CN111552843B - Fault prediction method based on weighted causal dependency graph - Google Patents
Fault prediction method based on weighted causal dependency graph Download PDFInfo
- Publication number
- CN111552843B CN111552843B CN202010327713.4A CN202010327713A CN111552843B CN 111552843 B CN111552843 B CN 111552843B CN 202010327713 A CN202010327713 A CN 202010327713A CN 111552843 B CN111552843 B CN 111552843B
- Authority
- CN
- China
- Prior art keywords
- event
- vertex
- causal dependency
- probability
- wcdgs
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/901—Indexing; Data structures therefor; Storage structures
- G06F16/9024—Graphs; Linked lists
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/903—Querying
Landscapes
- Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a fault prediction method based on a weighted causal dependency graph aiming at causal dependency among events in association rule mining. An abstract weighted causal dependency graph is used to represent event rules during the fault prediction phase. The method comprises the following steps: s1, constructing Weighted Causal Dependency Graphs (WCDGs) according to event rules; s2, continuously updating the weighted causal dependency graph before according to the event rule; and S3, predicting the probability of the occurrence of a subsequent event (fault event) when one event occurs based on the updated weighted causal dependency graph WCDGs. Compared with other prediction models, the method can store and update the event rules more easily, can realize efficient update of the rules in a self-starting mode in the whole life cycle of the system, meets the requirements of system dynamics, and greatly reduces time overhead while improving prediction accuracy.
Description
Technical Field
The invention belongs to the technical field of information systems, and particularly relates to a fault prediction method based on a Weighted Causal Dependency Graph (WCDGs).
Background
In recent years, as the network architecture becomes more complex and the scale is continuously enlarged, the reliability of the system is seriously affected, and the requirement on the accuracy of the fault prediction method is increasingly increased. At present, some fault prediction methods for network equipment and large-scale systems at home and abroad can be largely divided into three categories: a prediction method based on fault tracking, a prediction method based on state monitoring, and a prediction method based on event driving.
The fault prediction method based on event driving mostly adopts the methods of association rule mining and rule reasoning to predict the possible association fault of the system. However, the conventional association rule mining algorithm does not generally consider the occurrence sequence between events, cannot express causal dependency relationships between the events, and can seriously affect the accuracy and reliability of the fault prediction algorithm.
In order to solve the development state of the prior art, the prior patents and the prior documents are searched, compared and analyzed, and the following technical information with high relevance to the invention is screened out:
patent scheme 1:201310072577.9 fault prediction method and system
The invention provides a fault prediction method and a fault prediction system based on a kernel principal component analysis method. The invention adopts a KPCA-based fault reconstruction method aiming at fault prediction, can well solve the nonlinear problem of process data, excavate fault direction and estimate fault amplitude from data with hidden faults, simultaneously considers the multidimensional characteristics of the faults, and can obtain more accurate fault prediction results.
Patent scheme 2:201811229565.1 equipment fault prediction method based on gradient lifting decision tree
The invention provides an equipment fault prediction method based on a gradient lifting decision tree, which is used for ensuring the accuracy and reliability of a prediction result. The method comprises the following steps: constructing an equipment fault characteristic vector, and constructing a normalized equipment characteristic vector and a coded equipment characteristic vector through the equipment fault characteristic vector; training a decision tree classification model through fault large-class gradient boosting to obtain a final fault large-class prediction model; training a decision tree classification model through fault subclass gradient lifting to obtain a final fault subclass prediction model; and (4) constructing input characteristics of the equipment life prediction model to obtain a final equipment life prediction model.
The defects of the above patent scheme 1: according to the scheme, a KPCA fault reconstruction method is adopted for fault prediction, and the multidimensional characteristic of the fault is considered, so that the nonlinear problem of data can be solved well, and a more accurate prediction result can be obtained. However, the scheme mainly aims at fault prediction with rotating machinery, the expansibility is not strong, and a kernel principal component analysis method adopted by the scheme is simpler for a slightly complex fault prediction scene.
