CN118034990B - Concentrator verification method and system based on machine learning - Google Patents
Concentrator verification method and system based on machine learning Download PDFInfo
- Publication number
- CN118034990B CN118034990B CN202410432677.6A CN202410432677A CN118034990B CN 118034990 B CN118034990 B CN 118034990B CN 202410432677 A CN202410432677 A CN 202410432677A CN 118034990 B CN118034990 B CN 118034990B
- Authority
- CN
- China
- Prior art keywords
- concentrator
- user
- data
- determining
- preset condition
- 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
- 238000000034 method Methods 0.000 title claims abstract description 35
- 238000012795 verification Methods 0.000 title claims abstract description 31
- 238000010801 machine learning Methods 0.000 title claims abstract description 18
- 230000007246 mechanism Effects 0.000 claims abstract description 33
- 230000004044 response Effects 0.000 claims abstract description 9
- 230000008054 signal transmission Effects 0.000 claims description 17
- 230000003247 decreasing effect Effects 0.000 claims 1
- 239000013598 vector Substances 0.000 description 33
- 230000005540 biological transmission Effects 0.000 description 9
- 230000008569 process Effects 0.000 description 8
- 238000012549 training Methods 0.000 description 8
- 238000011144 upstream manufacturing Methods 0.000 description 7
- 238000007792 addition Methods 0.000 description 6
- 238000012986 modification Methods 0.000 description 6
- 230000004048 modification Effects 0.000 description 6
- 238000012545 processing Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 238000012544 monitoring process Methods 0.000 description 4
- 230000008439 repair process Effects 0.000 description 4
- 238000004891 communication Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 230000006872 improvement Effects 0.000 description 3
- 230000009467 reduction Effects 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000006978 adaptation Effects 0.000 description 1
- 230000000712 assembly Effects 0.000 description 1
- 238000000429 assembly Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000014509 gene expression Effects 0.000 description 1
- 238000011478 gradient descent method Methods 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 239000003607 modifier Substances 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
Landscapes
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The embodiment of the specification provides a concentrator verification method and system based on machine learning. The method comprises the following steps: acquiring a network map of the concentrator, wherein the network map comprises nodes and edges; determining relevant users and relevant edges from the nodes and edges based on the network map; acquiring actual working data of a concentrator, related users and related edges; and determining a reaction mechanism in response to the actual working data not meeting the preset condition.
Description
Technical Field
The present disclosure relates to the field of concentrators, and in particular, to a concentrator verification method and system based on machine learning.
Background
With the development of the internet of things and intelligent systems, a concentrator plays an important role in data transmission and communication as a key device. However, due to various factors, the concentrator may be subjected to various faults, including hardware faults, communication faults, software faults, and the like. In order to detect and diagnose these faults in time, verification techniques for concentrator faults are continually evolving and perfecting.
Therefore, it is desirable to provide a concentrator verification method and system based on machine learning, which can accurately determine a relevant attention range, reduce an attention data range, reduce the calculation amount, and improve the accuracy of determining a concentrator fault.
Disclosure of Invention
According to the concentrator verification method and system based on machine learning, the relevant attention range is accurately determined, the attention data range is reduced, the calculated amount is reduced, and meanwhile the accuracy of determining faults of the concentrator can be improved.
One of the embodiments of the present specification provides a machine learning-based concentrator verification method, which includes: acquiring a network map of the concentrator, wherein the network map comprises nodes and edges; determining relevant users and relevant edges from the nodes and edges based on the network map; acquiring actual working data of the concentrator, the related users and the related edges; and determining a reaction mechanism in response to the actual working data not meeting a preset condition.
One of the embodiments of the present specification provides a machine learning based concentrator verification system, the system comprising: the first acquisition module is used for acquiring a network map of the concentrator, wherein the network map comprises nodes and edges; a correlation determination module for determining a correlation user and a correlation edge from the nodes and the edges based on the network map; the second acquisition module is used for acquiring actual working data of the concentrator, the related users and the related edges; and the reaction mechanism determining module is used for determining a reaction mechanism in response to the fact that the actual working data does not meet the preset condition.
One of the embodiments of the present specification provides a machine learning based concentrator verification device, the device comprising a processor and a memory; the memory is configured to store instructions that, when executed by the processor, cause the apparatus to implement a concentrator verification method as described in any one of the preceding claims.
One of the embodiments of the present disclosure provides a computer readable storage medium, wherein the storage medium stores computer instructions, and when the computer reads the computer instructions in the storage medium, the computer executes the concentrator verification method according to any one of the above.
Drawings
The present specification will be further elucidated by way of example embodiments, which will be described in detail by means of the accompanying drawings. The embodiments are not limiting, in which like numerals represent like structures, wherein:
FIG. 1 is a schematic illustration of an application scenario of a concentrator verification system shown in some embodiments of the present description;
FIG. 2 is a block diagram of a concentrator verification system shown in accordance with some embodiments of the present description;
FIG. 3 is an exemplary flow chart of a concentrator verification method shown in accordance with some embodiments of the present description;
FIG. 4 is an exemplary flow chart for determining preset conditions according to some embodiments of the present description;
Fig. 5 is an exemplary flow chart for determining target preset conditions according to some embodiments of the present description.
Detailed Description
In order to more clearly illustrate the technical solutions of the embodiments of the present specification, the drawings that are required to be used in the description of the embodiments will be briefly described below. It is apparent that the drawings in the following description are only some examples or embodiments of the present specification, and it is possible for those of ordinary skill in the art to apply the present specification to other similar situations according to the drawings without inventive effort. Unless otherwise apparent from the context of the language or otherwise specified, like reference numerals in the figures refer to like structures or operations.
