WO2023179042A1 - Procédé de mise à jour de données, procédé de diagnostic de défaillance, dispositif électronique et support de stockage - Google Patents

Procédé de mise à jour de données, procédé de diagnostic de défaillance, dispositif électronique et support de stockage Download PDF

Info

Publication number
WO2023179042A1
WO2023179042A1 PCT/CN2022/130723 CN2022130723W WO2023179042A1 WO 2023179042 A1 WO2023179042 A1 WO 2023179042A1 CN 2022130723 W CN2022130723 W CN 2022130723W WO 2023179042 A1 WO2023179042 A1 WO 2023179042A1
Authority
WO
WIPO (PCT)
Prior art keywords
data
indicator
fault diagnosis
historical
historical indicator
Prior art date
Application number
PCT/CN2022/130723
Other languages
English (en)
Chinese (zh)
Inventor
刘煌
梁帅
Original Assignee
中兴通讯股份有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 中兴通讯股份有限公司 filed Critical 中兴通讯股份有限公司
Publication of WO2023179042A1 publication Critical patent/WO2023179042A1/fr

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F11/00—Error detection; Error correction; Monitoring
    • G06F11/07—Responding to the occurrence of a fault, e.g. fault tolerance
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F11/00—Error detection; Error correction; Monitoring
    • G06F11/30—Monitoring
    • G06F11/34—Recording or statistical evaluation of computer activity, e.g. of down time, of input/output operation ; Recording or statistical evaluation of user activity, e.g. usability assessment
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/23—Updating
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/24—Querying
    • G06F16/245—Query processing
    • G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/243—Classification techniques relating to the number of classes
    • G06F18/2433—Single-class perspective, e.g. one-against-all classification; Novelty detection; Outlier detection

