vector closest to s is used as the first initial cluster center.

[0133] After determining the first initial cluster center, the clustering network model is used to cluster the low-dimensional time series alarm information. This step is the key link in alarm information analysis, which aims to merge alarm information with similar characteristics into the same category through clustering algorithms, thereby revealing the potential structure and pattern in the alarm data.

[0134] As an achievable manner of the embodiment of the present disclosure, before calling the clustering network model to perform clustering processing on the low-dimensional time series alarm information according to the first initial cluster center, the method further includes:

[0135] Based on a preset similarity algorithm, a first similarity between the training warning information and the training cluster core is calculated, and based on the first similarity, the training warning information is classified into a training cluster corresponding to the training cluster core;

[0136] The training cluster centers are iteratively trained until the change value of the training cluster centers between two adjacent iterations is less than a first threshold, thereby obtaining a trained clustering network model.

[0137] Specifically, the clustering layer constructs the initialized cluster center w j After that, the clustering layer is trained by alternating the following two steps.

[0138] a. Calculate the input feature vector z i The probability that (the feature vector corresponding to the training warning information) belongs to cluster j (that is, the training cluster), z i The characteristic distance of the cluster center w j The closer, the i The greater the probability of belonging to category j. i When input to the clustering layer, the first similarity between the training alarm information and the training cluster center can be calculated by using but not limited to the distance measurement method. i With cluster heart j The distance d ij , then for the distance d ij Regularization processing. If z i Distance from cluster center w j The distance is shown in formula (3). ij , we can set a suitable threshold ρ, if ρ ≥ d ij , then the eigenvector x i Belongs to cluster j, otherwise, it does not.

[0139]

[0140] In the above formula, z i represents the i-th eigenvector.

[0141] b. Update the cluster center. For each cluster (i.e., the training cluster), calculate the mean distance from the data in the cluster to the (training cluster center) as the new cluster center w j , as shown in formula (4).

[0142]

[0143] The clustering layer is then trained iteratively until the error of the center point change is less than the specified threshold p (the first threshold).

[0144] The threshold calculation is implemented by formula (5).

[0145]

[0146] Among them, w i and w j They are the cluster center before and after updating respectively.

[0147] By applying this algorithm, large-scale time series data can be automatically clustered to obtain k-class alarm vector data. For unclassified data, the allocation strategy is used for clustering.

[0148] Step 206: Calculate the second similarity between the unclassified time series alarm information and the cluster center of each first cluster in the model classification result.

[0149] Step 207: classify the unclassified time series alarm information based on the clustering radius of each of the first clusters and the second similarity to obtain the time series alarm information classification result.

[0150] Specifically, in step 206 to step 207, the classification of the unclassified time series alarm information is completed by comparing the features of the unclassified time series alarm information with the feature centers of the existing clusters, in order to find out the similarity between the unclassified time series alarm information and each cluster.

[0151] The calculation of similarity is usually based on distance measurement or similarity function, which can quantify the closeness between two objects. In the context of time series alarm information, similarity may involve multiple dimensions such as the time interval between alarms, alarm frequency, and similarity of associated components. For an unclassified time series alarm information, find the first cluster (classified instance A) with the closest similarity distance. If the distance between the two is less than the cluster radius of the first cluster where A is located, the unclassified time series alarm information is classified into the same cluster as instance A.

[0152] For non-time series alarm information (data), non-time series alarm data of the same type have similar characteristics. Therefore, for non-time series data, the embodiment of the present disclosure adopts a clustering method of vector similarity. Feature engineering is used to convert non-time series alarm information into vector representation, and each feature of the alarm is numerically or coded. Then, feature distance is used as a similarity measure between vectors, and their similarity is measured by calculating the distance between vectors.

[0153] In order to clearly illustrate the embodiments of the present disclosure, this embodiment provides a flowchart of a method for clustering non-time-series alarm information.

[0154] like Figure 3 As shown, the method comprises the following steps:

[0155] Step 301 : calculating the attribute distance between different non-time series alarm information, wherein the attribute distance is the characteristic distance of the same attribute information in different non-time series alarm information.

