CN117118811A - Alarm analysis method for industrial alarm flooding - Google Patents
Alarm analysis method for industrial alarm flooding Download PDFInfo
- Publication number
- CN117118811A CN117118811A CN202311386097.XA CN202311386097A CN117118811A CN 117118811 A CN117118811 A CN 117118811A CN 202311386097 A CN202311386097 A CN 202311386097A CN 117118811 A CN117118811 A CN 117118811A
- Authority
- CN
- China
- Prior art keywords
- alarm
- node
- variable
- flooding
- network
- 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.)
- Pending
Links
- 238000004458 analytical method Methods 0.000 title claims abstract description 16
- 238000000034 method Methods 0.000 claims abstract description 72
- 238000012549 training Methods 0.000 claims abstract description 13
- 230000005540 biological transmission Effects 0.000 claims abstract description 6
- 230000000694 effects Effects 0.000 claims abstract description 6
- 238000012546 transfer Methods 0.000 claims description 26
- 238000005070 sampling Methods 0.000 claims description 10
- 238000004422 calculation algorithm Methods 0.000 claims description 5
- 238000012545 processing Methods 0.000 claims description 5
- 238000002372 labelling Methods 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 claims description 3
- 238000011160 research Methods 0.000 abstract description 4
- 238000012544 monitoring process Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 238000004519 manufacturing process Methods 0.000 description 4
- 238000004088 simulation Methods 0.000 description 3
- 230000002159 abnormal effect Effects 0.000 description 2
- 230000001364 causal effect Effects 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000009776 industrial production Methods 0.000 description 2
- 230000005856 abnormality Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 239000000498 cooling water Substances 0.000 description 1
- 239000008358 core component Substances 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 239000007788 liquid Substances 0.000 description 1
- 230000004807 localization Effects 0.000 description 1
- 230000002265 prevention Effects 0.000 description 1
- 238000003892 spreading Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/06—Management of faults, events, alarms or notifications
- H04L41/0631—Management of faults, events, alarms or notifications using root cause analysis; using analysis of correlation between notifications, alarms or events based on decision criteria, e.g. hierarchy, tree or time analysis
- H04L41/065—Management of faults, events, alarms or notifications using root cause analysis; using analysis of correlation between notifications, alarms or events based on decision criteria, e.g. hierarchy, tree or time analysis involving logical or physical relationship, e.g. grouping and hierarchies
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/06—Management of faults, events, alarms or notifications
- H04L41/069—Management of faults, events, alarms or notifications using logs of notifications; Post-processing of notifications
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The invention belongs to the technical field of industrial Internet, and discloses an alarm analysis method for industrial alarm flooding, which combines knowledge driving and data driving to construct an alarm propagation network, so as to realize the research on the root cause and indirect relation of the alarm propagation network and the alarm; the method comprises the steps of initializing the transmission relation of alarm data by utilizing data driving, training an initialized network structure by combining knowledge driving to obtain an alarm propagation network, and finally realizing alarm tracing by utilizing accurate reasoning, so as to help operators to realize alarm root positioning and know alarm variables possibly affected next, and prevent the occurrence of alarm domino effect, thereby effectively solving the problem of alarm flooding.
Description
Technical Field
The invention belongs to the technical field of industrial Internet, and particularly relates to an alarm analysis method for industrial alarm flooding.
Background
In recent years, the wide application of distributed control systems and data acquisition and monitoring systems in large industrial facilities has led to a gradual increase in the level of automation of production and management. However, since large industrial facilities contain a large amount of production equipment and complex production processes, once an abnormal condition occurs in one equipment, other related equipment is easily caused to be abnormal, and serious safety problems are caused for industrial production. The alarm monitoring system is used as one of the core components of modern large-scale complex industry, and plays a vital role in whether the industrial production can safely and stably run. Many problems in alarm monitoring, such as false alarm, associated alarm, missed alarm and the like, are well solved under the research of domestic and foreign scholars, but the current alarm flooding is still a difficult problem which puzzles operators.