Definitions

  • Embodiments of the present application relate to the field of communications, and in particular to a data update method, fault diagnosis method, electronic device and storage medium.
  • Big data systems are widely used not only in the Internet industry, but also in key fields such as government, finance, communications, aerospace, and military. This type of system usually has a massive user base of hundreds of millions. Service interruption or service quality degradation caused by any minor fault will bring huge losses. Moreover, such systems are large in scale, complex in composition and operation logic, resulting in frequent system failures. occurs, and it is difficult to find, locate and diagnose the fault after it occurs, and it is difficult to analyze and debug the system after it goes wrong.
  • the currently commonly used fault diagnosis method is fault diagnosis based on the graph neural network algorithm.
  • This method requires training the graph neural network based on existing abnormal data to obtain a fault diagnosis model, and completing the root cause of the fault through the trained fault diagnosis model and real-time data. diagnosis.
  • this method has low accuracy of fault diagnosis model due to the small sample size of abnormal data.
  • the main purpose of the embodiments of this application is to propose a data update method, fault diagnosis method, electronic device and storage medium, and update the fault diagnosis model through dynamic thresholds so that the fault diagnosis model can accurately locate faults and be applicable to various scenarios.
  • embodiments of the present application provide a data update method, which includes: obtaining historical indicator data of the business, classifying the historical indicator data according to the change pattern of the historical indicator data over time, and obtaining the classification results. ; Perform anomaly detection on the historical indicator data according to the anomaly detection algorithm corresponding to the classification result, and determine the abnormal data in the historical indicator data; update the preset indicators of the business according to the distribution of the abnormal data threshold.
  • Embodiments of the present application provide a fault diagnosis method, which includes: obtaining fault data of a service; when the indicator threshold of the service is not updated, analyzing the fault data according to a preset fault diagnosis model, generating and outputting Fault diagnosis results, wherein the preset fault diagnosis model includes a preset indicator threshold; when the indicator threshold of the service has been updated, the latest indicator threshold is obtained, and the preset indicator threshold is calculated based on the latest indicator threshold.
  • the fault diagnosis model is updated, the updated fault diagnosis model is used to analyze the fault data, and fault diagnosis results are generated and output, wherein the latest index threshold is obtained through the data update method described in the above embodiment.
  • an embodiment of the present application also provides an electronic device, including: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores information that can be processed by the at least one processor.
  • Instructions executed by a processor the instructions are executed by the at least one processor, so that the at least one processor can perform the data update method described in the above embodiment, or can perform the fault diagnosis described in the above embodiment. method.
  • embodiments of the present application also propose a computer-readable storage medium that stores a computer program.
  • the computer program is executed by a processor, the data update method described in the above embodiments is implemented, or the above implementation can be performed.
  • the fault diagnosis method described in the example is not limited to, but not limited to, but not limited to, but not limited to, but not limited to, but not limited to, but not limited to, but not limited to, but not limited to, but not limited to the above embodiments, the data update method described in the above embodiments.
  • This application proposes a data update method that obtains classification results by classifying historical indicator data of the business.
  • the classification results indicate the change pattern of historical indicator data over time.
  • the corresponding anomaly detection algorithm is selected according to the type of historical indicator data to analyze the historical indicators. Detect the data, identify abnormal data, and then determine or update the business indicator thresholds based on the distribution of abnormal data.
  • this application can periodically or continuously obtain historical indicator data of the business, determine the latest indicator threshold based on the latest historical indicator data, and obtain the current operating status of the business in a timely manner through this dynamic threshold.
  • a fault diagnosis method proposed in this application obtains the fault data of the service in real time, and then determines whether the indicator threshold of the service has been updated. If the indicator threshold of the service has not been updated, the preset fault diagnosis model and the preset indicator threshold are used. Analyze the fault data and obtain the fault diagnosis results (the cause of the fault). If the indicator threshold of the business has been updated, the preset fault diagnosis model is updated according to the latest indicator threshold, and the updated fault diagnosis model is used to complete fault diagnosis and determine the cause of the fault.
  • the fault diagnosis model of this application is not static, but is constantly updated based on the latest indicator thresholds, and the latest indicator thresholds are determined based on the latest historical indicator data. In this way, the fault diagnosis model can be timely based on the current operation of the business.
  • the adaptive adjustment of state changes makes the model more suitable for actual business scenarios, thereby improving the accuracy of fault diagnosis results.
  • Figure 1 is a flow chart of a data update method provided by an embodiment of the present application.
  • Figure 2 is a flow chart of a fault diagnosis method provided by an embodiment of the present application.
  • FIG. 3 is a schematic structural diagram of an electronic device provided by an embodiment of the present application.
  • the currently commonly used fault diagnosis method is fault diagnosis based on the graph neural network algorithm.