[0156] Specifically, in step 301, for the similarity clustering method, the attribute distance of the non-time series alarm information can be calculated using, but not limited to, the Euclidean distance function. Assume that the vectors of two different non-time series alarm information are represented as a i and a j , then first calculate the distance between two non-time series alarm information, which can be expressed as:

[0157]

[0158] Among them, dis(a i ,a j ) indicates warning information a i and a j The distance between them indicates that a in refers to the feature vector a i The n-th dimension feature of the warning information, m represents the feature dimension of the warning information.

[0159] Step 302: randomly select a second number of second initial cluster centers, and calculate a third similarity between the non-time series alarm information and the second initial cluster centers.

[0160] Step 303: Classify the non-time series alarm information into a second cluster corresponding to the second initial cluster center based on the third similarity and the attribute distance.

[0161] Step 304, iteratively update the second initial cluster center until the change value of the second initial cluster center in two adjacent iterations is less than a second threshold, and reclassify the non-time series alarm information into a new second cluster to obtain the classification result of the non-time series alarm information.

[0162] Specifically, in step 302 to step 304, assuming that the alarm vectors corresponding to the non-time-series alarm information are distributed in a space, firstly, k (a second number) of initial cluster centers (i.e., second initial cluster centers) are randomly selected, where k represents the number of clusters. 1 ,c 2 ,...,c k , and then use formula (4) to calculate each alarm vector ai The distance to the k cluster centers is selected, and the third closest similarity is selected to form a second cluster. When each alarm vector is divided into the corresponding second cluster, the first round of clustering ends and k second clusters S are formed. 1 ={S 11 ,S 12 ,...,S 1k}.

[0163] Based on the results of the first round of clustering, the cluster center is recalculated for each second cluster. j represents the jth clustering parameter. j As a new cluster center point. Finally, recalculate the similarity of the entire alarm vector object, re-divide the clusters, and repeat the whole process until the cluster center c j After n rounds of clustering iterations, the final k clusters S are obtained. n ={S n1 ,S n2 ,...,S nk}.

[0164] In order to clearly illustrate the embodiment of the present disclosure, this embodiment provides a flowchart of another method for alarm noise reduction.

[0165] like Figure 4 As shown, the method comprises the following steps:

[0166] Step 401, obtaining status information of system components, and analyzing the status information based on preset alarm rules to generate alarm information, wherein the alarm information includes sequential alarm information and non-sequential alarm information.

[0167] Step 402: Input the time series alarm information into a clustering network model for clustering processing to obtain a classification result of the time series alarm information.

[0168] Step 403: cluster the non-time series alarm information using a preset clustering algorithm to obtain a classification result of the non-time series alarm information.

[0169] For the description of steps 401 to 403, please refer to the above embodiment, which will not be described in detail in this embodiment.

[0170] Step 404 : perform similarity calculation on the cluster center of each first cluster in the time series alarm information classification result and the cluster center of each second cluster in the non-time series alarm information classification result to obtain a fourth similarity.

[0171] Step 405: merge the first cluster and the second cluster whose fourth similarity is greater than a third threshold to obtain a third cluster.

[0172] Step 406: Integrate the unmerged first cluster, the unmerged second cluster, and the third cluster to generate the alarm information classification result.

[0173] Specifically, in step 404 to step 406, the clustering result of the time series alarm information is where k 1 Indicates that there are k in the result 1 clusters, t1, t2 and tk 1 Represent the corresponding cluster centers respectively. The clustering result of non-time series alarm information is where k 2 Indicates that there are k in the result 2 clusters, g1, g2 and gk 2 They represent the corresponding cluster centers respectively.

[0174] After obtaining the clustering results of the time series alarm information and the non-time series alarm information, the cosine similarity is used to calculate S t The cluster centers and S g Take the similarity between cluster center t1 and cluster center g1 as an example, and use Simility(t1,g1) to represent the similarity between t1 and g1. The specific calculation method is expressed as:

[0175]

[0176] First, set a third threshold Φ, and then calculate S according to formula (6) t The cluster centers and S g The similarity of each cluster center in , when Simility(t1,g1)>Φ, the cluster S is considered t1 and cluster S g1 can be merged into one cluster, otherwise they cannot be merged. The final clustering result is expressed as S = {S 1 ,S 2 ,...,S k}, where k≤k 1 +k 2 .