It will be appreciated that "system," "apparatus," "unit" and/or "module" as used herein is one method for distinguishing between different components, elements, parts, portions or assemblies at different levels. However, if other words can achieve the same purpose, the words can be replaced by other expressions.
As used in this specification and the claims, the terms "a," "an," "the," and/or "the" are not specific to a singular, but may include a plurality, unless the context clearly dictates otherwise. In general, the terms "comprises" and "comprising" merely indicate that the steps and elements are explicitly identified, and they do not constitute an exclusive list, as other steps or elements may be included in a method or apparatus.
A flowchart is used in this specification to describe the operations performed by the system according to embodiments of the present specification. It should be appreciated that the preceding or following operations are not necessarily performed in order precisely. Rather, the steps may be processed in reverse order or simultaneously. Also, other operations may be added to or removed from these processes.
Fig. 1 is a schematic diagram of an application scenario of a concentrator verification system shown in some embodiments of the present description. As shown in fig. 1, a scenario 100 involved in a concentrator verification system may include a processor 110, a concentrator 120, a storage device 130, a network 140, and a user terminal 150.
In some embodiments, the processor 110 may be configured to process information and/or data related to the scene 100, e.g., the processor 110 may obtain a network map based on the data. In some embodiments, the processor 110 may be local or remote. For example, processor 110 may access information and/or data stored in storage device 130, user terminal 150 via network 140. As another example, the processor 110 may be directly connected to the storage device 130, the user terminal 150 to access stored information and/or data.
Concentrator 120 may refer to a device in a network for collecting, processing, and forwarding data/signals. The concentrator 120 may be at least one. The concentrator 120 may include a plurality of devices and the concentrator 120 may be connected to a plurality of users. In some embodiments, the concentrator 120 may be integrated with the processor 110.
The storage device 130 may be used to store data and/or instructions related to concentrator verification. In some embodiments, the storage device 130 may store data obtained/acquired from the user terminal 150. In some embodiments, the storage device 130 may store data and/or instructions that the processor 110 uses to execute or use to perform the exemplary methods described herein. In some embodiments, storage device 130 may be implemented on a cloud platform.
In some embodiments, the storage device 130 may be connected to the network 140 to communicate with one or more components of the scenario 100 (e.g., the processor 110, the concentrator 120, the user terminal 150). One or more components of the scenario 100 may access data or instructions stored in the storage device 130 via the network 140. In some embodiments, the storage device 130 may be directly connected to or in communication with one or more components of the scene 100 (e.g., the processor 110, the concentrator 120, the user terminal 150). In some embodiments, the storage device 130 may be part of the processor 110. In some embodiments, the storage device 130 may be a separate memory. The storage device 140 may store historical data, e.g., historical work data, user characteristic data, etc. The storage device 140 may also store machine learning models and the like.
The network 140 may facilitate the exchange of information and/or data. In some embodiments, one or more components of the scenario 100 (e.g., the processor 110, the concentrator 120, the user terminal 150) may send information and/or data to other components of the scenario 100 via the network 140. At least one concentrator may be included in the network 140, and the at least one concentrator may be connected to at least one user via the network 140.
The user terminal 150 may be a device used by a user to connect with the concentrator 120. User terminal 150 may also be a device of a user viewing the entire network. The user may be an upstream user or a downstream user of the concentrator 120.
FIG. 2 is a block diagram of a concentrator verification system shown in accordance with some embodiments of the present description.
As shown in fig. 2, the concentrator verification system 200 may include a first acquisition module 210, a correlation determination module 220, a second acquisition module 230, and a reaction mechanism determination module 240.
The first acquisition module 210 may be configured to acquire a network graph of where the concentrator is located, the network graph including nodes and edges.
The correlation determination module 220 may be used to determine correlation users and correlation edges from nodes and edges based on a network map.
The second acquisition module 230 may be used to acquire actual working data of the concentrator, the relevant user and the relevant edge.
The reaction mechanism determining module 240 may be configured to determine the reaction mechanism in response to the actual working data not meeting the preset condition.
In some embodiments, the preset condition includes at least one of a value of the working data being within a dynamic fluctuation range, a value of the working data being less than a dynamic threshold value, and a difference value, and the reaction mechanism determining module further includes: an ideal module for determining ideal operational data for the concentrator, associated users and associated edges at a set point/segment in time; and the difference module is used for responding to the fact that the difference between the actual working data and the ideal working data exceeds a difference value and making a corresponding reaction mechanism.
In some embodiments, the preset conditions are dynamically adjusted as target preset conditions.
In some embodiments, the reaction mechanism determination module 240 includes: the third acquisition module is used for acquiring history increase and decrease data; the increase and decrease user determining module is used for predicting the joining quantity of at least one group of new users, the increasing possibility, the cancellation quantity of at least one group of old users and the cancellation possibility of the new users through the prediction map model based on the historical increase and decrease data and the network map; the final determining module is used for determining the final joining quantity of the new user and the final cancellation quantity of the old user based on the joining quantity of the new user, the increasing possibility of the new user, the cancellation quantity of the old user and the cancellation possibility of the old user; the coefficient determining module is used for determining a dynamic adjustment coefficient based on the final joining quantity of the new user, the corresponding increasing possibility, the final cancellation quantity of the old user and the corresponding cancellation possibility; and the target condition determining module is used for redefining the preset condition based on the preset condition and the dynamic adjustment coefficient to serve as a target preset condition.
It should be understood that the above modules are merely a simple example of related modules mainly referred to in this specification, and do not represent a presentation of all relevant matters of the present application, and some modules and units are not shown in the present block diagram, which is not illustrated herein. And the above modules and units are not completely independent but may be related to each other.