  • This method often abstracts the big data system into a network graph and obtains exceptions from all nodes in the network graph (such as servers, clients, etc.) Data is used to train the pre-created graph neural network to obtain a fault diagnosis model.
  • a fault occurs, abnormal data is input into the fault diagnosis model in real time to complete the root cause diagnosis of the fault.
  • this method needs to prepare a large amount of abnormal data in the early training stage, and it is abnormal data that occurs when a fault occurs, and this kind of abnormal data often has a small sample size. If abnormal data with a small sample size is used for training, the fault diagnosis obtained The accuracy of the model is low; if the model is iteratively trained by accumulating abnormal data, the cost is extremely high and it takes a long time.
  • the embodiment of the present application relates to a data update method, as shown in Figure 1, including:
  • Step 101 Obtain the historical indicator data of the business, classify the historical indicator data according to the change pattern of the historical indicator data over time, and obtain the classification results.
  • the historical indicator data of the business includes: various indicator data related to the operation of the business itself and indicator data of the underlying platform on which the business operation depends.
  • [Business Service A] is located in the business center.
  • the various data required by business service A come from [Other Services 1] and [Other Services 2] in the data center, and the data center relies on the underlying PaaS platform to complete it.
  • the data center relies on the underlying PaaS platform to complete it.
  • the content of historical indicator data may be different.
  • the updated indicator thresholds must conform to the current operating status of the business as much as possible. Therefore, when obtaining historical indicator data, try to select the data closest to the current moment, that is, the latest historical indicator data. For example: For a certain business, it is necessary to obtain a week's historical indicator data to determine the indicator threshold. Then it is best to select the data of the week before the current time as the historical indicator data, rather than selecting the data of a certain week a year ago to calculate the indicator threshold. .
  • the indicator threshold of the business can be obtained through the data update method of this application. If the indicator threshold of the business has been determined, the index threshold of the business can be continuously updated through the data update method of this application.
  • Some indicator thresholds, or in other words, historical indicator thresholds are constantly updated. In other words, when this application obtains historical indicator data, it can be obtained continuously, in real time, or periodically.
  • classifying the historical indicator data according to the change pattern of the historical indicator data over time and obtaining the classification results includes: obtaining the autocorrelation coefficient amplitude spectrum of the historical indicator data, and based on the The autocorrelation coefficient amplitude spectrum determines whether the historical indicator data is periodic; obtains the mean and variance of the historical indicator data, and determines whether the historical indicator data is volatile based on the mean and variance; The indicator data is linearly fitted, a fitting function is obtained, and whether the historical indicator data has a trend is determined based on the fitting function.
  • the periodicity of historical indicator data first obtain the autocorrelation coefficient map of the historical indicator data, perform Fourier transform on the autocorrelation coefficient map to obtain the autocorrelation coefficient amplitude spectrum, and calculate the maximum value of the autocorrelation coefficient amplitude spectrum. , mean and variance. When the maximum value is greater than the sum of the mean and the variance, the historical indicator data is determined to be periodic data. When the maximum value is less than or equal to the sum of the mean and the variance, the historical indicator data is determined to be aperiodic data.
  • the autocorrelation coefficient of historical indicator data can also be directly calculated. If the autocorrelation coefficient is less than the preset threshold, the historical indicator data can be directly determined to be non-periodic data. However, when the autocorrelation coefficient is greater than the preset threshold, it cannot be determined that the historical indicator data is periodic data, because the autocorrelation coefficient of non-periodic data may also be greater than the preset threshold. In one embodiment, for historical indicator data whose autocorrelation coefficient is greater than a preset threshold, its autocorrelation coefficient amplitude spectrum can be obtained. If its amplitude spectrum has no peaks, it can be directly determined that the historical indicator data is aperiodic data.
  • the strength of the periodicity can be further determined, and different anomaly detection algorithms can be selected according to the strength of the periodicity. For example: for strongly periodic data, you can choose a simpler and less complex anomaly detection algorithm; for weakly periodic data, you can choose an anomaly detection algorithm with better detection results and stronger compatibility.
  • the maximum peak value of the main peak of the autocorrelation coefficient amplitude spectrum can be obtained, which is determined by the following formula:
  • value max is the maximum value of the amplitude spectrum
  • ⁇ is the mean value
  • ⁇ is the variance
  • peak max is the maximum peak value of the main wave peak
  • ⁇ is the preset discrimination rate.
  • the mean and variance of historical indicator data can be obtained. When the historical indicator data meets any of the following conditions, it is determined that it has no volatility. Otherwise, the historical indicator data has volatility:
  • ⁇ is the mean value of historical indicator data
  • ⁇ is the variance of historical indicator data
  • ⁇ is the preset adjustment parameter
  • the historical indicator data can be linearly fitted, a fitting function is obtained, and whether the overall data has an upward or downward trend is determined based on the type of the fitting function or the slope of the fitting function. trend.