Alarm flooding refers to a scenario where a large number of alarm signals occur resulting in an alarm rate that exceeds the processing capacity of the operator. The frequent occurrence and spreading of alarm flooding can lead to the loss of effectiveness of an alarm system, and can not effectively provide effective alarm monitoring information for operators, thereby being extremely easy to cause production accidents. Therefore, how to help operators cope with alarm flooding is a urgent problem to be solved.
In view of this problem, there are also not only solutions for solving different directions in the research and exploration of the prior art; as in patent application CN110209144a, it discloses a two-layer real-time monitoring and alarm tracing method based on dynamic and static cooperative difference analysis, which can monitor alarm flooding in real time and realize tracing; it does not take into account the ring network where variable transfer relationships may occur and only locates the device when ultimately locating the root of the fault, and it is not clear what the cause of the abnormality of the device is. Patent application CN113722140a discloses a method for diagnosing the root cause of industrial alarm flooding based on small sample learning and a storage medium, which only judges the cause of the fault, and does not consider the propagation process of the fault; meanwhile, the fault data is not further distinguished according to the high and low threshold values, so that the alarm can not be further determined, other devices can not be influenced, and the reason of the alarm can not be further determined.
In summary, the existing method can only solve the superficial fault location, but cannot take its root and cannot effectively generate the alarm propagation network, so that the problem of alarm flooding cannot be solved, and alarm prevention is realized.
Disclosure of Invention
In order to solve the technical problems, the invention provides an alarm analysis method for industrial alarm flooding, which combines knowledge driving and data driving to construct an alarm propagation network, so as to realize the research on the alarm propagation network, the root cause of the occurrence of the alarm and the indirect relation; the method comprises the steps of initializing the transmission relation of alarm data by utilizing data driving, training an initialized network structure by combining knowledge driving to obtain an alarm propagation network, and finally realizing alarm tracing by utilizing accurate reasoning, so as to help operators to realize alarm root positioning and know alarm variables possibly happening next, prevent the occurrence of alarm domino effect, and further effectively solve the problem of alarm flooding.
The invention relates to an alarm analysis method for industrial alarm flooding, which comprises the following steps:
step 1, acquiring a historical alarm event log, preprocessing the historical alarm event log, extracting an industrial alarm flooding sequence caused by a significant fault, and acquiring alarm process data;
step 2, determining a high threshold value and a low threshold value of each alarm variable, and labeling alarm process data to obtain alarm data;
step 3, for obtaining alarm data, identifying the relation between alarm variables by using a data driving method, and initializing an alarm propagation network;
step 4, further training the initialized alarm propagation network by using a knowledge-driven method to obtain a final alarm propagation network;
step 5, utilizing the finally obtained alarm propagation network to reversely infer alarm data, and determining an alarm root.
Further, in step 2, according toPrinciple is exactDetermining a high threshold and a low threshold of each alarm variable, and judging alarm process data and the high threshold and the low threshold to obtain alarm data; when the alarm process data is lower than the low threshold value, the alarm process data state is an alarm state, marked as 1, and marked with a low threshold value label; when the alarm process data is higher than the high threshold value, the alarm process data state is an alarm state, marked as 1, and marked with a high threshold value label; the rest alarm process data are in a normal state and recorded as 0; the high threshold and the low threshold are determined as follows:
，
，
wherein,mean value representing that the process variable is in a normal state, +.>Standard deviation indicating that the process variable is in a normal state, +.>Indicating that the variable is at time +.>Process data of->Representing a low threshold value of the variable, +.>Representing the high threshold of the variable.
Further, in step 3, for the alarm data obtained in step 2, the relation between fault variables is identified by using the transfer entropy between every two variables, and the formula of the transfer entropy is as follows:
，
wherein,expression sequence->At sampling instant +.>Value of->Expression sequence->At sampling instant +.>Value of->Expression sequence->At sampling instant +.>Value of->Expression sequence->At time->Probability of corresponding value, ++>Is an excessive probability;
after the transfer entropy between every two variables is calculated, the cause and effect are judged by the following method, ifThe transfer entropy of (2) is greater than->Is called->For reasons of (I)>Known as the effect, to initialize the alert propagation network.