[0177] Step 407: based on the alarm information classification result, deduplicate the alarm information in different alarm classifications.

[0178] Step 408: perform priority evaluation on the deduplicated alarm information classification and mark the corresponding priority.

[0179] Step 409: Call the alarm recovery rule corresponding to the alarm information, and perform fault recovery processing on the alarm information in the alarm classification in order of priority.

[0180] Specifically, in step 407 to step 409, after the classification results of the alarm information are obtained, the classification results are deduplicated to ensure that the same alarm information is not repeatedly responded to during the processing. Deduplication is a key step because in the actual operation and maintenance process, due to the complexity of the system or the sensitivity of the monitoring tool, repeated alarm information may appear. These duplicate information not only wastes resources, but also may interfere with the judgment and operation of the operation and maintenance personnel.

[0181] The specific method of deduplication processing can be customized according to the characteristics of the alarm information. For example, duplicate items can be identified based on the unique identifier of the alarm information (such as ID, timestamp and key attribute combination), and one of them can be retained as a representative. In this way, redundant data in the alarm information can be effectively eliminated, and the efficiency and accuracy of subsequent processing can be improved. After the deduplication processing is completed, the priority evaluation is performed on the classification of the deduplicated alarm information. The priority evaluation is determined based on factors such as the urgency, impact scope, and duration of the alarm information. By comprehensively considering these factors, a priority label such as high, medium, low or urgent, ordinary, etc. can be assigned to each alarm classification. The purpose of priority evaluation is to guide subsequent fault recovery processing. In the case of limited resources, giving priority to alarm information with higher priority can ensure the stable operation of key business systems and reduce potential losses. Therefore, this step is crucial to improving the effectiveness and efficiency of alarm management. After the priority evaluation is completed, the alarm recovery rules corresponding to the alarm information are called. These rules are a set of pre-defined operating procedures and policies that guide operation and maintenance personnel on how to handle different types of alarm information. Alarm recovery rules may include restarting services, switching backup devices, adjusting system parameters, etc., aiming to restore the normal operation of the system as soon as possible. According to the priority order, the alarm information in the alarm classification is processed for fault recovery. For alarm information with higher priority, more resources and efforts can be invested, but not limited to, to ensure that it is handled in a timely and effective manner. For alarm information with lower priority, a more relaxed processing strategy can be adopted, or it can be included in the pending list for follow-up tracking.

[0182] It should be noted that the embodiments of the present disclosure may include multiple steps. For the convenience of description, these steps are numbered, but these numbers do not limit the execution time slots or execution order between the steps; these steps can be implemented in any order, and the embodiments of the present disclosure do not limit this.

[0183] Corresponding to the above-mentioned alarm noise reduction method, the present invention also provides an alarm noise reduction device. Since the device embodiment of the present invention corresponds to the above-mentioned method embodiment, the details not disclosed in the device embodiment can be referred to the above-mentioned method embodiment, and will not be repeated in the present invention.

[0184] Figure 5 A schematic diagram of the structure of a device for alarm noise reduction provided by an embodiment of the present disclosure is shown in FIG. Figure 5 As shown, including:

[0185] The acquisition unit 51 is used to acquire the status information of the system components, and analyze the status information based on the preset alarm rules to generate alarm information, wherein the alarm information includes sequential alarm information and non-sequential alarm information;

[0186] A first clustering unit 52 is used to input the time series alarm information into a clustering network model for clustering processing to obtain a classification result of the time series alarm information;

[0187] A second clustering unit 53 is used to perform clustering processing on the non-time series alarm information using a preset clustering algorithm to obtain a classification result of the non-time series alarm information;

[0188] The merging unit 54 is used to merge the time series alarm information classification result and the non-time series alarm information classification result to obtain an alarm information classification result.