Fig. 3 is an exemplary flow chart of a concentrator verification method shown in accordance with some embodiments of the present description. The process 300 may be performed by a processor. As shown in fig. 3, the process 300 may include the steps of:
step 310, a network map of the concentrator is obtained.
The network map may be a map of the connection relationship to the network in which the concentrator is located. The network graph includes nodes and edges. Nodes in the network map may include receiving points, transmitting points of data/signals.
The nodes may include subscribers and concentrators. The concentrator may refer to a device that provides centralized management, distribution, and forwarding of data/signals. In the current network, there may be at least one concentrator. The user may comprise at least one upstream user and/or at least one downstream user. Upstream users may be directed to users of the source from which the concentrator provides data or information. The upstream users may include various databases, various devices, etc. data sources. Downstream users may refer to terminals, applications, etc. that receive data or information from the concentrator. The downstream user can acquire the data acquired from the upstream user through the concentrator for data processing, storage, distribution or display.
The node attributes corresponding to the nodes may include node attributes of the user and node attributes of the concentrator. The node attributes of the concentrator may include physical attributes of the concentrator and time points/segments, corresponding repair information, and the like. The node attributes of the user may include the point in time/segment and the corresponding user characteristics. The user characteristics are characteristics related to user usage. The user characteristics may include user category, usage habits, physical attributes of the user terminal device, repair information, and the like. The user categories may be residential users, general server users, databases, etc., the usage habits may include peak categories of user usage, which may include points in time or periods of data/signal transmission over a pre-set peak threshold. Wherein the peak threshold may be set manually. For example, 7:00 a.m. to 10:00 a.m. are periods of time when the user is using the network more frequently, and thus the peak class is AM 7:00-PM10:00. The physical attributes of the concentrator, the user terminal device may include age, model number of the terminal device, physical address, etc. The report information may include the frequency, number of times, and time nodes of the history report to the server. The data/signal transmission quantity of the concentrator is reduced due to the fault problem of the user, which is the normal condition, and the misjudgment on the fault of the concentrator is reduced by adding the report and repair information. In some embodiments, the node attributes may be obtained from a storage device or connection with corresponding historical data.
Edges in the network map may include transmission paths for data, including wireless transmission paths and wired transmission paths. The edge attributes corresponding to the edges may include the physical attributes of the edges, which may include the connection manner with the concentrator, the composition of the connection lines, the service life, the connection length (of the concentrator and the corresponding user), the error reporting information, and the like. The edge attributes may also be obtained from a storage device or view the corresponding work log.
The network map can be updated in real time according to information such as network registration and the like, so that the influence of adding new users and/or logging out old users on the accuracy of determining the concentrator faults is reduced.
At step 320, relevant users and relevant edges are determined from the nodes and edges based on the network map.
The relevant user may refer to a user that exceeds a traffic threshold with the current concentrator data/signal transmission. The relevant edge may refer to the edge through which there is currently data/signal traffic exceeding the traffic threshold. The flow threshold may be set manually.
In some embodiments, all first users and first edges connected to the current concentrator may be determined based on the network map.
The first edge may refer to an edge that is directly connected to the current concentrator in the network map. The connection may include a wired connection and a wireless connection. The first user may refer to a device connected to the concentrator via a first edge. Data/signal transmission between the user equipment and the concentrator can be achieved by means of the connection device.
Because the online and offline states of the concentrator, connection side, user, etc. devices are dynamically changed during different periods of time, the user who is in data/signal transmission with the concentrator is also dynamically changed. In some embodiments, the relevant user and the relevant edge may be dynamically determined from the first user, the first edge. For example, the data/signal transmission amounts of the downstream users are different between the working time and the off-hours on the working day, and the data/signal transmission amounts of the nodes of the corresponding home are smaller than those of the off-hours because the users are not at home during the working time. As another example, the first user connected to the concentrator may have reduced data/signal transmission with the concentrator due to physical causes such as equipment failure, power outage, etc. on the first side or itself. When the data/signal traffic between the user and the concentrator is less than the traffic threshold, the first user connected to the concentrator is not the relevant user.
In some embodiments, node attributes of the first user may be obtained based on a network map, an online probability of the first user at a preset time point or a time period is determined through an online probability model based on user features of the first user, and a related user and a related edge in the preset time point or the time period are determined based on the online probability of the first user.
The online probability model may be a machine learning model, for example, a deep neural network. The input of the feature layer may include first user features corresponding to the concentrator and node attributes of the concentrator, and the output may be an online probability of each first user. In some embodiments, a first user having an online probability greater than an online threshold may be determined to be a relevant user and a relevant edge based on the concentrator and a first edge of the relevant user. The online threshold may be set manually. The online probability model may be obtained through training.
Through some embodiments of the present disclosure, a first user may be determined based on a network map, and a relevant user may be determined from the first user, so that a range of users concerned may be reduced, and users having a greater influence on the working data of the concentrator may be accurately determined.
In step 330, the actual working data of the concentrator, the relevant user and the relevant edge is obtained.
The actual operating data for the relevant user, the relevant edge may include data/signal transmission rate, operating voltage, operating current, temperature, etc. The operational data of the concentrator may include data/signal transmission rate, forwarding rate, delay, packet loss rate, operating voltage, operating current, temperature, etc. The actual working data of the concentrator, the relevant users and the relevant edges may correspond to the current point in time. The transmission rate represents the maximum data transmission rate supported by the concentrator, typically in units of Mbps or Gbps. The higher the transmission rate, the more data processing capacity the concentrator has. A forwarding rate may refer to a rate at which a concentrator can process and forward data/signals, typically expressed in terms of the number of packets forwarded per second. The higher the forwarding rate, the more data/signal processing capacity the concentrator has. The delay may be the time required for data/signal transmission from the input port to the output port. Packet loss rate may refer to the proportion of data/signal lost during the data/signal transmission. The lower the packet loss rate, the more reliable the data/signal transmission.