Further, the step 4 specifically includes:
step 4-1 processing the annular path phenomenon existing in the initial alarm propagation network by using a ring-removing method;
and 4-2, training the processed alarm propagation network by using the proposed improved K2 algorithm to obtain an optimal alarm propagation path, thereby obtaining a final alarm propagation network.
Further, the step 4-1 specifically comprises: the alarm propagation network is represented by a directed acyclic graph,wherein V represents all nodes in the alarm propagation network, namely equipment affected by the alarm, E represents all edges in the directed acyclic graph, and represents the transfer relation between alarms;
defining each alarm variable of the alarm as a node, wherein in the training process, the node, namely a child node, needs to be selected first; defining a node directly connected with the selected node as a father node; defining a node directly connected with a parent node of the selected node as an ancestor node;
judging whether the child node belongs to the ancestor node by recording the ancestor node of each node, namely the node directly or indirectly connected with the father node and the value of the transfer entropy between each node and the child node, and selecting the value with the maximum transfer entropy between the nodes each time; if yes, discarding the relation between the nodes; otherwise, the relation between the nodes is reserved; the process is repeated until all nodes are judged to be finished.
Further, the step 4-2 specifically comprises: calculating the proportion of time occupied by the variable in different states by adopting an improved K2 scoring function, and obtaining more alarm propagation paths when the correctness of the alarm propagation paths is ensured; regarding different network structures obtained through training, taking the network structure with the highest score as a final alarm transmission network;
the K2 scoring function after improvement is as follows:
，
wherein,and->Respectively representing the number of father nodes and the number of directed edges for each child node +.>Representing the number of nodes>Representing the variable->Parent node set,/->Is a variable->First->Each child node takes the value +.>Time weight occupied by time, +.>And adding the time weights occupied by each child node in different states.
Further, the step 5 specifically includes: and determining an alarm propagation path of a certain alarm occurrence possibility according to an alarm propagation network, respectively calculating posterior probabilities of different propagation paths, then calculating average posterior probabilities of alarm states of different paths, and taking a path with the maximum average alarm posterior probability as a correct propagation path, wherein a root node of the path is an alarm root.
The beneficial effects of the invention are as follows: the method provided by the invention further distinguishes the fault data according to the high and low thresholds in the data processing stage, so that the reason why the fault occurs can be positioned more clearly later, and the alarm propagation network can obtain more detailed information finally; for the problem that an alarm propagation network is difficult to obtain due to the lack of an efficient and stable modeling method, so that the alarm variable possibly generated in the next process cannot be known and the problem that multiple variables are difficult to process and alarm simultaneously is difficult to occur, the invention combines a data driving mode and a knowledge driving mode to construct the alarm propagation network, the causality of the alarm propagation network is primarily judged by using the data driving mode, and the whole fault propagation network can be further judged by the knowledge driving mode; the adopted training process can avoid the condition of looping between alarm variables, and the proposed improved K2 algorithm can keep more correct fault propagation paths in the final training process; the localization of the alarm root cause can be combined with accurate reasoning after the alarm propagation network is obtained, thereby effectively helping operators to cope with the problem of alarm flooding.
Drawings
FIG. 1 is a flow chart of an alarm analysis method for industrial alarm flooding;
FIG. 2 is a schematic diagram of a preliminary alarm propagation network;
FIG. 3 is a schematic diagram of the final alarm propagation network;
FIG. 4 is a pictorial view of an alarm propagation network learned by BIC in the prior art;
FIG. 5 is a diagram of an alarm propagation network obtained by TE learning in the prior art;
fig. 6 is a diagram of an alarm propagation network obtained by K2 learning in the prior art.
Detailed Description
In order that the invention may be more readily understood, a more particular description of the invention will be rendered by reference to specific embodiments that are illustrated in the appended drawings.