  • Step 102 Detect the historical indicator data according to the anomaly detection algorithm corresponding to the classification result, and determine the abnormal data in the historical indicator data.
  • a regression algorithm is used to perform anomaly detection on the historical indicator data; when the classification result is that the historical indicator data is aperiodic data, a clustering algorithm is used to detect the historical indicator data. Anomaly detection is performed on the indicator data; when the classification result is that the historical indicator data is non-volatile data, the box plot algorithm is used to detect anomalies in the historical indicator data; when the classification result is that the historical indicator data is trend data, time series decomposition is used The algorithm performs anomaly detection on historical indicator data.
  • the corresponding anomaly detection algorithm is selected according to the type of historical indicator data.
  • the data trend shows a certain periodic pattern over time and there is a correlation between the data, so the regression algorithm is used to detect anomalies.
  • Detection and regression algorithms can be Histogram-based decision tree algorithms, Leaf-wise algorithms with depth restrictions, unilateral gradient sampling algorithms, etc.
  • non-periodic historical indicator data the data is volatile, but the fluctuation patterns are messy and non-periodic, so a clustering algorithm is used to detect anomalies.
  • the clustering algorithm can be Gaussian mixture model algorithm, K-means clustering algorithm, mean shift clustering algorithm, density-based clustering algorithm, agglomerative hierarchical clustering algorithm, etc.
  • Time series decomposition algorithms can be X11 decomposition algorithm (X11 decomposition), classic decomposition algorithm, SEATS decomposition algorithm (SEATS decomposition), STL decomposition algorithm (STL decomposition), etc.
  • Step 103 Update the business indicator threshold according to the distribution of abnormal data.
  • updating the preset indicator threshold of the business according to the distribution of the abnormal data includes: determining the upper quartile and lower quartile of the abnormal data according to the distribution of the abnormal data. number and interquartile range; determine the current indicator threshold of the business based on the upper quartile, the lower quartile and the interquartile range, and use the current indicator threshold to replace the preset indicator threshold.
  • the lower quartile Q1, the median Q2, and the upper quartile Q3 are obtained.
  • the 25th, 50th, and 75th percentile numbers are the lower quartile Q1, the median Q2, and the upper quartile Q3 respectively.
  • the upper quartile and the lower quartile The difference between the quartiles is the interquartile range IQR, the upper limit of the indicator threshold is Q3+1.5 ⁇ IQR, and the lower limit of the indicator threshold is Q3+1.5 ⁇ IQR.
  • the data update method is explained.
  • a clustering algorithm is used for anomaly detection, that is, the entire CPU usage data is clustered according to the set number of categories, and then the abnormal ones are found according to the discrimination rules. That category or categories and label them. In order to achieve the purpose of anomaly detection.
  • This application proposes a data update method that obtains classification results by classifying historical indicator data of the business.
  • the classification results indicate the change pattern of historical indicator data over time.
  • the corresponding anomaly detection algorithm is selected according to the type of historical indicator data to analyze the historical indicators. Detect the data, identify abnormal data, and then determine or update the business indicator thresholds based on the distribution of abnormal data.
  • this application can periodically or continuously obtain historical indicator data of the business, determine the latest indicator threshold based on the latest historical indicator data, and obtain the current operating status of the business in a timely manner through this dynamic threshold.
  • the embodiment of the present application relates to a fault diagnosis method, as shown in Figure 2, including:
  • Step 201 Obtain service fault data.
  • the fault data includes: log data, alarm data, indicator data, third-party operation and maintenance data, etc.
  • the specific fault data acquisition time can be to obtain the business fault data in real time when the fault occurs, and then store the fault data so that it can be called during subsequent fault diagnosis. It can also be obtained from relevant equipment or platforms during fault diagnosis.
  • fault diagnosis methods are deployed in the data center in the form of modules.
  • the data center is based on the data warehouse and data platform, and produces data into data API services to provide it to the business in a more efficient way.
  • [Business Service A] is located in the business center.
  • the various data required by business service A come from [Other Services 1] and [Other Services 2] in the data center.
  • the data center relies on the underlying PaaS platform to complete industry applications. Support, then when obtaining fault data, you need to obtain fault data of the PaaS platform, fault data of the data center, fault data of the business center, etc.
  • Step 202 When the service indicator threshold is not updated, the fault data is analyzed according to a preset fault diagnosis model, and a fault diagnosis result is generated and output, where the preset fault diagnosis model includes the preset indicator threshold.
  • Step 203 When the indicator threshold of the service has been updated, the latest indicator threshold is obtained, the preset fault diagnosis model is updated according to the latest indicator threshold, and the fault data is analyzed using the updated fault diagnosis model. Analyze, generate and output fault diagnosis results, where the latest indicator threshold is obtained through the data update method described in the above embodiment.
  • the update of the indicator threshold is obtained through the data update method described in the above embodiment, and the data update method can be deployed on a certain device or platform in the form of a module, and the fault diagnosis method can also be deployed on a certain device or platform in the form of a module. On the same or different device or platform as the data update method.
  • the preset indicator threshold in the fault diagnosis model (the preset indicator threshold can be initially manually set, or it can be the indicator threshold at a historical moment obtained through the data update method), and the updated fault diagnosis model is obtained.
