CN111400284B - Method for establishing dynamic anomaly detection model based on performance data - Google Patents
Method for establishing dynamic anomaly detection model based on performance data Download PDFInfo
- Publication number
- CN111400284B CN111400284B CN202010198615.5A CN202010198615A CN111400284B CN 111400284 B CN111400284 B CN 111400284B CN 202010198615 A CN202010198615 A CN 202010198615A CN 111400284 B CN111400284 B CN 111400284B
- Authority
- CN
- China
- Prior art keywords
- performance data
- performance
- data
- time
- acquisition
- Prior art date
- Legal status (The legal status 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 status listed.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; 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
- G06F16/21—Design, administration or maintenance of databases
- G06F16/211—Schema design and management
- G06F16/212—Schema design and management with details for data modelling support
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Software Systems (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Databases & Information Systems (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Artificial Intelligence (AREA)
- Debugging And Monitoring (AREA)
Abstract
The application belongs to the technical field of information, and relates to a dynamic modeling technology, in particular to a method for establishing a dynamic anomaly detection model based on performance data, which comprises a performance data collector and a performance processing server, wherein the data collector collects data according to the collection frequency defined in a retention strategy used by the measurement of an object; the performance processing server is responsible for storing the performance data sent by the performance data collector, and storing the acquisition frequency time stamp, the object, the measurement type and the aggregate value which accord with the retention policy into the time sequence database. The application uses the performance engine to perform configurable collection and retention strategy on the performance data, uses the time sequence database storage of the dynamic expansion of the object to efficiently support the dynamic anomaly detection. The problems of low efficiency in processing performance data, interface solidification, poor adaptability, poor expansibility and the like in the prior art are solved.
Description
Technical Field
The application belongs to the technical field of information, and relates to a dynamic modeling technology, in particular to a method for establishing a dynamic anomaly detection model based on performance data.
Background
With equipment and application, the explosive quantity of intelligent Internet of things is increased and scenes are rich, the traditional database is difficult to process historical performance data efficiently, and static modeling cannot meet the requirements of effective anomaly analysis and accurate anomaly positioning on the performance data. The application mainly provides dynamic modeling for a single resource measurement object, and improves variability of anomaly detection modeling and accuracy of anomaly determination.
The information age is moving into a new stage, and innovation and development of technology are accompanied by massive growth of equipment, applications and data. Conventional monitoring and performance systems have evolved for decades, and the tools and means for centralized monitoring have grown more and more, and the management and analysis schemes for performance data have been relatively single. The existing performance analysis and management is usually another integrated module of a relatively independent centralized monitoring and resource or configuration system, mainly comprising the following four-point functions, and the functional flow sequence is as follows:
1) Controlling the association of performance data with resources, supporting configuration of association rules related to resource attributes.
2) The measurement instance object, which may be accurate to the resource instance, is statically defined and a threshold or interval is established.
3) If the performance data does not exceed the established threshold or interval, the data will be stored as normal performance value data. If the performance data exceeds the established threshold or interval, an anomaly is located and an anomaly alert is generated.
4) If integrated with the monitoring alarm processing and flow platform, the alarms can be processed conventionally, including alarm analysis, conversion of alarms into fault worksheets, alarm troubleshooting repair task processing, etc.
In the prior art, a performance management and analysis module only supports static modeling of resource objects, and sets a threshold or interval for measurement and resource instances to be accurate, so that abnormality is judged. The static anomaly detection model lacks dynamic perception capability, the prior art does not have predictive analysis processing capability, anomaly detection dynamic modeling cannot be performed through historical performance data of resources, and rules and machine learning cannot be implanted in the model to perform predictive analysis. The anomaly analysis of the existing performance data would face accuracy problems and cannot develop multidimensional analysis and prediction.
Disclosure of Invention
The application aims at the problems and provides a method for establishing a dynamic anomaly detection model based on performance data.