[0189] The present disclosure provides a device for alarm noise reduction, which obtains the status information of system components, and analyzes the status information based on preset alarm rules to generate alarm information, wherein the alarm information includes sequential alarm information and non-sequential alarm information; the sequential alarm information is input into a clustering network model for clustering processing to obtain a sequential alarm information classification result; the non-sequential alarm information is clustered using a preset clustering algorithm to obtain a non-sequential alarm information classification result; the sequential alarm information classification result and the non-sequential alarm information classification result are merged to obtain an alarm information classification result. Compared with the related art, the embodiment of the present disclosure can more accurately capture different characteristics in the alarm data by distinguishing the alarm information into sequential alarm information and non-sequential alarm information, and performing clustering processing on them respectively. In particular, for time series alarm data, the use of clustering network models for processing helps to fully extract and utilize the characteristics of time series data, thereby improving the accuracy of alarm suppression; by accurately capturing the characteristics of time series alarm data, it can more effectively handle the slight deviations introduced by time factors and improve the actual effectiveness of alarm suppression. By classifying and processing alarm information, it can significantly improve the accuracy and efficiency of alarm management, optimize the alarm rule system, and better cope with data scenarios with strict real-time requirements. This will help the operation and maintenance team to more effectively deal with the alarm storm phenomenon and improve the overall operation and maintenance level.

[0190] Furthermore, in a possible implementation of this embodiment, as Figure 6 As shown, the device also includes:

[0191] a deduplication unit 55, configured to perform deduplication processing on alarm information in different alarm classifications based on the alarm information classification result after the merging unit 54 merges the time series alarm information classification result and the non-time series alarm information classification result;

[0192] An evaluation unit 56, configured to perform priority evaluation on the deduplicated alarm information classification and mark the corresponding priority;

[0193] The recovery unit 57 is used to call the alarm recovery rule corresponding to the alarm information, and perform fault recovery processing on the alarm information in the alarm classification in order of priority.

[0194] Furthermore, in a possible implementation of this embodiment, as Figure 6 As shown, the first clustering unit 52 includes:

[0195] A dimension reduction module 521 is used to perform sampling and dimension reduction processing on the time series alarm information to obtain low-dimensional time series alarm information;

[0196] A first clustering module 522, configured to input the low-dimensional time series alarm information into the clustering network model, perform clustering processing on the low-dimensional time series alarm information, and obtain a model classification result;

[0197] The classification module 523 is used to classify the unclassified time series alarm information into the cluster category in the model classification result according to the model classification result, so as to obtain the time series alarm information classification result.

[0198] Furthermore, in a possible implementation of this embodiment, the dimension reduction module 521 is further used to:

[0199] According to a preset number of sampling times and a preset sampling ratio, the timing alarm information is randomly sampled to obtain a sampling set;

[0200] Divide each time series alarm information in the sampling set into a first number of data frames, and calculate the arithmetic mean of each of the data frames;

[0201] The arithmetic mean of each of the time series alarm information is arranged in chronological order to generate the low-dimensional time series alarm information.

[0202] Furthermore, in a possible implementation of this embodiment, the first clustering module 522 is further configured to:

[0203] Calculating a cluster center coefficient between the low-dimensional time series alarm information and a first initial cluster center, and determining the first initial cluster center based on the cluster center coefficient;

[0204] According to the first initial cluster center, the clustering network model is called to perform clustering processing on the low-dimensional time series alarm information to obtain the model classification result.

[0205] Furthermore, in a possible implementation of this embodiment, as Figure 6 As shown, the device also includes:

[0206] The first training unit 58 is used to calculate the first similarity between the training alarm information and the training cluster center based on a preset similarity algorithm before the first clustering unit calls the clustering network model to perform clustering processing on the low-dimensional time series alarm information according to the first initial cluster center, and classify the training alarm information into the training cluster cluster corresponding to the training cluster center based on the first similarity;

[0207] The second training unit 59 is used to iteratively train the training cluster centers until the change value of the training cluster centers between two consecutive iterations is less than a first threshold, thereby obtaining a trained clustering network model.

[0208] Furthermore, in a possible implementation of this embodiment, the classification module 523 is further configured to:

[0209] Calculating a second similarity between the unclassified time series alarm information and the cluster center of each first cluster in the model classification result;

[0210] Based on the clustering radius of each of the first clusters and the second similarity, the unclassified time series alarm information is classified to obtain the time series alarm information classification result.