In some embodiments, various network monitoring tools, sensors, etc. may be provided at the concentrator, user device, etc. The data can be collected through the network monitoring tool, and working information such as real-time data flow of the concentrator and the like can be provided. Monitoring information on the aspects of node current, voltage, temperature and the like can be obtained through the sensor. For example, a network monitoring tool such as WIRESHARK, SOLARWINDS or the like may be used to monitor the transmission rate, forwarding rate of the concentrator. For another example, the packet loss rate may be calculated by sending a packet via Ping command or a specialized network tool and checking for replies. For another example, the log file of the concentrator may be periodically checked to obtain the historical working data. For another example, data of the operating voltage, current and temperature can be obtained by the voltmeter, ammeter and thermometer that they are loaded with.
According to some embodiments of the present disclosure, only the working data of the relevant user and the relevant edge is obtained, but not the working data of all nodes in the network, so that the obtained data volume is reduced, and the processing speed is improved.
And step 340, determining a reaction mechanism in response to the actual working data not meeting the preset condition.
The preset condition refers to a condition for judging whether the concentrator is out of order. In some embodiments, the preset condition may include at least one of the value of the operational data being within a dynamic fluctuation range, the value of the operational data being less than a dynamic threshold. The dynamic fluctuation range refers to the numerical range of working data of the concentrator and related users in normal operation. The dynamic threshold value refers to the maximum limit of the numerical value of working data when the concentrator and the user node equipment are in normal operation. The dynamic fluctuation range, dynamic threshold may be preset based on a priori knowledge or historical data. When the working data does not meet the preset conditions, the concentrator can be determined to have faults with high probability, and a corresponding reaction mechanism needs to be determined.
In some embodiments, the preset conditions may include a difference value, which may determine ideal working data for the concentrator, associated users, and associated edges at a set point/segment in time; and responding to the difference between the actual working data and the ideal working data exceeding the difference value, making a corresponding reaction mechanism, and making the reaction mechanism based on preset conditions. For more details, see fig. 4 of the present description and the description thereof.
In some embodiments, when it is determined that the concentrator is malfunctioning, the reaction mechanism includes: reboot concentrator, firmware upgrade, check power and connections, replace hardware, etc. Some temporary problems, such as software errors or memory overflows, can be solved by simply restarting the concentrator. If the problem with the concentrator is due to a software bug or an unstable firmware version, it may be considered to upgrade the concentrator's firmware to the latest version to fix the known problem. In some embodiments, the problem may be due to a power failure or connection problem, the power supply of the concentrator may be ensured to be normal by checking the power supply and connection, and the connection with other devices is stable. If the problem with the concentrator is due to hardware damage, the damaged components or the entire concentrator device may need to be replaced.
According to some embodiments of the present disclosure, the relevant attention scope is accurately determined, so that the attention data scope is reduced, the calculation amount is reduced, and meanwhile, the accuracy of determining the concentrator fault is improved.
Fig. 4 is an exemplary flow chart for determining preset conditions according to some embodiments of the present description. As shown in fig. 4, the flow 400 may be performed by a processor. The process 400 may include the steps of:
at step 410, ideal operational data for the concentrator, associated users, and associated edges at the set point/segment is determined.
Ideal working data refers to standard values of working data of the concentrator, the user and the side when the concentrator operates normally. The ideal working data may correspond to a standard range of values for the working data.
In some embodiments, the ideal working data for the set point in time/segment may refer to a working data standard value for the preset point in time/segment. The preset time period may be determined based on actual requirements. For example, if it is desired to monitor the future 24 hours of operation with a high emphasis, the corresponding future preset time period may be 24 hours.
In some embodiments, the ideal operational data for the set point in time/segment may include the concentrator and user operational data corresponding to the point in time/segment where the concentrator is fault free. In some embodiments, the ideal working data may be determined based on historical data. For example, a reference work database may be constructed based on historical data, the reference work database comprising a plurality of data sets, each data set comprising a historical time period of operation without anomalies and its corresponding historical time characteristics, and reference work data corresponding to the historical time period. The reference work database may be searched based on a time characteristic of a future preset time period, at least one data set meeting the reference condition is determined, and ideal work data is determined based on the reference work data corresponding to the at least one data set. For example, the average value of the aforementioned reference work data is determined as ideal work data. The time characteristic may include at least a season in which the time period is located, a time range of the time period, and the like. The time characteristics can be used for determining a historical time period which is more similar to a future preset time period to be determined in the reference work database, and the reference work data with more reference value can be determined based on the more similar historical time period.
In some embodiments, the ideal operating data may also be determined based on a predictive model. The predictive model may be a machine learning model.
In some embodiments, the inputs to the predictive model include at least actual operational data and network profiles for the concentrator, associated users, and associated edges, and the outputs include ideal operational data.
In some cases, changes in upstream and downstream users and corresponding edges (including failures, unused) can affect the concentrator. Therefore, the working data and the user characteristics of the upstream and downstream users can be used as references to judge whether the operation of the concentrator is abnormal, so that the judgment result is determined more comprehensively and systematically, and the judgment accuracy is improved.