In order to achieve the aim, the application adopts the technical proposal that,
a method for building dynamic abnormality detection model based on performance data comprises a performance data collector and a performance processing server,
the data collector collects data at a collection frequency defined in a retention policy used by the metrics of the object;
the performance processing server is responsible for storing the performance data sent by the performance data collector, and storing the acquisition frequency time stamp, the object, the measurement type and the aggregate value which accord with the retention policy into the time sequence database.
Preferably, the reservation policy supports that different reservation rules are used for each acquisition period definition, including data acquisition frequency and storage reservation time, and all the reservation rules in one reservation policy use the same aggregation processing mode.
Preferably, the performance data collector aggregates the real-time performance data according to the collected frequency, and then sends the aggregated data to the performance processing server.
Preferably, the aggregation processing calculates the real-time data according to a specified aggregation mode, and supports a latest value, a minimum value, a maximum value and an average aggregation mode; the calculated value is used as performance data to be sent to a performance processing server, the collection frequency of a retention strategy is inversely proportional to retention timeliness, and the more dense the collection frequency is, the smaller the storage timeliness is; the measurement of an object can define a plurality of acquisition time periods, and each acquisition time period has a retention rule, so that a historical acquisition time period is needed to judge the retention rule applicable to the current situation; the processing logic is as follows:
1) Querying a retention policy of the object metric;
2) Inquiring and acquiring a current acquisition period by accessing the stored data of the object measurement;
3) And matching a reservation rule applicable to the current acquisition period, and using the acquisition frequency in the reservation rule to aggregate real-time data according to an aggregation processing mode of the reservation strategy to generate performance data.
Preferably, the time sequence database belongs to a storage application, and is different from the traditional relational database, the time sequence database is modeled according to objects, the measurement type can be dynamically expanded, and the time sequence database is stored according to time sequences and measurement values; the application dynamically models and updates data based on the object, the object metric type, and the retention policy.
Preferably, the data is stored and the abnormality detection process is performed. The abnormality detection model uses interval detection, raviness detection, linear detection, and model detection. The data will match the triggering conditions of the different anomaly detection models. If the matching is successful, an abnormality calculation and processing task is generated, an abnormality alarm is generated by the timing task according to the model calculation abnormality score, and the user-defined arrangement processing of the abnormality alarm is supported by the arrangement device.
Compared with the prior art, the application has the advantages and positive effects that,
1. the application uses the performance engine to perform configurable collection and retention strategy on the performance data, uses the time sequence database storage of the dynamic expansion of the object to efficiently support the dynamic anomaly detection. The problems of low efficiency in processing performance data, interface solidification, poor adaptability, poor expansibility and the like in the prior art are solved.
2. The application can set up the dynamic abnormal triggering conditions of interval detection, gully detection, linear detection and model detection for the object of the measurement example accurate to the resource example, and solves the problems that the prior art only supports static modeling of the resource object and cannot develop multidimensional analysis and prediction.
3. The application supports multi-level triggering conditions and arranging flow setting, generates abnormal alarms with different severity and applies different alarm processing flows including fault work order dispatching and alarm obstacle avoidance repair task processing. The method solves the problems that the prior art only supports a single group of static threshold values or intervals, the condition for generating the abnormal alarm is single, and the severity of the abnormal alarm cannot be divided according to different resource objects, measurement types and historical data conditions. Only a single alarm processing flow is supported, abnormal alarms with different severity degrees generated by different measurement types cannot be subjected to flow arrangement, and automatic work order distribution and fault restoration cannot be realized.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are needed in the description of the embodiments will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present application, and other drawings may be obtained according to these drawings without inventive effort to a person skilled in the art.
FIG. 1 is a schematic diagram of a dynamic modeling performance processing engine architecture;
FIG. 2 is a flow chart of anomaly detection;
FIG. 3 is a flow chart of a trigger condition hierarchy and orchestration flow.
Detailed Description
In order that the above objects, features and advantages of the application will be more clearly understood, a further description of the application will be rendered by reference to the appended drawings and examples. It should be noted that, without conflict, the embodiments of the present application and features in the embodiments may be combined with each other.
In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present application, however, the present application may be practiced otherwise than as described herein, and therefore the present application is not limited to the specific embodiments of the disclosure that follow.