As shown in fig. 1, an alarm analysis method for industrial alarm flooding comprises the following steps:
step 1, acquiring a historical alarm event log, preprocessing the historical alarm event log, extracting an industrial alarm flooding sequence caused by a significant fault, and acquiring alarm process data;
step 2, determining a high threshold value and a low threshold value of each alarm variable, and labeling alarm process data to obtain alarm data;
step 3, for obtaining alarm data, identifying the relation between alarm variables by using a data driving method, and initializing an alarm propagation network;
step 4, further training the obtained initialized alarm propagation network by using a knowledge-driven method to obtain a final alarm propagation network;
and 5, utilizing the finally obtained alarm propagation network to reversely infer alarm data, and determining an alarm source.
In the embodiment, the method is verified by adopting a Tenn Islaman process, and the Tenn Islaman process is simulated by a simulink module of Matlab. And generating an alarm time log by triggering faults. Considering that the Tennesse Islaman process involves more module variables, the simulation verification is mainly performed by selecting variables of A feed, D feed, E feed, reactor feed speed, reactor internal pressure value, reactor liquid level, reactor internal temperature and cooling water outlet temperature, and the specific process variable description is shown in table 1.
Table 1 selected process variable description
。
After the experimental module was selected, taking the example of a loss of feed a as an example, the simulation was performed for 72 hours, and a fault was introduced after the system was operated for 12 hours to stabilize. Data sampling interval the time period was set to 10 minutes, a total of 360 sets of data.
And further processing and converting the variable process data obtained through simulation into alarm data. Meanwhile, in order to distinguish alarm data into high-threshold alarm and low-threshold alarm, the high-low threshold of each variable is obtained by using a 3 sigma criterion, and the high-low threshold of each variable is shown in table 2.
TABLE 2 threshold setting of process variables
。
The transfer entropy can be suitable for calculating linear variables and calculating nonlinear variables under the condition of more variables; the final alarm propagation network needs to determine the causal transfer relation of the alarm, and the transfer entropy between every two variables is calculated by using the calculated alarm data.
Two alarm variables x and y in the alarm are calculated respectivelyTransfer entropy sum->If the transfer entropy of (a)The transfer entropy of (2) is greater than->X is called a factor and y is called a fruit; the transfer entropy calculation formula is as follows:
，
wherein,representation of sequence->At sampling instant +.>Value of->Expression sequence->At sampling instant +.>Value of->Expression sequence->At sampling instant +.>Value of->Expression sequence->At time->The probability of the corresponding value is determined, and (2)>Is an excessive probability.
For the obtained transfer entropy between the nodes, the causal relationship between every two nodes can be further determined according to the size of the transfer entropy between the nodes, and the judgment result is shown in fig. 2.
For the primarily obtained alarm propagation network, in order to avoid the annular phenomenon of the finally generated alarm propagation network, the node with larger transmission information among alarm variables is reserved to solve the problem that an alarm transmission path is looped. The method comprises the following steps: judging whether the child node belongs to the ancestor node by recording the ancestor node of each node, namely the node directly or indirectly connected with the father node, and the value of the transfer entropy between each node and the child node, and selecting the value with the maximum transfer entropy between the nodes each time; if yes, discarding the relation between the nodes; otherwise, the relation between the nodes is reserved; and repeating the process until all nodes are judged to be finished. Thereby avoiding the occurrence of a ring network. The whole network is then trained by the proposed modified K2 algorithm, resulting in a final alarm propagation network as shown in fig. 3.
Finally, by utilizing the learned network parameters, the root cause of the alarm generation is determined to be caused by the loss of the feed A by calculating the posterior probability of each path.
Comparing the method of the invention with the network propagation path accuracy obtained by BIC, TE and K2 learning, the alarm propagation network obtained by each algorithm learning is shown in figures 4 to 6, and the learning condition statistics of each method is shown in table 3.
Table 3 learning condition statistics for each method
；
As can be seen from Table 3, the accuracy of the method of the present invention is higher.