In some embodiments, the initial predictive model may be trained based on a plurality of second samples with second labels to obtain a trained predictive model. For example, inputting a second sample into the initial predictive model, constructing a loss function based on the output of the initial predictive model and the second label; and based on the loss function, iteratively updating parameters in the initial prediction model by a gradient descent method until a preset condition is met, and obtaining a trained prediction model after training is finished. The preset conditions may include convergence of the loss function, reaching of the iteration number to a threshold value, and the like.
In some embodiments, the second sample includes sample actual working data and a sample network map for the concentrator, the associated user, and the associated edge. The second sample may be determined based on historical data. In some embodiments, the second tag includes sample ideal work data. The second tag may be determined based on operational data of the concentrator during normal operation in the historical data.
In step 420, a corresponding reaction mechanism is made in response to the actual working data differing from the ideal working data by more than a difference value.
In some embodiments, a fault database may be preset, where the fault database includes at least one set of historical fault vectors, where any set of historical fault vectors includes historical actual working data of the concentrator and the first node when a certain historical time point/segment of the concentrator fails, and each set of historical fault vectors corresponds to a historical reaction mechanism. Any set of historical fault vectors includes: time point/segment, concentrator historical actual working data vector, historical working data vector of first node connected with concentrator. In some embodiments, the corresponding concentrator operational feature vector, the first node operational feature vector, is determined by a feature model based on the concentrator, first node operational data. The feature model is a machine learning model and can be obtained through historical data training. A certain historical time point/segment, a historical reaction mechanism can be manually established with numbers/letters. For example, a set of historical fault vectors in the fault database is (1, σ, τ), where the historical fault vectors correspond to a historical reaction mechanism A, where 1 may represent 0:00-02:00, σ represents the historical actual operational data vectors of the concentrator of 0:00-02:00, τ represents the historical actual operational data vectors of the nodes of 0:00-02:00, and A represents restarting the concentrator.
In some embodiments, the concentrator operation feature vector and the relevant node operation feature vector of the current time point/section are determined through the feature model based on the operation data of the current time point/section, the concentrator and the relevant node, and the current fault vector is determined, and the details of the current fault vector can refer to the historical fault vector. And comparing the similarity between the current fault vector and the historical fault vector in the fault database to determine the difference, and determining at least one group of historical fault vectors with the difference exceeding the difference value, thereby determining the corresponding reaction mechanism. In some embodiments, a historical fault vector with the highest similarity to the current fault vector is used as the target fault vector, and a historical reaction mechanism corresponding to the target fault vector is used as the current reaction mechanism. The difference between the actual working data and the ideal working data may be represented by a similarity between vectors, e.g., the similarity and the difference are inversely proportional, or other preset relationships. The higher the similarity, the smaller the difference between the actual working data and the ideal working data. The correspondence between the similarity and the difference may be preset.
The difference value may be manually set or may be dynamically changed, and the details may be found in fig. 5.
In some embodiments, one historical fault vector corresponds to a plurality of reaction mechanisms, for example, the concentrator is restarted, the fault is repaired, and hardware needs to be replaced to repair the fault, because a time correspondence is set in the fault vector, in some embodiments, differences of the concentrator working feature vector of the current time point/section, the working feature vector of the relevant node and the historical actual working feature vector of the concentrator and the first node in the corresponding historical fault vector can be compared first, then at least one target fault vector is determined, and the historical reaction mechanism corresponding to the target fault vector with the latest time is taken as the current reaction mechanism.
Fig. 5 is an exemplary flow chart for determining target preset conditions according to some embodiments of the present description. The process 500 may be performed by a processor.
Since the joining of new users and the cancellation of old users may affect the working data (e.g., the amount of data/signal transmitted) of the concentrator, and since the joining of new users and the cancellation of old users may not be updated in the system in time, there may be errors in determining the relevant users based on the network map and setting the corresponding preset conditions. Therefore, the addition quantity of the new user and the cancellation quantity of the old user can be predicted based on the historical increase and decrease data, the influence of the new user and the cancellation quantity of the old user on the working data of the concentrator can be determined by predicting the addition quantity of the new user and the cancellation quantity of the old user, and a proper amount of the preset conditions are dynamically adjusted, so that misjudgment on the faults of the concentrator is reduced.
In some embodiments, dynamically adjusting the preset condition may include adjusting a dynamic fluctuation range, a dynamic threshold, and/or a variance value. In some embodiments, the dynamic adjustment may adjust the dynamic fluctuation range, dynamic threshold, and/or variance value based on determining the dynamic adjustment coefficient. The dynamic adjustment coefficients may include at least one of an increase coefficient, a decrease coefficient. Increasing the coefficient may refer to adding a new user such that the preset condition changes. The reduction coefficient may refer to a coefficient by which an old user logs off such that a preset condition changes.
As shown in FIG. 5, the number of joins 541-1 and its increasing likelihood 541-2, the number of deregistrations 542-1 and its deregistration likelihood 542-2 of at least one group of old users for at least one group of new users may be predicted by the predictive pattern model 530 based on the historical increase and decrease data 510 and the network pattern 520, and the final number of joins 551-1 and the final number of deregistrations 552-1 for the new users and the old users may be determined based on the number of joins 541-1 and its increasing likelihood 541-2, the number of deregistrations 542-1 for the old users and the number of deregistrations 542-2. In some embodiments, the number of new users with the highest increasing probability may be taken as the final joining number of new users, and the number of old users with the highest log-out probability may be taken as the final log-out number of old users. Wherein the increase probability and the cancellation probability are numbers between 0 and 1.
The historical delta data may include historical points/segments of time and corresponding numbers of newly joined users and/or numbers of logged-off users.