The defects of the above patent scheme 2: according to the scheme, a gradient lifting decision tree model is adopted, the equipment fault feature vector is constructed, and the decision model is continuously improved by utilizing a combination mode of gradient lifting and regression decision tree, so that the accuracy and reliability of the prediction model can be better ensured. However, in the patent scheme, the association rule is not mined in consideration of the interdependency and the association among the events, so that the extracted knowledge is limited, and the prediction accuracy is limited.
And the fault prediction aims to predict whether the system will be in fault in a future period of time based on the history of the system and the current state information in the observation time. The main idea of the event-driven fault prediction method is to realize fault prediction by associated rule mining and rule reasoning of log events. Such methods generally include the following three steps: 1) Preprocessing a log: identification, screening and filtering of representative events are performed. 2) And (3) association rule mining: and mining a frequent event sequence occurring together with the fault events to generate a fault prediction rule. 3) And (3) online fault prediction: the purpose of prediction is achieved through reasonable rule reasoning by carrying out pattern matching on the event monitored in real time and the rule in the knowledge base.
However, the conventional association rule mining algorithm usually does not consider the occurrence sequence of events, so the mined association rule only represents the event correlation determined by statistical association, and cannot express the causal dependency relationship between events (namely, the event a causes the event b to occur), so that the causal dependency relationship cannot be used as the basis for fault prediction. In addition, the causal graph and the production rule are two different knowledge representation methods, and the latter has the defects of non-intuition, lack of flexibility, low reasoning efficiency and the like.
Disclosure of Invention
The technical problem to be solved by the invention is to provide a fault prediction method based on a weighted causal dependency graph aiming at the problems in the background art, the method designs a new Weighted Causal Dependency Graph (WCDGs) to represent event rules, the WCDGs are continuously and automatically updated in the whole life cycle of a cluster system, and the fault prediction is realized by utilizing the forward reasoning of the causal graph.
The technical scheme adopted by the invention is as follows:
a fault prediction method based on a weighted causal dependency graph comprises the following steps:
(1) Building a group of Weighted Causal Dependency Graphs (WCDGs) according to an event rule, wherein each Weighted Causal Dependency Graph (WCDGs) represents the relevant condition of an event, creating a WCDGs index, and storing the position of the event in the WCDGs;
(2) Continuously updating the Weighted Causal Dependency Graphs (WCDGs) created in the step (1) according to event rules;
(3) Predicting the probability of occurrence of a subsequent fault event when an event occurs based on the updated weighted causal dependency graph WCDGs; the method comprises the steps of deducing a fatal event which possibly occurs in a prediction time window and the occurrence probability of the fatal event according to an event observed in the current observation time window, sending an alarm once the occurrence probability exceeds a predefined prediction probability threshold, and giving prediction detailed information.
Wherein, the step (1) comprises the following steps:
s11, constructing a group of Weighted Causal Dependency Graphs (WCDGs) according to event rules, wherein each weighted causal dependency graph represents the relevant condition of an event; wherein the weighted causal dependency graph is a directed dependency graph (V, G), each vertex in the graph represents an event, its child vertices are the posterior events of it in the event rule, each vertex is weighted by an event weight; the edges in the weighted causal dependency graph represent the causal dependency of two events connected by the edges, with the pointer direction in chronological order; each edge has four attributes of a head vertex, a tail vertex, a support count and a confidence coefficient, wherein the head vertex and the tail vertex respectively represent two events in the event rule, the support count of the edge represents the support degree of the event rule, and the confidence coefficient of the edge represents the confidence coefficient of the event rule, namely the association strength between the two events;
s12, comparing the support counts of the sides with circulation, and deleting the sides with low support;
s13, creating a WCDGs index [ WCDG entry vertex, WCDG _ ID ], and saving the position of an event in the WCDGs.