[0211] Furthermore, in a possible implementation of this embodiment, as Figure 6 As shown, the second clustering unit 53 includes:

[0212] A first calculation module 531 is used to calculate the attribute distance between different non-time series alarm information, wherein the attribute distance is the characteristic distance of the same attribute information in different non-time series alarm information;

[0213] The second clustering module 532 is used to perform clustering processing on the non-time series alarm information based on the attribute distance to obtain the classification result of the non-time series alarm information.

[0214] Furthermore, in a possible implementation of this embodiment, the second clustering module 532 is further configured to:

[0215] Randomly selecting a second number of second initial cluster centers, and calculating a third similarity between the non-time series alarm information and the second initial cluster centers;

[0216] Based on the third similarity and the attribute distance, classifying the non-time series alarm information into a second cluster corresponding to the second initial cluster center;

[0217] The second initial cluster center is iteratively updated until the change value of the second initial cluster center in two adjacent iterations is less than a second threshold, and the non-time series alarm information is reclassified into a new second cluster to obtain the classification result of the non-time series alarm information.

[0218] Furthermore, in a possible implementation of this embodiment, as Figure 6 As shown, the merging unit 54 includes:

[0219] The second calculation module 541 is used to calculate the similarity between the cluster center of each first cluster in the time series alarm information classification result and the cluster center of each second cluster in the non-time series alarm information classification result to obtain a fourth similarity;

[0220] A merging module 542, configured to merge the first cluster and the second cluster whose fourth similarity is greater than a third threshold, to obtain a third cluster;

[0221] The generating module 543 is used to integrate the unmerged first cluster, the unmerged second cluster, and the third cluster to generate the alarm information classification result.

[0222] It should be noted that the above explanation of the method embodiment is also applicable to the device of this embodiment, and the principle is the same, which is not limited in this embodiment.

[0223] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium and a computer program product.

[0224] Figure 7 A schematic block diagram of an example electronic device 600 that can be used to implement an embodiment of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present disclosure described and / or required herein.

[0225] like Figure 7 As shown, the device 600 includes a computing unit 601, which can perform various appropriate actions and processes according to a computer program stored in a ROM (Read-Only Memory) 602 or a computer program loaded from a storage unit 608 to a RAM (Random Access Memory) 603. In the RAM 603, various programs and data required for the operation of the device 600 can also be stored. The computing unit 601, the ROM 602, and the RAM 603 are connected to each other via a bus 604. An I / O (Input / Output) interface 605 is also connected to the bus 604.

[0226] A number of components in the device 600 are connected to the I / O interface 605, including: an input unit 606, such as a keyboard, a mouse, etc.; an output unit 607, such as various types of displays, speakers, etc.; a storage unit 608, such as a disk, an optical disk, etc.; and a communication unit 609, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 609 allows the device 600 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0227] The computing unit 601 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 601 include, but are not limited to, a CPU (Central Processing Unit), a GPU (Graphic Processing Units), various dedicated AI (Artificial Intelligence) computing chips, various computing units running machine learning model algorithms, a DSP (Digital Signal Processor), and any appropriate processor, controller, microcontroller, etc. The computing unit 601 performs the various methods and processes described above, such as the method of alarm noise reduction. For example, in some embodiments, the method of alarm noise reduction may be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as a storage unit 608. In some embodiments, part or all of the computer program may be loaded and / or installed on the device 600 via the ROM 602 and / or the communication unit 609. When the computer program is loaded into the RAM 603 and executed by the computing unit 601, one or more steps of the method described above may be performed. Alternatively, in other embodiments, the computing unit 601 may be configured to execute the aforementioned alarm noise reduction method in any other appropriate manner (for example, by means of firmware).