The predictive atlas model may be a machine learning model, e.g., a graph neural network model, a deep neural network, etc. The inputs to the predictive atlas model may include the network atlas and historical increase and decrease data, and the outputs may be the likelihood that new users are increasing and old users are logging out at corresponding points in time/time.
In some embodiments, the model may be trained using a plurality of labeled training samples by a plurality of methods (e.g., gradient descent) so that parameters of the predictive atlas model may be learned. And when the trained model meets the preset conditions, training is finished, and a trained prediction spectrum model is obtained.
The training samples may include multiple sets of historical network patterns at the same point/segment in time and historical increase and decrease data. The labels of the training samples can be the number of historical joins of the historical new user and the corresponding historical increasing likelihood, the number of log-outs of the historical old user and the corresponding historical log-out likelihood. For example, multiple sets of the same time point/segment may correspond to 3 months of each year. The labels of the training samples can be obtained through manual labeling. For example, the number of historical additions may be 2, 4, 8,2 with a corresponding historical increase likelihood of 50%.
In some embodiments, dynamic adjustment coefficients 560 are determined based on a final joined number 551-1 of new users, a corresponding increased likelihood 551-2, and/or a final de-registered number 552-1 of old users, a corresponding de-registered likelihood 552-2. In some embodiments, an increase look-up table may be preset to determine an increase factor based on the number of new users 'final joins and their corresponding increase likelihoods, and a decrease factor based on the number of old users' final log-outs and their corresponding log-outs by logging-out the look-up table. Further, the increment table may include the final increment amount, the increment likelihood, and the corresponding increment coefficient. The log off look-up table may include the final log off number, log off likelihood, and corresponding reduction factor. The addition of the lookup table, the cancellation of the lookup table may be based on a priori experience or manually set. The increase coefficient may be a number greater than 1 and the decrease coefficient may be a number between 0 and 1.
In some embodiments, the preset condition may be redetermined as the target preset condition 580 based on the preset condition 570, the dynamic adjustment coefficient 580. The target preset condition is a condition after the preset condition is adjusted based on the dynamic adjustment coefficient.
In some embodiments, the target preset condition may be redetermined by formula (1);
Wherein, Preset condition for target,/>For dynamic adjustment of coefficients,/>Is the value of the preset condition.
For example, in the preset condition, the dynamic threshold requirement of the packet loss rate is 10%, after the new user increases, the increase coefficient is determined to be 1.1, and the target preset condition is 101.1=11%, which indicates that after the new user increases, the transmission amount of data and/or signals increases, and the corresponding increase of the packet loss rate is normal, and the concentrator failure cannot be judged when the qini packet loss rate exceeds the preset condition.
In some embodiments, new user addition and old user cancellation occur simultaneously, which may be determined based on the comparison results by comparing the product of the final join number and the likelihood of addition, the product of the final cancel number and the likelihood of cancellation.
If the comparison result is that the product of the final joining number and the increase possibility is greater than and/or equal to the product of the final cancellation number and the cancellation possibility, the absolute value of the difference between the final joining number and the final cancellation number is taken as the final joining number, and (increase possibility+cancellation possibility)/increase possibility is taken as the final increase possibility, and the final increase coefficient is determined by the increase comparison table.
If the comparison result is that the product of the final joining number and the increasing possibility is smaller than the product of the final cancellation number and the cancellation possibility, taking the absolute value of the difference value of the final joining number and the final cancellation number as the final cancellation number, taking the cancellation possibility/(the increasing possibility+the cancellation possibility) as the final cancellation possibility, and determining the final reduction coefficient through the cancellation comparison table.
In some embodiments of the present application, the history increase and decrease data of the history time point/section can obtain the rule of new user joining or old user logging off, and increase the accuracy of predicting new joining users and/or logging off users. The new user is added and the old user is logged out to generate influence on the working data of the concentrator, which is a normal condition, so that misjudgment on the fault of the concentrator is reduced by predicting the new user to be added and the old user to be logged out. In some embodiments, the user characteristics may be obtained from a network map or memory.
It should be noted that, the advantages that may be generated by different embodiments may be different, and in different embodiments, the advantages that may be generated may be any one or a combination of several of the above, or any other possible advantages that may be obtained.
While the basic concepts have been described above, it will be apparent to those skilled in the art that the foregoing detailed disclosure is by way of example only and is not intended to be limiting. Although not explicitly described herein, various modifications, improvements, and adaptations to the present disclosure may occur to one skilled in the art. Such modifications, improvements, and modifications are intended to be suggested within this specification, and therefore, such modifications, improvements, and modifications are intended to be included within the spirit and scope of the exemplary embodiments of the present invention.
Meanwhile, the specification uses specific words to describe the embodiments of the specification. Reference to "one embodiment," "an embodiment," and/or "some embodiments" means that a particular feature, structure, or characteristic is associated with at least one embodiment of the present description. Thus, it should be emphasized and should be appreciated that two or more references to "an embodiment" or "one embodiment" or "an alternative embodiment" in various positions in this specification are not necessarily referring to the same embodiment. Furthermore, certain features, structures, or characteristics of one or more embodiments of the present description may be combined as suitable.
Furthermore, the order in which the elements and sequences are processed, the use of numerical letters, or other designations in the description are not intended to limit the order in which the processes and methods of the description are performed unless explicitly recited in the claims. While certain presently useful embodiments have been discussed in the foregoing disclosure, by way of various examples, it is to be understood that such details are merely illustrative and that the appended claims are not limited to the disclosed embodiments, but, on the contrary, are intended to cover all modifications and equivalent arrangements included within the spirit and scope of the embodiments of the present disclosure. For example, while the system components described above may be implemented by hardware devices, they may also be implemented solely by software solutions, such as installing the described system on an existing server or mobile device.