  • the preset fault diagnosis model includes multiple fault diagnosis templates, and the fault diagnosis templates are created according to a rule engine and preset fault diagnosis conditions.
  • the rule engine can be a Drools engine, an Open Tablets engine, an Easy Rules engine, etc.
  • the rule engine converts preset fault diagnosis conditions into fault diagnosis templates.
  • a fault diagnosis model usually includes multiple fault diagnosis templates.
  • the thresholds in the rule engine are generally fixed values, which cannot satisfy fault diagnosis in various scenarios of the same business. Therefore, this application also combines the data update method to obtain dynamic thresholds. In this way, the fault diagnosis model of this application does not need to collect a large amount of abnormal data. Conduct training and adaptively adjust the threshold according to the current operating status of the business. The entire fault diagnosis method has low cost, short model construction time and high accuracy.
  • the fault diagnosis method of this application is explained by taking the sudden drop in the number of Yarn's running tasks as an example.
  • the preset fault diagnosis conditions include: (1) If the task submission service is down, then the number of Yarn's running tasks will be caused. The reason for the decrease is that the submitting node is down; (2) If the submitting task service is not down, but the Pg service is abnormal, then the reason for the decrease in the number of Yarn running tasks is that the Pg service is abnormal; (3) If the submitting task service is different from the Pg service None of them are abnormal, but the task initialized at the current moment is abnormal, so the reason for the decrease in the number of running tasks in Yarn is the task initialization exception. Create multiple fault diagnosis templates through the above fault diagnosis conditions and rule engine, and create a fault diagnosis model based on multiple fault diagnosis templates.
  • the indicator data involved in this failure are (1) the number of submitted task services; (2) the number of Pg service processes; (3) the number of initial tasks.
  • the corresponding indicator thresholds are fixed values, so there is no need for data update methods to update. And whether there is any abnormality in the initial number of tasks in the current hour (how many tasks are initialized before how many tasks need to be submitted), This cannot be judged using a fixed threshold, because different nodes have different task volumes, so the task volume is different at different hours every day, and the number of tasks at the same hour on different days is not necessarily the same, so the data update method is used to continuously update
  • the indicator threshold of the number of initial tasks is continuously updated, and the fault diagnosis model is continuously updated to complete the root cause diagnosis of the sudden drop in the number of running tasks in Yarn.
  • a fault diagnosis method proposed in this application obtains the fault data of the service in real time, and then determines whether the indicator threshold of the service has been updated. If the indicator threshold of the service has not been updated, the preset fault diagnosis model and the preset indicator threshold are used. Analyze the fault data and obtain the fault diagnosis results (the cause of the fault). If the indicator threshold of the business has been updated, the preset fault diagnosis model is updated according to the latest indicator threshold, and the updated fault diagnosis model is used to complete fault diagnosis and determine the cause of the fault.
  • the fault diagnosis model of this application is not static, but is constantly updated based on the latest indicator thresholds, and the latest indicator thresholds are determined based on the latest historical indicator data. In this way, the fault diagnosis model can be timely based on the current operation of the business.
  • the adaptive adjustment of state changes makes the model more suitable for actual business scenarios, thereby improving the accuracy of fault diagnosis results.
  • the embodiment of the present application relates to an electronic device, as shown in Figure 3, including: at least one processor 301; and a memory 302 communicatively connected with the at least one processor 301; wherein the memory 302 stores data that can Instructions executed by the at least one processor 301, the instructions are executed by the at least one processor 301, so that the at least one processor 301 can perform the fault diagnosis method of the above embodiment, or can perform the above embodiment data update method.
  • the bus can include any number of interconnected buses and bridges.
  • the bus connects one or more processors and various circuits of the memory together.
  • the bus may also connect various other circuits together such as peripherals, voltage regulators, and power management circuits, which are all well known in the art and therefore will not be described further herein.
  • the bus interface provides the interface between the bus and the transceiver.
  • a transceiver may be one element or may be multiple elements, such as multiple receivers and transmitters, providing a unit for communicating with various other devices over a transmission medium.
  • the data processed by the processor is transmitted over the wireless medium through the antenna. Further, the antenna also receives the data and transmits the data to the processor.
  • the processor is responsible for managing the bus and general processing, and can also provide a variety of functions, including timing, peripheral interfaces, voltage regulation, power management, and other control functions.
  • Memory can be used to store data used by the processor when performing operations.
  • Embodiments of the present application relate to a computer-readable storage medium storing a computer program.
  • the computer program is executed by the processor, the above fault diagnosis method or data update method is implemented.
  • the program is stored in a storage medium and includes several instructions to cause a device ( It may be a microcontroller, a chip, etc.) or a processor (processor) that executes all or part of the steps of the method described in each embodiment of the application.
  • the aforementioned storage media include: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), magnetic disk or optical disk and other media that can store program code. .