[0228] Various embodiments of the systems and techniques described above herein may be implemented in digital electronic circuit systems, integrated circuit systems, FPGAs (Field Programmable Gate Arrays), ASICs (Application-Specific Integrated Circuits), ASSPs (Application Specific Standard Products), SOCs (System On Chips), CPLDs (Complex Programmable Logic Devices), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include: being implemented in one or more computer programs that may be executed and / or interpreted on a programmable system including at least one programmable processor that may be a dedicated or general-purpose programmable processor that may receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0229] The program code for implementing the method of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0230] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or device, or any suitable combination of the foregoing. More specific examples of machine-readable storage media may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a RAM, a ROM, an EPROM (Electrically Programmable Read-Only-Memory) or a flash memory, an optical fiber, a CD-ROM (Compact Dis sc Read-Only Memory), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0231] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (Cathode-Ray Tube) or LCD (Liquid Crystal Display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0232] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: LAN (Local Area Network), WAN (Wide Area Network), the Internet, and blockchain networks.

[0233] A computer system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The relationship between the client and the server is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services ("Virtual Private Server", or "VPS" for short). The server may also be a server of a distributed system, or a server combined with a blockchain.

[0234] It should be noted that artificial intelligence is a discipline that studies how computers can simulate certain human thought processes and intelligent behaviors (such as learning, reasoning, thinking, planning, etc.), and includes both hardware-level and software-level technologies. Artificial intelligence hardware technologies generally include technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, and big data processing; artificial intelligence software technologies mainly include computer vision technology, speech recognition technology, natural language processing technology, as well as machine learning / deep learning, big data processing technology, knowledge graph technology, and other major directions.

[0235] The various numerical numbers such as first and second involved in the present disclosure are only for the convenience of description and are not used to limit the scope of the embodiments of the present disclosure, but also indicate the order of precedence.

[0236] At least one in the present disclosure may also be described as one or more, and a plurality may be two, three, four or more, which is not limited in the present disclosure. In the embodiments of the present disclosure, for a technical feature, the technical features in the technical feature are distinguished by "first", "second", "third", "A", "B", "C" and "D", etc., and there is no order of precedence or size between the technical features described by the "first", "second", "third", "A", "B", "C" and "D".

[0237] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in this disclosure can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved, and this document does not limit this.

[0238] The above specific implementations do not constitute a limitation on the protection scope of the present disclosure. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present disclosure shall be included in the protection scope of the present disclosure.

Claims

1. A method for reducing alarm noise, characterized in that: include: Acquire status information of system components, and analyze the status information based on preset alarm rules to generate alarm information, wherein the alarm information includes sequential alarm information and non-sequential alarm information; Inputting the time series alarm information into a clustering network model for clustering processing to obtain a classification result of the time series alarm information; Using a preset clustering algorithm to perform clustering processing on the non-time series alarm information to obtain a classification result of the non-time series alarm information; The time series alarm information classification result and the non-time series alarm information classification result are combined to obtain an alarm information classification result.

2. The method according to claim 1, characterized in that After merging the time series alarm information classification result and the non-time series alarm information classification result, the method further includes: Based on the alarm information classification result, deduplication processing is performed on the alarm information in different alarm classifications; Prioritize the warning information after deduplication and mark the corresponding priority; The alarm recovery rule corresponding to the alarm information is called, and fault recovery processing is performed on the alarm information in the alarm classification in sequence according to the priority order.

3. The method according to claim 1, characterized in that The step of inputting the time series alarm information into a clustering network model for clustering processing to obtain a classification result of the time series alarm information includes: Performing sampling and dimensionality reduction processing on the time series alarm information to obtain low-dimensional time series alarm information; Inputting the low-dimensional time series alarm information into the clustering network model, performing clustering processing on the low-dimensional time series alarm information, and obtaining a model classification result; According to the model classification result, the unclassified time series alarm information is classified into the cluster category in the model classification result to obtain the time series alarm information classification result.

4. The method according to claim 3, characterized in that The sampling and dimensionality reduction processing is performed on the time series alarm information to obtain low-dimensional time series alarm information, including: According to a preset number of sampling times and a preset sampling ratio, the timing alarm information is randomly sampled to obtain a sampling set; Divide each time series alarm information in the sampling set into a first number of data frames, and calculate the arithmetic mean of each of the data frames; The arithmetic mean of each of the time series alarm information is arranged in chronological order to generate the low-dimensional time series alarm information.