Likewise, it should be noted that in order to simplify the presentation disclosed in this specification, and thereby aid in understanding one or more embodiments, various features are sometimes grouped together in a single embodiment, figure, or description thereof for the purpose of the preceding description of the embodiments of the present specification. This method of disclosure does not imply that the subject matter of the present description requires more features than are set forth in the claims. Indeed, less than all of the features of a single embodiment disclosed above.
In some embodiments, numbers describing the components, number of attributes are used, it being understood that such numbers being used in the description of embodiments are modified in some examples by the modifier "about," approximately, "or" substantially. Unless otherwise indicated, "about," "approximately," or "substantially" indicate that the number allows for a 20% variation. Accordingly, in some embodiments, numerical parameters set forth in the specification and claims are approximations that may vary depending upon the desired properties sought to be obtained by the individual embodiments. In some embodiments, the numerical parameters should take into account the specified significant digits and employ a method for preserving the general number of digits. Although the numerical ranges and parameters set forth herein are approximations that may be employed in some embodiments to confirm the breadth of the range, in particular embodiments, the setting of such numerical values is as precise as possible.
Each patent, patent application publication, and other material, such as articles, books, specifications, publications, documents, etc., referred to in this specification is incorporated herein by reference in its entirety. Except for application history documents that are inconsistent or conflicting with the content of this specification, documents that are currently or later attached to this specification in which the broadest scope of the claims to this specification is limited are also. It is noted that, if the description, definition, and/or use of a term in an attached material in this specification does not conform to or conflict with what is described in this specification, the description, definition, and/or use of the term in this specification controls.
Finally, it should be understood that the embodiments described in this specification are merely illustrative of the principles of the embodiments of this specification. Other variations are possible within the scope of this description. Thus, by way of example, and not limitation, alternative configurations of embodiments of the present specification may be considered as consistent with the teachings of the present specification. Accordingly, the embodiments of the present specification are not limited to only the embodiments explicitly described and depicted in the present specification.
Claims (10)
1. A machine learning based concentrator verification method, the method comprising:
acquiring a network map of the concentrator, wherein the network map comprises nodes and edges;
Determining, based on the network map, relevant users and relevant edges from the nodes and the edges, the relevant users referring to users who have data/signal transmissions with the current concentrator that exceed a traffic threshold, the relevant edges referring to edges through which the current data/signal transmissions exceed the traffic threshold, comprising:
Acquiring node attributes of a first user based on the network map, determining online probability of the first user at a preset time point or time period through an online probability model based on user characteristics of the first user, and determining the related user and the related edge in the preset time point or time period based on the online probability of the first user, wherein the online probability model is a machine learning model, and the first user is a user directly connected with the concentrator in the network map;
acquiring actual working data of the concentrator, the related users and the related edges;
and determining a reaction mechanism in response to the actual working data not meeting a preset condition.
2. The concentrator verification method of claim 1, wherein the predetermined condition includes at least one of a value of the operational data being within a dynamic fluctuation range, a value of the operational data being less than a dynamic threshold value, and a difference value, and wherein determining the reaction mechanism in response to the actual operational data not meeting the predetermined condition comprises:
determining ideal operational data for the concentrator, the associated user, and the associated edge at a set point/segment in time;
And responding to the difference between the actual working data and the ideal working data exceeding the difference value, and making a corresponding reaction mechanism.
3. The concentrator calibration method of claim 1, wherein the preset condition is dynamically adjusted as a target preset condition.
4. A concentrator verification method as claimed in claim 3 wherein said dynamically adjusting said preset condition as a target preset condition comprises:
acquiring history increase and decrease data;
predicting the joining quantity of at least one group of new users, the increasing possibility of the joining quantity, the cancellation quantity of at least one group of old users and the cancellation possibility of the new users through a prediction map model based on the historical increasing and decreasing data and the network map;
Determining a final joining number of the new user and a final cancellation number of the old user based on the joining number of the new user and the increased likelihood thereof, the cancellation number of the old user and the cancellation likelihood thereof;
determining a dynamic adjustment coefficient based on the final joining number of the new user, the corresponding increased likelihood and/or the final cancellation number of the old user, the corresponding cancellation likelihood;
and re-determining the preset condition based on the preset condition and the dynamic adjustment coefficient, and taking the preset condition as a target preset condition.
5. A machine learning based concentrator verification system, the system comprising:
The first acquisition module is used for acquiring a network map where the concentrator is located, and the network map comprises nodes and edges;
A correlation determination module, configured to determine, based on the network map, a correlation user and a correlation edge from the node and the edge, where the correlation user refers to a user that has data/signal transmission with a current concentrator that exceeds a traffic threshold, and the correlation edge refers to an edge through which a current data/signal transmission amount exceeds the traffic threshold, and includes:
Acquiring node attributes of a first user based on the network map, determining online probability of the first user at a preset time point or time period through an online probability model based on user characteristics of the first user, and determining the related user and the related edge in the preset time point or time period based on the online probability of the first user, wherein the online probability model is a machine learning model, and the first user is a user directly connected with the concentrator in the network map;
The second acquisition module is used for acquiring actual working data of the concentrator, the related users and the related edges;
and the reaction mechanism determining module is used for determining a reaction mechanism in response to the fact that the actual working data does not meet the preset condition.
6. The concentrator verification system of claim 5, wherein the preset condition includes at least one of a value of the operational data being within a dynamic fluctuation range, a value of the operational data being less than a dynamic threshold, and a difference value, the reaction mechanism determination module further comprising:
An ideal module for determining ideal working data of the concentrator, the relevant users and the relevant edges at a set point/segment in time;
And the difference module is used for responding to the fact that the difference between the actual working data and the ideal working data exceeds the difference value and making a corresponding reaction mechanism.