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Databases & Information Systems (AREA)
  • Quality & Reliability (AREA)
  • Software Systems (AREA)
  • Artificial Intelligence (AREA)
  • Mathematical Physics (AREA)
  • Computational Linguistics (AREA)
  • Computer Hardware Design (AREA)
  • Fuzzy Systems (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Probability & Statistics with Applications (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • Debugging And Monitoring (AREA)

Abstract

L'invention concerne un procédé de mise à jour de données, un procédé de diagnostic de défaillance, un dispositif électronique et un support de stockage, qui se rapportent au domaine des communications. Le procédé de mise à jour de données consiste à : acquérir des données d'index historiques d'un service, selon une règle de modification des données d'index historiques au fil du temps, classifier les données d'index historiques et acquérir un résultat de classification (101) ; selon un algorithme de détection d'anomalies correspondant au résultat de classification, effectuer une détection d'anomalies sur les données d'index historiques et déterminer des données anormales dans les données d'index historiques (102) ; et selon la distribution des données anormales, mettre à jour un seuil d'index prédéfini du service (103).
PCT/CN2022/130723 2022-03-21 2022-11-08 Procédé de mise à jour de données, procédé de diagnostic de défaillance, dispositif électronique et support de stockage WO2023179042A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202210283753.2A CN116821141A (zh) 2022-03-21 2022-03-21 数据更新方法、故障诊断方法、电子设备和存储介质
CN202210283753.2 2022-03-21

Publications (1)

Publication Number Publication Date
WO2023179042A1 true WO2023179042A1 (fr) 2023-09-28

Family

ID=88099730

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/130723 WO2023179042A1 (fr) 2022-03-21 2022-11-08 Procédé de mise à jour de données, procédé de diagnostic de défaillance, dispositif électronique et support de stockage

Country Status (2)

Country Link
CN (1) CN116821141A (fr)
WO (1) WO2023179042A1 (fr)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117591607A (zh) * 2024-01-19 2024-02-23 杭州青橄榄网络技术有限公司 数据同步管理方法与系统