Wherein, the step (3) comprises the following steps:
s31, defining a prediction probability threshold value p th Observation time window Δ t d And a prediction time window Δ t p ；
S32, when an event occurs, searching an index of the event in the updated weighted causal dependency graph WCDG, searching a matched weighted causal dependency graph, and finding out an inlet vertex of the WCDG as a head vertex;
s33, calculating the probability of the head vertex and all the connected sub-vertices, and selecting the maximum probability as the probability of the head vertex; the probability calculation formula of the head vertex is as follows:
wherein, power (e) i ) Representing an event e i The calculation formula of (c) is:
power(e i )＝w(e i )×lgd
wherein, w (e) i ) Representing an event e i Weight of (c), corr (e) i →e j ) Representing an event e i And event e j The strength of the causal relationship between the two elements is calculated by the following formula:
corr(e i →e j )＝δ×confidence(e i →e j )
wherein, confidence (e) i →e j ) Representing an event e i And event e j The confidence of the edge between, δ representing the time attenuation factor;
s34, when the probability of the vertex at the head is higher than the prediction probability threshold value p th If so, marking the sub-vertex with the maximum probability corresponding to the head vertex, and judging as a predicted event;
s35, taking the sub-vertex marked in the S34 as a head vertex, and returning to the step S33 until the probability that no event occurs is greater than p th And outputting all predicted events, including the type, level and occurrence position of the events.
Compared with the prior art, the invention has the advantages that:
compared with other prediction models, the method can store and update the event rules more easily, can realize efficient update of the rules in a self-starting mode in the whole life cycle of the system, meets the requirements of system dynamics, and greatly reduces time overhead while improving prediction accuracy.
Drawings
FIG. 1 is a flow chart of a method for predicting a fault based on a weighted causal dependency graph according to the present invention;
FIG. 2 is a diagram illustrating a WCDGs of a log causal dependency graph based on a weighted causal dependency graph failure prediction method according to the present invention.
Detailed Description
The present invention will be further described with reference to fig. 1 to 2.
A failure prediction method based on a weighted causal dependency graph is described as follows:
s1, constructing Weighted Causal Dependency Graphs (WCDGs) according to event rules. The weighted causal dependency graph is a directed dependency graph (V, G), wherein V represents a vertex in the directed graph and represents an event; g represents an edge in the directed graph, representing the causal relevance of two events connected by the edge;
one vertex in the weighted causal dependency graph represents an event whose child vertices are a posteriori events in the event rules. The vertices are weighted by event weights. The edges in the weighted causal dependency graph represent the causal dependency of two events connected by an edge, with the pointer direction in chronological order. Each edge has four attributes, head vertex, tail vertex, support count, and confidence. The head vertex and the tail vertex respectively represent two events in the event rule, the support count of the edge represents the support degree of the event rule, and the confidence degree of the edge represents the confidence degree of the event rule, namely the association strength between the two events. In fig. 2, circles are made to represent events, numeral references represent transaction numbers, and lines between events represent the order of occurrence of events. The method specifically comprises the following steps:
s11, constructing a group of WCDGs according to the generated event rule, wherein each WCDGs represents the relevant condition of an event tag.
S12, deleting the WCDGs. WCDGs generated directly from event rules may have some loops in the graph. For example, { A → B } and { B → A } may occur simultaneously, but this violates causal theory. To prune the edges, we compare their support counts. Lower support counts indicate that there is a weak statistical dependency between events, and their co-occurrence may be only coincidental. Therefore, we delete the less supported edges from the graph.
And S13, creating WCDGs indexes and saving the positions of the events in the WCDGs. [ WCDG _ ID, WCDG entry vertex ] is the index of the WCDG. The event location can be quickly retrieved by indexing.
And S2, continuously updating the weighted causal dependency graph before according to the event rule.