Embodiment 1 as shown in fig. 1 and 2, the present application provides a method for building a dynamic anomaly detection model based on performance data, including a performance data collector and a performance processing server,
the data collector collects data at a collection frequency defined in a retention policy used by the metrics of the object;
the performance processing server is responsible for storing the performance data sent by the performance data collector, and storing the acquisition frequency time stamp, the object, the measurement type and the aggregate value which accord with the retention policy into the time sequence database.
Preferably, the reservation policy supports that different reservation rules are used for each acquisition period definition, including data acquisition frequency and storage reservation time, and all the reservation rules in one reservation policy use the same aggregation processing mode.
Preferably, the performance data collector aggregates the real-time performance data according to the collected frequency, and then sends the aggregated data to the performance processing server.
Preferably, the aggregation processing calculates the real-time data according to a specified aggregation mode, and supports a latest value, a minimum value, a maximum value and an average aggregation mode; the calculated value is used as performance data to be sent to a performance processing server, the collection frequency of a retention strategy is inversely proportional to retention timeliness, and the more dense the collection frequency is, the smaller the storage timeliness is; the measurement of an object can define a plurality of acquisition time periods, and each acquisition time period has a retention rule, so that a historical acquisition time period is needed to judge the retention rule applicable to the current situation; the processing logic is as follows:
1) Querying a retention policy of the object metric;
2) Inquiring and acquiring a current acquisition period by accessing the stored data of the object measurement;
3) And matching a reservation rule applicable to the current acquisition period, and using the acquisition frequency in the reservation rule to aggregate real-time data according to an aggregation processing mode of the reservation strategy to generate performance data.
Preferably, the time sequence database belongs to a storage application, and is different from the traditional relational database, the time sequence database is modeled according to objects, the measurement type can be dynamically expanded, and the time sequence database is stored according to time sequences and measurement values; the application dynamically models and updates data based on the object, the object metric type, and the retention policy.
Preferably, the data is stored and the abnormality detection process is performed. The abnormality detection model uses interval detection, raviness detection, linear detection, and model detection. The data will match the triggering conditions of the different anomaly detection models. If the matching is successful, an abnormality calculation and processing task is generated, an abnormality alarm is generated by the timing task according to the model calculation abnormality score, and the user-defined arrangement processing of the abnormality alarm is supported by the arrangement device.
The interval detection can be configured with a plurality of detection intervals, and each interval can have independent interval parameters, interval calculation formulas, tolerance ranges, anomaly scores corresponding to the interval calculation formulas and processing flows of corresponding anomaly alarms. If the value of the performance data falls outside the interval, a corresponding anomaly score is generated, and if the difference between the performance value of the next acquisition frequency and the current performance value accords with the floating range, a new processing task is not generated, and the last anomaly score is directly used. If the performance value of the next acquisition frequency is larger than the floating range and does not belong to the normal interval, the abnormal score is recalculated, and the processing task of abnormal alarm is generated.
The detection of ravines is applicable to anomaly analysis of data loss, ravines refer to time windows of performance data loss. And detecting the gully, and triggering an abnormal alarm when the gully reaches a certain value, namely the duration of the performance data loss window reaches a certain value. Different abnormal score values can be mapped by configuring different sizes of ravines, namely different time window durations, the larger the ravines are, the higher the abnormal score is, the higher the severity of abnormal alarms is, and different abnormal alarm processing flows can be configured and triggered.
The linear detection can be used for carrying out abnormal alarm prediction on performance data in a certain time range in the future. Linear probing requires the use of relevant historical performance data, the range of use of the configuration historical performance data, the range of future effective predictions, thresholds, and confidence levels. By configuring historical performance data in a range, linear estimation, growth or reduction is performed in a future effective prediction range. If the anomaly score is calculated to possibly reach the anomaly alarm threshold value in the future effective prediction range, an anomaly alarm is generated. The linear detection supports a plurality of different threshold configurations, corresponding to abnormal alarms and processing flows of different severity levels respectively.