5. The method according to claim 3, characterized in that: The step of inputting the low-dimensional time series alarm information into the clustering network model, performing clustering processing on the low-dimensional time series alarm information, and obtaining a model classification result includes: Calculating a cluster center coefficient between the low-dimensional time series alarm information and a first initial cluster center, and determining the first initial cluster center based on the cluster center coefficient; According to the first initial cluster center, the clustering network model is called to perform clustering processing on the low-dimensional time series alarm information to obtain the model classification result.

6. The method according to claim 5, characterized in that Before calling the clustering network model to perform clustering processing on the low-dimensional time series alarm information according to the first initial cluster center, the method further includes: Based on a preset similarity algorithm, a first similarity between the training warning information and the training cluster core is calculated, and based on the first similarity, the training warning information is classified into a training cluster corresponding to the training cluster core; The training cluster centers are iteratively trained until the change value of the training cluster centers between two adjacent iterations is less than a first threshold, thereby obtaining a trained clustering network model.

7. The method according to claim 3, characterized in that The step of classifying the unclassified time series alarm information into the cluster category in the model classification result according to the model classification result to obtain the time series alarm information classification result includes: Calculating a second similarity between the unclassified time series alarm information and the cluster center of each first cluster in the model classification result; Based on the clustering radius of each of the first clusters and the second similarity, the unclassified time series alarm information is classified to obtain the time series alarm information classification result.

8. The method according to claim 1, characterized in that The clustering process of the non-time-series alarm information by using a preset clustering algorithm to obtain a classification result of the non-time-series alarm information includes: Calculating the attribute distance between different non-time series alarm information, wherein the attribute distance is the characteristic distance of the same attribute information in different non-time series alarm information; Based on the attribute distance, clustering processing is performed on the non-time series alarm information to obtain a classification result of the non-time series alarm information.

9. The method according to claim 8, characterized in that The clustering process of the non-time series alarm information based on the attribute distance to obtain the classification result of the non-time series alarm information includes: Randomly selecting a second number of second initial cluster centers, and calculating a third similarity between the non-time series alarm information and the second initial cluster centers; Based on the third similarity and the attribute distance, classifying the non-time series alarm information into a second cluster corresponding to the second initial cluster center; The second initial cluster center is iteratively updated until the change value of the second initial cluster center in two adjacent iterations is less than a second threshold, and the non-time series alarm information is reclassified into a new second cluster to obtain the classification result of the non-time series alarm information.

10. The method according to claim 1, characterized in that The merging of the time series alarm information classification result and the non-time series alarm information classification result to obtain the alarm information classification result includes: Calculating the similarity between the cluster center of each first cluster in the time series alarm information classification result and the cluster center of each second cluster in the non-time series alarm information classification result to obtain a fourth similarity; Merging the first cluster and the second cluster whose fourth similarity is greater than a third threshold to obtain a third cluster; The unmerged first cluster, the unmerged second cluster, and the third cluster are integrated to generate the alarm information classification result.

11. A device for reducing alarm noise, characterized in that: include: An acquisition unit, used to acquire status information of system components, and analyze the status information based on preset alarm rules to generate alarm information, wherein the alarm information includes sequential alarm information and non-sequential alarm information; A first clustering unit, used for inputting the time series alarm information into a clustering network model for clustering processing to obtain a classification result of the time series alarm information; A second clustering unit is used to perform clustering processing on the non-time series alarm information by using a preset clustering algorithm to obtain a classification result of the non-time series alarm information; The merging unit is used to merge the time series alarm information classification result and the non-time series alarm information classification result to obtain the alarm information classification result.

12. An electronic device, characterized in that: include: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 10.

13. A non-transitory computer-readable storage medium storing computer instructions, characterized in that: The computer instructions are used to cause the computer to execute the method according to any one of claims 1-10.

14. A computer program product, characterized in that A computer program is included which, when executed by a processor, implements the method according to any one of claims 1 to 10.

Citation Information

Cited By

  • Short text-oriented alarm information rapid convergence method and system

    CN120910699A