7. The concentrator verification system of claim 5, wherein the preset condition is dynamically adjusted as a target preset condition.
8. The concentrator verification system of claim 7, wherein the reaction mechanism determination module comprises:
The third acquisition module is used for acquiring history increase and decrease data;
An increase and decrease user determination module for predicting the number of new users added, the increase probability thereof, the number of old users logged out, and the logging out probability thereof of at least one group through a prediction map model based on the history increase and decrease data and the network map;
A final determining module for determining a final joining number of the new user and a final cancellation number of the old user based on the joining number of the new user and the increasing possibility thereof, the cancellation number of the old user and the cancellation possibility thereof;
The coefficient determining module is used for determining a dynamic adjustment coefficient based on the final joining quantity of the new user, the corresponding increasing possibility, the final cancellation quantity of the old user and the corresponding cancellation possibility;
and the target condition determining module is used for redetermining the preset condition based on the preset condition and the dynamic adjustment coefficient to serve as a target preset condition.
9. A concentrator verification device, the device comprising a processor and a memory; the memory for storing instructions that, when executed by the processor, cause the apparatus to implement the concentrator verification method of any one of claims 1 to 4.
10. A computer readable storage medium storing computer instructions which, when read by a computer in the storage medium, operate the concentrator verification method of any one of claims 1 to 4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410432677.6A CN118034990B (en) | 2024-04-11 | 2024-04-11 | Concentrator verification method and system based on machine learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410432677.6A CN118034990B (en) | 2024-04-11 | 2024-04-11 | Concentrator verification method and system based on machine learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN118034990A CN118034990A (en) | 2024-05-14 |
| CN118034990B true CN118034990B (en) | 2024-06-18 |
Family
ID=90984427
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410432677.6A Active CN118034990B (en) | 2024-04-11 | 2024-04-11 | Concentrator verification method and system based on machine learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN118034990B (en) |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110311709A (en) * | 2019-06-10 | 2019-10-08 | 国网浙江省电力有限公司嘉兴供电公司 | Power information acquisition system fault distinguishing method |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| GB201616027D0 (en) * | 2016-09-21 | 2016-11-02 | Rolls-Royce Ltd | Electrical fault detection method |
| FR3069327B1 (en) * | 2017-07-19 | 2019-08-30 | Sagemcom Energy & Telecom Sas | METHOD FOR COLLECTING INFORMATION OF AN ELECTRICAL COUNTER ASSEMBLY |
| CN209088949U (en) * | 2018-11-28 | 2019-07-09 | 国网安徽省电力公司 | Bimodulus communication intelligent concentrator |
| CN112114230A (en) * | 2020-09-18 | 2020-12-22 | 中电装备山东电子有限公司 | Distribution network monitoring system for low-voltage distribution area and distribution area topology identification method |
-
2024
- 2024-04-11 CN CN202410432677.6A patent/CN118034990B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110311709A (en) * | 2019-06-10 | 2019-10-08 | 国网浙江省电力有限公司嘉兴供电公司 | Power information acquisition system fault distinguishing method |
Non-Patent Citations (1)
| Title |
|---|
| "基于22标准集中器的高频采集技术研究";滕培青，谷呈星;《现代信息科技》;20240410;第8卷(第7期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN118034990A (en) | 2024-05-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US9778639B2 (en) | Systems and methods for adaptively updating equipment models | |
| US10444746B2 (en) | Method for managing subsystems of a process plant using a distributed control system | |
| US20020022969A1 (en) | Remote automated customer support for manufacturing equipment | |
| CN110968061B (en) | Equipment fault early warning method and device, storage medium and computer equipment | |
| AU2020309879B2 (en) | Failure probability assessment system and method therefor | |
| EP3163519A1 (en) | Methods for detecting one or more aircraft anomalies and devices thereof | |
| US11830091B2 (en) | Methods for production line electricity management based on an industrial internet of things, systems and storage mediums thereof | |
| JPWO2019116418A1 (en) | Fault analyzer, fault analysis method and fault analysis program | |
| CN111338876B (en) | Decision method, system and storage medium for fault mode and influence analysis | |
| CN113191074A (en) | Machine room power supply parameter detection method for data center | |
| CN112182233B (en) | Knowledge base for storing equipment fault records, and method and system for assisting in positioning equipment faults by using knowledge base | |
| CN113740666B (en) | Method for positioning root fault of storm alarm in power system of data center | |
| CN111473471A (en) | Multi-linkage energy consumption metering method and system | |
| CN118034990B (en) | Concentrator verification method and system based on machine learning | |
| US11392437B1 (en) | Cold start and adaptive server monitor | |
| CN113010394A (en) | Machine room fault detection method for data center | |
| CN114564349B (en) | Monitoring method and device of server, electronic equipment and storage medium | |
| CN105847978B (en) | A kind of formation gathering method and system based on smart television | |
| JP2022037107A (en) | Failure analysis device, failure analysis method, and failure analysis program | |
| CN113507397A (en) | Method for collecting terminal equipment state automatic inspection based on cloud operation and maintenance | |
| CN115994136B (en) | Energy data cleaning method and system based on energy network topological relation | |
| CN113469453B (en) | Elevator evaluation method and elevator evaluation device based on information physical system | |
| CN117387925B (en) | Damper performance continuous test method and system | |
| JP2020004028A (en) | Network device information management device and network device information management method | |
| CN117498918B (en) | Signal receiver control method and system based on satellite signal intensity |
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 |