The model detection mainly uses a machine learning training mechanism, supports a scripted self-defined model algorithm, and acquires a more accurate abnormality detection model by measuring abnormality monitoring for a period of time, training data and comparing results. Model detection requires defining a time window for training, with early warning conditions such as below expected values, above expected values or not within an expected range of values. Model detection supports defining multiple levels of abnormal alarm triggering conditions and process flows by standard deviation numbers from expected values. The greater the standard deviation from the expected value, the higher the severity of the anomaly alert.
The model detects the built-in prediction mode, uses a probability index weighting moving average line algorithm, and performs the partial prediction analysis on the normal time sequence performance data. This model also belongs to machine learning, requires a certain training range, and uses a defined concept index weighted moving average line algorithm to analyze and predict abnormal alarms in the configured data training application effective period. The probability index weighted moving average line algorithm has mainly two parameters, alpha and beta. The alpha value controls the amount of data used for the weighted average calculation, and the greater the alpha value, the more data used for the weighted average calculation, and the smoother the weighted average line. The higher the beta value, the lower the tolerance of the beta value to the difference, the lower the tolerance, and the sensitivity to the difference from the expected value is improved.
When a scripted custom model is introduced, the content defined by the model is consistent with the parameters and modes of a prediction mode, normal performance data is required to be used for training to obtain a normal performance data model, whether new performance data meets a trigger condition or not is judged in the validity period of the model, namely, whether the data is different from the expected value of the model or not is judged, and the severity is defined according to standard deviation.
The dynamic model supports configuration of multi-level abnormal triggering conditions, mapping of abnormal alarms with different severity, subsequent alarm processing through an arrangement tool, obstacle removal repair and other operations.
Finally, the following needs to be added:
1. and compressing and storing the performance data according to a retention policy by adopting a performance acquisition server and a time sequence database. The method ensures that the high-efficiency performance data acquisition and storage support dynamic modeling, and is a key point of the proposal, and protection needs to be given.
2. The dynamic model for abnormality detection is established, and comprises interval detection, gully detection, linear detection and model detection, the abnormal triggering conditions of multiple levels and the arrangement flows corresponding to the triggering conditions are supported, different alarm processing and obstacle removing restoration flows are defined, and the method is a key point of the proposal and needs to be protected.
3. The model detection supports scripted custom model expansion, a machine learning model algorithm is introduced, a normal performance data filter, training time, effective aging and the like are set, an expected value of the model is dynamically generated, a standard deviation number is used for defining a multi-level triggering condition, the dynamic model of abnormal detection is intelligently enriched, and the model is a key point of the proposal and needs to be protected.
The present application is not limited to the above-mentioned embodiments, and any equivalent embodiments which can be changed or modified by the technical content disclosed above can be applied to other fields, but any simple modification, equivalent changes and modification made to the above-mentioned embodiments according to the technical substance of the present application without departing from the technical content of the present application still belong to the protection scope of the technical solution of the present application.