The foregoing is merely a preferred embodiment of the present invention, and is not intended to limit the present invention, and all equivalent variations using the description and drawings of the present invention are within the scope of the present invention.
Claims (7)
1. An alarm analysis method for industrial alarm flooding is characterized by comprising the following steps:
step 1, acquiring a historical alarm event log, preprocessing the historical alarm event log, extracting an industrial alarm flooding sequence caused by a significant fault, and acquiring alarm process data;
step 2, determining a high threshold value and a low threshold value of each alarm variable, and labeling alarm process data to obtain alarm data;
step 3, for obtaining alarm data, identifying the relation between alarm variables by using a data driving method, and initializing an alarm propagation network;
step 4, further training the initialized alarm propagation network by using a knowledge-driven method to obtain a final alarm propagation network;
and 5, utilizing the finally obtained alarm propagation network to reversely infer alarm data, and determining an alarm source.
2. The method for alarm analysis of industrial alarm flooding of claim 1, wherein in step 2, the method is based onDetermining a high threshold and a low threshold of each alarm variable in principle, and judging alarm process data and the high threshold and the low threshold to obtain alarm data; when the alarm process data is lower than the low threshold value, the alarm process data state is an alarm state, marked as 1, and marked with a low threshold value label; when the alarm process data is higher than the high threshold value, the alarm process data state is an alarm state, marked as 1, and marked with a high threshold value label; the rest alarm process data are in a normal state and recorded as 0; the high threshold and the low threshold are determined as follows:
，
，
wherein,mean value representing that the process variable is in a normal state, +.>Representing the standard deviation of the process variable in a normal state,indicating that the variable is at time +.>Process data of->Representing a low threshold value of the variable, +.>Representing the high threshold of the variable.
3. The alarm analysis method for industrial alarm flooding according to claim 1, wherein in step 3, the relation between fault variables is identified by using a transfer entropy between two variables for the alarm data obtained in step 2, and the transfer entropy is expressed as follows:
，
wherein the method comprises the steps of the process comprises,expression sequence->At sampling instant +.>Value of->Expression sequence->At sampling instant +.>Value of->Expression sequence->At sampling instant +.>Value of->Expression sequence->At time->Probability of corresponding value , / / >Is an excessive probability;
after the transfer entropy between every two variables is calculated, the cause and effect are judged by the following method, ifThe transfer entropy of (2) is greater than->Is called->For reasons of (I)>Known as the effect, to initialize the alert propagation network.
4. The method for alarm analysis of industrial alarm flooding according to claim 1, wherein step 4 is specifically:
step 4-1, processing a ring path phenomenon existing in an initial alarm propagation network by using a ring-removing method;
and 4-2, training the processed alarm propagation network by using the proposed improved K2 algorithm to obtain an optimal alarm propagation path, thereby obtaining a final alarm propagation network.
5. The method for alarm analysis of industrial alarm flooding of claim 4, wherein step 4-1 is specifically: the alarm propagation network is represented by a directed acyclic graph,wherein V represents all nodes in the alarm propagation network, namely equipment affected by the alarm, E represents all edges in the directed acyclic graph, and represents the transfer relation between alarms;
defining each alarm variable of the alarm as a node, wherein in the training process, the node, namely a child node, needs to be selected first; defining a node directly connected with the selected node as a father node; defining a node directly connected with a parent node of the selected node as an ancestor node;
judging whether the child node belongs to the ancestor node by recording the ancestor node of each node, namely the node directly or indirectly connected with the father node and the value of the transfer entropy between each node and the child node, and selecting the value with the maximum transfer entropy between the nodes each time; if yes, discarding the relation between the nodes; otherwise, the relation between the nodes is reserved; the process is repeated until all nodes are judged to be finished.
6. The method for alarm analysis of industrial alarm flooding of claim 5, wherein step 4-2 is specifically: calculating the proportion of time occupied by the variable in different states by adopting an improved K2 scoring function, and obtaining more alarm propagation paths when the correctness of the alarm propagation paths is ensured; regarding different network structures obtained through training, taking the network structure with the highest score as a final alarm transmission network;
the K2 scoring function after improvement is as follows:
，
wherein,and->Respectively representing the number of father nodes and the number of directed edges for each child node +.>Representing the number of nodes>Representing the variable->Parent node set,/->Is a variable->First->Each child node takes the value +.>Time weight occupied by time, +.>And adding the time weights occupied by each child node in different states.
7. The method for alarm analysis of industrial alarm flooding of claim 5, wherein step 5 is specifically: and determining an alarm propagation path of a certain alarm occurrence possibility according to an alarm propagation network, respectively calculating posterior probabilities of different propagation paths, then calculating average posterior probabilities of alarm states of different paths, and taking a path with the maximum average alarm posterior probability as a correct propagation path, wherein a root node of the path is an alarm root.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311386097.XA CN117118811A (en) | 2023-10-25 | 2023-10-25 | Alarm analysis method for industrial alarm flooding |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311386097.XA CN117118811A (en) | 2023-10-25 | 2023-10-25 | Alarm analysis method for industrial alarm flooding |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117118811A true CN117118811A (en) | 2023-11-24 |
Family
ID=88798811
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311386097.XA Pending CN117118811A (en) | 2023-10-25 | 2023-10-25 | Alarm analysis method for industrial alarm flooding |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117118811A (en) |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102436740A (en) * | 2011-09-29 | 2012-05-02 | 东南大学 | Automatic detection method of traffic incident on highway |
| CN107784314A (en) * | 2016-08-26 | 2018-03-09 | 北京协同创新智能电网技术有限公司 | Normal the abnormal data division methods and system of a kind of multivariable warning system |
| CN108922139A (en) * | 2018-07-23 | 2018-11-30 | 山东科技大学 | A kind of online test method that industry alarm is spread unchecked |
| CN108922140A (en) * | 2018-08-07 | 2018-11-30 | 山东科技大学 | It is a kind of based on N-gram model industry alarm spread unchecked prediction technique |
| EP3651121A1 (en) * | 2018-11-09 | 2020-05-13 | Siemens Aktiengesellschaft | Assistance system for analysis of a trained causal network |
| US20200151574A1 (en) * | 2018-11-12 | 2020-05-14 | Fujitsu Limited | Computer-readable recording medium having stored therein learning program, learning method, and learning apparatus |
| CN113722140A (en) * | 2021-08-30 | 2021-11-30 | 中国地质大学（武汉） | Industrial alarm flooding root cause diagnosis method based on small sample learning and storage medium |
| US20220358266A1 (en) * | 2021-05-07 | 2022-11-10 | Dalian University Of Technology | Method and system of sudden water pollutant source detection by forward-inverse coupling |
| CN115758908A (en) * | 2022-11-29 | 2023-03-07 | 中国地质大学（武汉） | Alarm online prediction method under alarm flooding condition based on deep learning |
| WO2023124778A1 (en) * | 2021-12-29 | 2023-07-06 | 浙江中控技术股份有限公司 | Real-time alarm tracing apparatus and method in process industry production procedure |
-
2023
- 2023-10-25 CN CN202311386097.XA patent/CN117118811A/en active Pending
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102436740A (en) * | 2011-09-29 | 2012-05-02 | 东南大学 | Automatic detection method of traffic incident on highway |
| CN107784314A (en) * | 2016-08-26 | 2018-03-09 | 北京协同创新智能电网技术有限公司 | Normal the abnormal data division methods and system of a kind of multivariable warning system |
| CN108922139A (en) * | 2018-07-23 | 2018-11-30 | 山东科技大学 | A kind of online test method that industry alarm is spread unchecked |
| CN108922140A (en) * | 2018-08-07 | 2018-11-30 | 山东科技大学 | It is a kind of based on N-gram model industry alarm spread unchecked prediction technique |
| EP3651121A1 (en) * | 2018-11-09 | 2020-05-13 | Siemens Aktiengesellschaft | Assistance system for analysis of a trained causal network |
| US20200151574A1 (en) * | 2018-11-12 | 2020-05-14 | Fujitsu Limited | Computer-readable recording medium having stored therein learning program, learning method, and learning apparatus |
| US20220358266A1 (en) * | 2021-05-07 | 2022-11-10 | Dalian University Of Technology | Method and system of sudden water pollutant source detection by forward-inverse coupling |
| CN113722140A (en) * | 2021-08-30 | 2021-11-30 | 中国地质大学（武汉） | Industrial alarm flooding root cause diagnosis method based on small sample learning and storage medium |
| WO2023124778A1 (en) * | 2021-12-29 | 2023-07-06 | 浙江中控技术股份有限公司 | Real-time alarm tracing apparatus and method in process industry production procedure |
| CN115758908A (en) * | 2022-11-29 | 2023-03-07 | 中国地质大学（武汉） | Alarm online prediction method under alarm flooding condition based on deep learning |
Non-Patent Citations (4)
| Title |
|---|
| 吴水龙;郭阳;曲广浩;杨虎林;陈矿;刘书杰;: "基于Savitzky-Golay滤波器的多变量监控报警方法研究", 计算机与应用化学, no. 05 * |
| 柯永琦: "基于传递熵与改进 K2 算法生成报警传播网络", 《计算机系统应用》, vol. 32, no. 02, pages 1 - 4 * |
| 柯永琦: "基于传递熵与改进K2算法生成报警传播网络", 《计算机系统应用》, vol. 32, no. 02, pages 1 - 4 * |
| 王文铎: "基于深度学习的工业过程报警因果分析与溯源", 《中国优秀硕士学位论文全文数据库》, no. 02 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108460207A (en) | A kind of fault early warning method of the generating set based on operation data model | |
| CN110287552A (en) | Based on the motor bearings fault diagnosis method and system for improving random forests algorithm | |
| CN103218689A (en) | Analyzing method and analyzing device for operator state assessment reliability | |
| CN113282635A (en) | Micro-service system fault root cause positioning method and device | |
| Wunderlich et al. | Structure learning methods for Bayesian networks to reduce alarm floods by identifying the root cause | |
| CN116660672B (en) | Power grid equipment fault diagnosis method and system based on big data | |
| CN103675355A (en) | Anemometer monitoring method and system | |
| CN109861825B (en) | Internal attack detection method based on weighting rule and consistency in CPS system | |
| CN112632845B (en) | Data-based mini-reactor online fault diagnosis method, medium and equipment | |
| Bodla et al. | Logistic regression and feature extraction based fault diagnosis of main bearing of wind turbines | |
| CN114721336A (en) | Information security event early warning method for technological parameters of instrument control system | |
| CN104571084B (en) | Main air compressor fault rootstock depth diagnostic method and device | |
| CN112288298A (en) | Desulfurization system health condition evaluation method based on online data | |
| CN108037387B (en) | Equipment fault analysis method and device based on clustering | |
| CN114519382A (en) | Nuclear power plant key operation parameter extraction and abnormity monitoring method | |
| CN103675356A (en) | Anemometer fault detection method and system on the basis of particle swarm optimization | |
| CN113313365A (en) | Degradation early warning method and device for primary air fan | |
| CN103675357A (en) | Anemometer fault warning method and system | |
| Hu et al. | Cause and effect analysis for decision support in alarm floods | |
| CN117118811A (en) | Alarm analysis method for industrial alarm flooding | |
| CN116701846A (en) | Hydropower station dispatching operation data cleaning method based on unsupervised learning | |
| CN115560962A (en) | Fan abnormal state detection method based on multi-level space-time diagram neural network | |
| CN112699598A (en) | Intelligent diagnosis method and device for abnormal oil temperature of gear box | |
| Zavisca et al. | A bayesian network approach to accident management and estimation of source terms for emergency planning | |
| CN112115128A (en) | 5G real-time dynamic fault processing system based on multi-dimensional sensor |
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 |