And S3, predicting the probability of the occurrence of a subsequent event (fault event) when one event occurs based on the updated weighted causal dependency graph WCDGs. I.e. according to the current observation time window deltat d Internally observed event inference predicted time window Δ t p In case of a possible occurrence of a fatal event and the probability of its occurrence, once a predefined prediction probability threshold p is exceeded th An alert is issued and details of the prediction are given (including the type, level and location of occurrence of the event).
The method specifically comprises the following steps:
s31, defining a prediction probability threshold value p th Observation time window Δ t d And a prediction time window Δ t p ；
S32, when the current observation time window delta t d When certain event is observed to happen, a prediction time window delta t is deduced p Searching the indexes of the events in the updated Weighted Causal Dependency Graph (WCDGs), searching a matched weighted causal dependency graph, and finding out an entrance vertex of the WCDG as a head vertex;
s33, calculating the probability of the head vertex and all the connected sub-vertices, and selecting the maximum probability as the probability of the head vertex; the probability calculation formula of the head vertex is as follows:
wherein, power (e) i ) Representing an event e i The calculation formula of (c) is:
power(e i )＝w(e i )×lgd
wherein, w (e) i ) Representing an event e i Weight of (c), corr (e) i →e j ) Representing an event e i And event e j The degree of causality is calculated according to the following formula:
corr(e i →e j )＝δ×confidence(e i →e j )
wherein, confidence (e) i →e j ) Representing an event e i And event e j The confidence of the edges between, delta, represents the time-decaying factor;
s34, when the probability of the vertex at the head is higher than the prediction probability threshold value p th If so, marking the sub-vertex with the maximum probability corresponding to the head vertex, and judging as a predicted event;
s35, taking the sub-vertex marked in the S34 as a head vertex, and returning to the step S33 until the probability that no event occurs is greater than p th And outputting all predicted events, including the type, level and occurrence position of the events.
Claims (1)
1. A fault prediction method based on a weighted causal dependency graph is characterized by comprising the following steps:
(1) Building a group of Weighted Causal Dependency Graphs (WCDGs) according to an event rule, wherein each Weighted Causal Dependency Graph (WCDGs) represents the relevant condition of an event, creating a WCDGs index, and storing the position of the event in the WCDGs;
(2) Continuously updating the Weighted Causal Dependency Graphs (WCDGs) created in the step (1) according to event rules;
(3) Predicting the probability of occurrence of a subsequent fault event when one event occurs based on the updated Weighted Causal Dependency Graphs (WCDGs); deducing a fatal event which possibly occurs in a prediction time window and the occurrence probability thereof according to an event observed in the current observation time window, sending an alarm once the occurrence probability exceeds a predefined prediction probability threshold, and giving detailed prediction information;
wherein, the step (1) comprises the following steps:
s11, constructing a group of Weighted Causal Dependency Graphs (WCDGs) according to event rules, wherein each weighted causal dependency graph represents the relevant condition of an event; wherein the weighted causal dependency graph is a directed dependency graph (V, G), each vertex in the graph represents an event, its child vertices are the posterior events of it in the event rule, each vertex is weighted by an event weight; the edges in the weighted causal dependency graph represent the causal dependency of two events connected by the edges, with the pointer direction in chronological order; each edge has four attributes of a head vertex, a tail vertex, a support count and a confidence coefficient, wherein the head vertex and the tail vertex respectively represent two events in the event rule, the support count of the edge represents the support degree of the event rule, and the confidence coefficient of the edge represents the confidence coefficient of the event rule, namely the association strength between the two events;
s12, comparing the support counts of the sides with circulation, and deleting the sides with low support;
s13, establishing a WCDGs index [ a WCDG entrance vertex and a WCDG _ ID ], and storing the position of an event in the WCDGs;
wherein, the step (3) comprises the following steps:
s31, defining a prediction probability threshold value p th Observation time window Δ t d And a prediction time window Δ t p ；
S32, when the current observation time window delta t d When an event is observed to occur, searching an index of the event in the updated Weighted Causal Dependency Graph (WCDG), searching a matched weighted causal dependency graph, and finding an inlet vertex of the WCDG as a head vertex;
s33, calculating the probability of the head vertex and all the connected sub-vertices, and selecting the maximum probability as the probability of the head vertex; the probability calculation formula of the head vertex is as follows:
wherein, power (e) i ) Representing an event e i The calculation formula of the influence strength of (c) is as follows:
power(e i )＝w(e i )×lgd
wherein, w (e) i ) Representing an event e i Weight of (c), corr (e) i →e j ) Representing an event e i And event e j The degree of causality is calculated according to the following formula:
corr(e i →e j )＝δ×confidence(e i →e j )
wherein, confidence (e) i →e j ) Representing an event e i And event e j The confidence of the edge between, δ representing the time attenuation factor;
s34, when the probability of the vertex at the head is higher than the prediction probability threshold value p th If so, marking the sub-vertex corresponding to the head vertex and having the maximum probability, and judging as the prediction time window delta t p Predicted events within;
s35, taking the sub-vertex marked in the S34 as a head vertex, and returning to the step S33 until the probability that no event occurs is greater than p th And outputting all predicted events, including the type, level and occurrence position of the events.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010327713.4A CN111552843B (en) | 2020-04-23 | 2020-04-23 | Fault prediction method based on weighted causal dependency graph |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010327713.4A CN111552843B (en) | 2020-04-23 | 2020-04-23 | Fault prediction method based on weighted causal dependency graph |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111552843A CN111552843A (en) | 2020-08-18 |
| CN111552843B true CN111552843B (en) | 2023-03-31 |
Family
ID=71998502
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010327713.4A Active CN111552843B (en) | 2020-04-23 | 2020-04-23 | Fault prediction method based on weighted causal dependency graph |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111552843B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113822430B (en) * | 2020-12-28 | 2024-05-21 | 京东科技控股股份有限公司 | Event reasoning method, device, computer equipment and storage medium |
| CN112365344B (en) * | 2021-01-11 | 2021-04-27 | 支付宝(杭州)信息技术有限公司 | Method and system for automatically generating business rules |
| CN113420917B (en) * | 2021-06-18 | 2023-10-27 | 广东工业大学 | Method for predicting future faults of service system, computer equipment and storage medium |
| CN115146271B (en) * | 2022-09-02 | 2022-11-25 | 浙江工业大学 | APT (advanced persistent threat) source tracing and researching method based on causal analysis |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101196743A (en) * | 2007-12-26 | 2008-06-11 | 西安交通大学 | Dynamoelectric system safety analyzing device and method based on cause-effect network model |
| WO2015176565A1 (en) * | 2014-05-22 | 2015-11-26 | 袁志贤 | Method for predicting faults in electrical equipment based on multi-dimension time series |
| CN105117771A (en) * | 2015-07-28 | 2015-12-02 | 北京理工大学 | Agricultural machinery fault identification method based on association rule directed acyclic graph |
| CN108304317A (en) * | 2017-12-28 | 2018-07-20 | 东南大学 | A kind of single process procedures static slicing method and system executing frequency based on path |
| CN108320040A (en) * | 2017-01-17 | 2018-07-24 | 国网重庆市电力公司 | Acquisition terminal failure prediction method and system based on Bayesian network optimization algorithm |
| CN108762238A (en) * | 2018-06-15 | 2018-11-06 | 东北大学 | A kind of wet metallurgical leaching process method for diagnosing faults based on DCD |
| CN109886313A (en) * | 2019-01-28 | 2019-06-14 | 东北大学 | A kind of Dynamic Graph clustering method based on density peak |