Claims (1)
1. A method for establishing a dynamic abnormality detection model based on performance data comprises a performance data collector and a performance processing server, and is characterized in that,
the data collector collects data at a collection frequency defined in a retention policy used by the metrics of the object;
the performance processing server is responsible for storing the performance data sent by the performance data collector, and storing the acquisition frequency time stamp, the object, the measurement type and the aggregate value which accord with the retention strategy into a time sequence database;
different reservation rules are used for supporting the definition of each acquisition period in the reservation strategy, including data acquisition frequency and storage reservation timeliness, and all the reservation rules in one reservation strategy use the same aggregation processing mode;
the performance data collector aggregates the real-time performance data according to the collected frequency, and then sends the aggregated data to the performance processing server;
the aggregation processing calculates the real-time data according to a specified aggregation mode, and supports a latest value, a minimum value, a maximum value and an average aggregation mode; the calculated value is sent to a performance processing server as performance data, the retention policy acquisition frequency is inversely proportional to retention time, the more densely the acquisition frequency,
the storage time can be reduced; the measurement of an object can define a plurality of acquisition time periods, and each acquisition time period has a retention rule, so that a historical acquisition time period is needed to judge the retention rule applicable to the current situation; the processing logic is as follows:
1) Querying a retention policy of the object metric;
2) Inquiring and acquiring a current acquisition period by accessing the stored data of the object measurement;
3) Matching a reservation rule applicable to the current acquisition period, and using the acquisition frequency in the reservation rule to aggregate real-time data according to an aggregation processing mode of a reservation strategy to generate performance data;
the time sequence database belongs to storage application, is different from the traditional relational database, is modeled according to objects, can dynamically expand measurement types, and is stored according to time sequences and measurement values; the application dynamically models and updates data based on the object, the object metric type, and the retention policy.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010198615.5A CN111400284B (en) | 2020-03-20 | 2020-03-20 | Method for establishing dynamic anomaly detection model based on performance data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010198615.5A CN111400284B (en) | 2020-03-20 | 2020-03-20 | Method for establishing dynamic anomaly detection model based on performance data |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111400284A CN111400284A (en) | 2020-07-10 |
| CN111400284B true CN111400284B (en) | 2023-09-12 |
Family
ID=71434444
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010198615.5A Active CN111400284B (en) | 2020-03-20 | 2020-03-20 | Method for establishing dynamic anomaly detection model based on performance data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111400284B (en) |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2012047757A1 (en) * | 2010-10-04 | 2012-04-12 | Avocent | System and method for monitoring and managing data center resources in real time incorporating manageability subsystem |
| CN103246695A (en) * | 2013-03-15 | 2013-08-14 | 山西省电力公司大同供电分公司 | Integrating method of HighSoon real-time database and IES600P system |
| CN105868073A (en) * | 2016-03-25 | 2016-08-17 | 山东超越数控电子有限公司 | Data center energy saving strategy implementation method based on time series analysis |
| CN106612199A (en) * | 2015-10-26 | 2017-05-03 | 华耀(中国)科技有限公司 | Network monitoring data collection and analysis system and method |
| KR20180076172A (en) * | 2016-12-27 | 2018-07-05 | 주식회사 씨에스리 | Apparatus and method for detecting anomaly of database system |
| CN109343995A (en) * | 2018-10-25 | 2019-02-15 | 金税信息技术服务股份有限公司 | Intelligent O&M analysis system based on multi-source heterogeneous data fusion, machine learning and customer service robot |
| CN109656793A (en) * | 2018-11-22 | 2019-04-19 | 安徽继远软件有限公司 | A kind of information system performance stereoscopic monitoring method based on multi-source heterogeneous data fusion |
| CN109871401A (en) * | 2018-12-26 | 2019-06-11 | 北京奇安信科技有限公司 | A method and device for detecting abnormality in time series |
| CN110136795A (en) * | 2019-05-20 | 2019-08-16 | 合肥工业大学 | Construction method of time sequence database for cognitive early warning |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8554699B2 (en) * | 2009-10-20 | 2013-10-08 | Google Inc. | Method and system for detecting anomalies in time series data |