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20170094537A1 (en) * 2015-09-24 2017-03-30 Futurewei Technologies, Inc. System and Method for a Multi View Learning Approach to Anomaly Detection and Root Cause Analysis
CN110807024A (zh) * 2019-10-12 2020-02-18 广州市申迪计算机系统有限公司 动态阈值异常检测方法、系统、存储介质及智能设备
CN112712113A (zh) * 2020-12-29 2021-04-27 广州品唯软件有限公司 一种基于指标的告警方法、装置及计算机系统
CN113010389A (zh) * 2019-12-20 2021-06-22 阿里巴巴集团控股有限公司 一种训练方法、故障预测方法、相关装置及设备
CN113887616A (zh) * 2021-09-30 2022-01-04 海看网络科技(山东)股份有限公司 一种epg连接数的实时异常检测系统及方法

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20170094537A1 (en) * 2015-09-24 2017-03-30 Futurewei Technologies, Inc. System and Method for a Multi View Learning Approach to Anomaly Detection and Root Cause Analysis
CN110807024A (zh) * 2019-10-12 2020-02-18 广州市申迪计算机系统有限公司 动态阈值异常检测方法、系统、存储介质及智能设备
CN113010389A (zh) * 2019-12-20 2021-06-22 阿里巴巴集团控股有限公司 一种训练方法、故障预测方法、相关装置及设备
CN112712113A (zh) * 2020-12-29 2021-04-27 广州品唯软件有限公司 一种基于指标的告警方法、装置及计算机系统
CN113887616A (zh) * 2021-09-30 2022-01-04 海看网络科技(山东)股份有限公司 一种epg连接数的实时异常检测系统及方法

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117591607A (zh) * 2024-01-19 2024-02-23 杭州青橄榄网络技术有限公司 数据同步管理方法与系统
CN117591607B (zh) * 2024-01-19 2024-05-07 杭州青橄榄网络技术有限公司 数据同步管理方法与系统

Also Published As

Publication number Publication date
CN116821141A (zh) 2023-09-29

Similar Documents

Publication Publication Date Title
US10311044B2 (en) Distributed data variable analysis and hierarchical grouping system
US11403164B2 (en) Method and device for determining a performance indicator value for predicting anomalies in a computing infrastructure from values of performance indicators
CN103513983B (zh) 用于预测性警报阈值确定工具的方法和系统
US20190080253A1 (en) Analytic system for graphical interpretability of and improvement of machine learning models
US11288577B2 (en) Deep long short term memory network for estimation of remaining useful life of the components
US20170097980A1 (en) Detection method and information processing device
US20230267032A1 (en) Using an event graph schema for root cause identification and event classification in system monitoring
WO2021105927A1 (fr) Surveillance et analyse de performances d'apprentissage machine
WO2021002780A1 (fr) Système de surveillance de qualité et de processus reposant sur un apprentissage machine
CN111949429A (zh) 基于密度聚类算法的服务器故障监测方法及系统
WO2023179042A1 (fr) Procédé de mise à jour de données, procédé de diagnostic de défaillance, dispositif électronique et support de stockage
CN115699209A (zh) 用于人工智能(ai)模型选择的方法
CN116416884B (zh) 一种显示器模组的测试装置及其测试方法
CN115705501A (zh) 机器学习数据处理管道的超参数空间优化
US11792081B2 (en) Managing telecommunication network event data
US20240142922A1 (en) Analysis method, analysis program and information processing device
CN114202256A (zh) 架构升级预警方法、装置、智能终端及可读存储介质
CN117540826A (zh) 机器学习模型的优化方法、装置、电子设备及存储介质
CN111400122A (zh) 一种硬盘健康度评估方法及装置
US20230036483A1 (en) Data Anomaly Forecasting From Data Record Meta-Statistics
Landin et al. An intelligent monitoring algorithm to detect dependencies between test cases in the manual integration process
US11711257B1 (en) Systems and methods for automating incident severity classification
US12007829B2 (en) Extended dynamic intelligent log analysis tool
US11954615B2 (en) Model management for non-stationary systems
CN117539948B (zh) 基于深度神经网络的业务数据检索方法及装置

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22933093

Country of ref document: EP

Kind code of ref document: A1