| CN110795327A (en) * | 2018-08-01 | 2020-02-14 | 中国矿业大学 | Weighted dependency graph-based concurrent program execution trajectory static simplification method |
| WO2020071054A1 (en) * | 2018-10-02 | 2020-04-09 | 株式会社日立製作所 | Fault factor priority presentation device |
-
2020
- 2020-04-23 CN CN202010327713.4A patent/CN111552843B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101196743A (en) * | 2007-12-26 | 2008-06-11 | 西安交通大学 | Dynamoelectric system safety analyzing device and method based on cause-effect network model |
| WO2015176565A1 (en) * | 2014-05-22 | 2015-11-26 | 袁志贤 | Method for predicting faults in electrical equipment based on multi-dimension time series |
| CN105117771A (en) * | 2015-07-28 | 2015-12-02 | 北京理工大学 | Agricultural machinery fault identification method based on association rule directed acyclic graph |
| CN108320040A (en) * | 2017-01-17 | 2018-07-24 | 国网重庆市电力公司 | Acquisition terminal failure prediction method and system based on Bayesian network optimization algorithm |
| CN108304317A (en) * | 2017-12-28 | 2018-07-20 | 东南大学 | A kind of single process procedures static slicing method and system executing frequency based on path |
| CN108762238A (en) * | 2018-06-15 | 2018-11-06 | 东北大学 | A kind of wet metallurgical leaching process method for diagnosing faults based on DCD |
| CN110795327A (en) * | 2018-08-01 | 2020-02-14 | 中国矿业大学 | Weighted dependency graph-based concurrent program execution trajectory static simplification method |
| WO2020071054A1 (en) * | 2018-10-02 | 2020-04-09 | 株式会社日立製作所 | Fault factor priority presentation device |
| CN109886313A (en) * | 2019-01-28 | 2019-06-14 | 东北大学 | A kind of Dynamic Graph clustering method based on density peak |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111552843A (en) | 2020-08-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111552843B (en) | Fault prediction method based on weighted causal dependency graph | |
| Kuncheva | Classifier ensembles for detecting concept change in streaming data: Overview and perspectives | |
| US9633198B2 (en) | Detecting anomalous process behavior | |
| CN113887616B (en) | Real-time abnormality detection method for EPG connection number | |
| CN113282461B (en) | Alarm identification method and device for transmission network | |
| CN110084326B (en) | Industrial equipment anomaly detection method based on fuzzy set | |
| CN112528519A (en) | Method, system, readable medium and electronic device for engine quality early warning service | |
| CN106600115A (en) | Intelligent operation and maintenance analysis method for enterprise information system | |
| Kadwe et al. | A review on concept drift | |
| Kao et al. | Quality prediction modeling for multistage manufacturing based on classification and association rule mining | |
| CN110210169B (en) | LSTM-based shield tunneling machine fault prediction method | |
| CN111538311B (en) | Flexible multi-state self-adaptive early warning method and device for mechanical equipment based on data mining | |
| CN110826237B (en) | Wind power equipment reliability analysis method and device based on Bayesian belief network | |
| CN114595623A (en) | XGboost algorithm-based unit equipment reference value prediction method and system | |
| CN113064873B (en) | Log anomaly detection method with high recall rate | |
| CN115508672A (en) | Power grid main equipment fault tracing reasoning method, system, equipment and medium | |
| CN115791174B (en) | Rolling bearing abnormality diagnosis method, system, electronic equipment and storage medium | |
| CN112380274A (en) | Control process-oriented anomaly detection system | |
| CN114444582A (en) | Mechanical equipment fault diagnosis method based on convolutional neural network and Bayesian network | |
| CN116457802A (en) | Automatic real-time detection, prediction and prevention of rare faults in industrial systems using unlabeled sensor data | |
| CN110580213A (en) | Database anomaly detection method based on cyclic marking time point process | |
| CN113268370B (en) | Root cause alarm analysis method, system, equipment and storage medium | |
| CN110781206A (en) | Method for predicting whether electric energy meter in operation fails or not by learning meter-dismantling and returning failure characteristic rule | |
| Xu et al. | TLS-WGAN-GP: A generative adversarial network model for data-driven fault root cause location | |
| CN116304835B (en) | AI-based dynamic equipment working condition monitoring management method, system, equipment and medium |
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 |