| US20110153603A1 (en) * | 2009-12-17 | 2011-06-23 | Yahoo! Inc. | Time series storage for large-scale monitoring system |
| US10410135B2 (en) * | 2015-05-21 | 2019-09-10 | Software Ag Usa, Inc. | Systems and/or methods for dynamic anomaly detection in machine sensor data |
| US10917419B2 (en) * | 2017-05-05 | 2021-02-09 | Servicenow, Inc. | Systems and methods for anomaly detection |
| US11461290B2 (en) * | 2018-04-12 | 2022-10-04 | Vmware, Inc. | System and method for run-time adaptable policy engine for heterogeneous managed entities |
| US11042314B2 (en) * | 2018-07-31 | 2021-06-22 | The Silk Technologies Ilc Ltd | Generation, validation and implementation of storage-orchestration strategies using virtual private array (VPA) in a dynamic manner |
-
2020
- 2020-03-20 CN CN202010198615.5A patent/CN111400284B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2012047757A1 (en) * | 2010-10-04 | 2012-04-12 | Avocent | System and method for monitoring and managing data center resources in real time incorporating manageability subsystem |
| CN103246695A (en) * | 2013-03-15 | 2013-08-14 | 山西省电力公司大同供电分公司 | Integrating method of HighSoon real-time database and IES600P system |
| CN106612199A (en) * | 2015-10-26 | 2017-05-03 | 华耀(中国)科技有限公司 | Network monitoring data collection and analysis system and method |
| CN105868073A (en) * | 2016-03-25 | 2016-08-17 | 山东超越数控电子有限公司 | Data center energy saving strategy implementation method based on time series analysis |
| KR20180076172A (en) * | 2016-12-27 | 2018-07-05 | 주식회사 씨에스리 | Apparatus and method for detecting anomaly of database system |
| CN109343995A (en) * | 2018-10-25 | 2019-02-15 | 金税信息技术服务股份有限公司 | Intelligent O&M analysis system based on multi-source heterogeneous data fusion, machine learning and customer service robot |
| CN109656793A (en) * | 2018-11-22 | 2019-04-19 | 安徽继远软件有限公司 | A kind of information system performance stereoscopic monitoring method based on multi-source heterogeneous data fusion |
| CN109871401A (en) * | 2018-12-26 | 2019-06-11 | 北京奇安信科技有限公司 | A method and device for detecting abnormality in time series |
| CN110136795A (en) * | 2019-05-20 | 2019-08-16 | 合肥工业大学 | Construction method of time sequence database for cognitive early warning |
Non-Patent Citations (1)
| Title |
|---|
| 柴亚刚 ; .基于时序数据库的分布式网络波动监控系统.中国传媒科技.2018,(03),全文. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111400284A (en) | 2020-07-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN118214168B (en) | A real-time status monitoring method for power distribution circuit | |
| CN116976707B (en) | User electricity consumption data anomaly analysis method and system based on electricity consumption data acquisition | |
| CN108446864B (en) | Big data analysis-based fault early warning system and method for rail transit equipment | |
| CN118427761B (en) | Intelligent building energy consumption anomaly analysis system and method | |
| CN117932501B (en) | Electric energy meter running state management method and system | |
| CN108491861A (en) | Power transmission and transformation equipment state abnormal patterns recognition methods based on multi-source multi-parameter fusion and device | |
| CN114978956B (en) | Method and device for detecting abnormal mutation points of performance of intelligent city network equipment | |
| CN111241208A (en) | A kind of abnormal monitoring method and device for periodic time series data | |
| CN117560300B (en) | Intelligent internet of things flow prediction and optimization system | |
| CN115225986B (en) | OSU bandwidth self-adaptive adjustment method and device | |
| CN118642922B (en) | A computer operation status monitoring and management system | |
| CN115129011B (en) | Industrial resource management method based on edge calculation | |
| CN118965171A (en) | Online intelligent monitoring method for anti-electricity theft based on abnormal electricity consumption data | |
| CN119492945B (en) | A bus duct trend analysis early warning method, system and storage medium | |
| CN120336129B (en) | Supervision data quality monitoring system and method based on artificial intelligence | |
| CN118245945A (en) | 5G communication-based power distribution fault early warning system and method | |
| CN120185193A (en) | Intelligent power distribution cabinet state dynamic monitoring method and system | |
| CN118487917A (en) | Internet of things alarm system based on autoregressive comprehensive moving average model | |
| CN111400284B (en) | Method for establishing dynamic anomaly detection model based on performance data | |
| CN118014386B (en) | Data trend analysis system, method and medium for regional electricity users | |
| CN118611273A (en) | Adaptive topology electric energy meter remote monitoring method and platform | |
| CN118473967A (en) | Home wide network quality degradation identification method | |
| CN111125195A (en) | Data anomaly detection method and device | |
| CN116720983A (en) | Power supply equipment abnormality detection method and system based on big data analysis | |
| CN113822587B (en) | Factory capacity evaluation method based on bus current data |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |