CN112654060A - Device abnormality detection method and system - Google Patents
Device abnormality detection method and system Download PDFInfo
- Publication number
- CN112654060A CN112654060A CN202011505191.9A CN202011505191A CN112654060A CN 112654060 A CN112654060 A CN 112654060A CN 202011505191 A CN202011505191 A CN 202011505191A CN 112654060 A CN112654060 A CN 112654060A
- Authority
- CN
- China
- Prior art keywords
- devices
- model
- data
- abnormality
- prediction model
- 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.)
- Granted
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W24/00—Supervisory, monitoring or testing arrangements
- H04W24/06—Testing, supervising or monitoring using simulated traffic
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
- H04W4/30—Services specially adapted for particular environments, situations or purposes
- H04W4/38—Services specially adapted for particular environments, situations or purposes for collecting sensor information
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Testing And Monitoring For Control Systems (AREA)
Abstract
The invention provides a device abnormality detection method and system, comprising the following steps: k +1 similar devices, wherein one of said similar devices is a first device and the remaining K of said similar devices are second devices; a collection component that collects data from the K +1 similar devices; a communication component that transmits the data to a cloud platform; the cloud platform processes data, and the cloud platform has the following processing method: at least one group of parameters with normalization exist between the first device and the plurality of second devices, the normalization coefficient of the parameters is solved through the K regression models, and the normalization coefficient is utilized to generate a special prediction model of the first device; a plurality of data items from the first device are input, an abnormality in the state of the first device is monitored and determined using a prediction model dedicated to the first device, and detection information is output when the abnormality is detected.
Description
Technical Field
The present invention relates to a device for detecting abnormality of one or more states of a device, and more particularly, to a device abnormality detecting method and system.
Background
In recent years, in the field of machine learning using artificial intelligence, robots are beginning to autonomously determine their own operating states and detect machine abnormalities with high accuracy. State reference maintenance technology is known, in which a plurality of sensors are installed in an energy conversion device for converting fuel into kinetic energy, thermal energy, or the like, to measure the state of each part of the device, and whether the state of the device is abnormal or not is determined based on sensor data.
Patent document CN102460529B discloses a device abnormality monitoring method and system for solving the problem of data shortage by generating an abnormality determination model of a device for data of a plurality of other similar devices of a detection device and accumulating a model for diagnosing the cause of an abnormality by abnormality instances of the plurality of other similar devices.
However, the following problems still remain with the method disclosed in patent document CN 102460529B:
1. data volume is deficient: the patent document CN102460529B uses data of a plurality of similar devices to generate an abnormality determination model to solve the problem of data shortage, but the conditions that the similar devices limited by the abnormality determination model need to meet are too harsh, and there is a limitation in solving the problem of data shortage by this method.
2. Abnormal cases learn slowly: since the probability of abnormality of the apparatus is relatively low, when there is little or no abnormality, learning of a judgment model of the apparatus or the like is not advanced, that is, when a cause diagnosis model of an abnormal state is further generated, a model for comparing and referring to the abnormality diagnosis in the future is stored. However, if an abnormality rarely occurs, the accumulation of the diagnostic model may not progress.
Therefore, it is necessary to design an anomaly detection system and an anomaly detection method to solve the above problems.
Disclosure of Invention
An object of the present invention is to provide a device abnormality detection method for performing abnormality processing for detecting and determining a state of at least one device among K +1 devices, which are to be processed by using similar K +1 devices, based on a plurality of data items of the respective devices obtained by measuring states of the K +1 devices using sensors, the method comprising:
one first device T of the K +1 devices is a target of detection and determination, and the remaining K second devices are targets of data for acquiring a dedicated determination model for generating the first device T, the dedicated determination model for the first device T being used for monitoring and determination of the first device T, data generated by the K +1 devices being all transmitted to a cloud platform, the cloud platform processing the data in accordance with a device abnormality detection method;
the device abnormality detection method includes the steps of:
a preprocessing step of normalizing the data items generated by at least one of the K second devices;
a first step of generating K regression models of K second devices as individual device-specific prediction models of the respective second devices and a common global prediction model of similar devices from a plurality of data items of the K second devices in a normal state, the plurality of data items including normalized data items, and generating a specific prediction model of the first device; and
a second step of inputting a plurality of data items from the first device in units of a predetermined time, monitoring and determining an abnormality in the state of the first device using a prediction model dedicated to the first device, and outputting detection information when the abnormality is detected;
wherein K is a natural number greater than 2
Preferably, the normalization process includes the steps of:
solving coefficients and intercepts of regression expressions of the K regression models according to historical data of the K second devices;
and solving a plurality of groups of measurement normalization coefficients according to the coefficients and the intercept, and carrying out normalization processing on the measurement normalization coefficients to obtain a prediction normalization coefficient.
Preferably, generating the first device-specific predictive model comprises the steps of: and generating a similar device common overall prediction model according to the feature item value or the setting environment measured value of each of the K second devices, substituting the prediction normalization coefficient into the similar device common overall prediction model, and acquiring the coefficient and intercept of the regression expression of the regression model of the first device, thereby generating the first device-specific prediction model.
Preferably, the preprocessing step includes a step of classifying the explanatory variables into collinear explanatory variables having a large cross-correlation strength and independent explanatory variables other than the collinear explanatory variables; in the step of generating the K regression models, a set of regression models is generated for each of the collinearity explanatory variables, and the target variable is predicted from the collinearity explanatory variable and the other independent explanatory variables.
Preferably, the second step includes a step of calculating a group deviation degree in which one or more deviation degrees obtained for each of the collinearity explanatory variables are combined, the step of calculating the predicted value of the target variable includes calculating a group of the predicted values of the target variable by inputting the collinearity explanatory variable and another independent explanatory variable to the first device-specific prediction model, the step of calculating the deviation degree includes calculating a group of the deviation degrees for the group of the predicted values, and the step of detecting an abnormality of the first device includes comparing a group deviation degree in which one or more deviation degrees included in the group of the deviation degrees are combined with a threshold value, thereby detecting the abnormality of the first device.
Provided is a device abnormality detection system including: similar K +1 devices as processing targets, an abnormality process of detecting and determining a state of at least one device among the K +1 devices from a plurality of data items of the respective devices obtained by measuring states of the K +1 devices using sensors;
a collection component configured to collect data from the K +1 similar devices;
a communication component configured to send the data to a cloud platform;
a cloud platform for receiving the data and processing the data to implement device anomaly detection, the cloud platform having:
an individual device-specific determination model generation unit that performs the following processing: generating K regression models of K second devices as individual device-specific prediction models of the respective second devices based on a plurality of data items when the respective second devices of the K second devices are normal;
an overall device determination model generation unit that performs the following processing: a group of parameters with normalization exists between the first device and the plurality of second devices, normalized coefficients of the parameters are solved through the K regression models, and a common overall prediction model of the similar devices for predicting the coefficient and intercept of the regression model of each co-linear explanatory variable of the group of regression models is generated according to the characteristic item values or environment measurement values of the K second devices and the normalized coefficients;
a determination model storage unit that selects a cloud storage partition for storing the individual device-specific prediction model and the similar device-shared global prediction model, based on the K individual device-specific prediction models and the similar device-shared global prediction model;
an abnormality determination module that performs the following processing: a plurality of data items from the first device are input in units of a predetermined time, an abnormality in the state of the first device is monitored and determined using a prediction model dedicated to the first device, and detection information is output when the abnormality is detected.
Preferably, the device abnormality detection system further includes an abnormality cause diagnosis, the abnormality cause diagnosis including the steps of:
s501: when the anomaly detection module detects system anomaly, the anomaly diagnosis module starts to collect data of each sensor uploaded to the cloud end by the anomaly time period device, and establishes an anomaly map model according to the obtained data;
s502: extracting normal graph models and abnormal models stored in other devices from the cloud platform for differential processing;
s503: by comparing with the abnormal model stored in the cloud platform, if the similar device has the abnormality, the abnormal model can be directly compared with the cloud data, and if the similar device has no consistent model, the reason of the abnormality needs to be manually determined, and the diagnosis is uploaded to the cloud end for learning.
Preferably, the communication component provides a software mechanism to dynamically link the underlying cloud gateway.
Preferably, the system further comprises a notification component configured to send a notification of the abnormality detection result and the abnormality cause diagnosis result of the system to a client device.
Preferably, the cloud platform processing data includes: the plurality of data items of each of the K second devices are classified into a target variable and an explanatory variable other than the target variable in the regression analysis.
The detection of the device using the abnormality detection system as described above has the following advantages:
1. the cloud platform is utilized to cancel the limitation of data regions, and the data available range is widened;
2. the cloud agent provides a software mechanism to dynamically link the basic cloud gateway, so that the remote monitoring service is expanded under the condition of not forcing the complete reprogramming of the application and the interface thereof;
3. the normalization coefficient is utilized to enlarge the equipment range of available data and further solve the problem of data shortage.
Drawings
FIG. 1 is a schematic view of an anomaly detection system;
FIG. 2 is a schematic view of an individual device determination model generation section;
FIG. 3 is a flow chart of individual device decision model generation;
FIG. 4 is a flowchart of overall device decision model generation;
FIG. 5 is a flow diagram of an anomaly determination module;
fig. 6 is an overview of a cloud computing architecture.
Detailed Description
The present invention will be described in further detail below by way of examples with reference to the accompanying drawings, which are illustrative of the present invention and are not to be construed as limiting the present invention.
As shown in fig. 1, the present invention provides an abnormality detection system including: the off-line K +1 similar devices are processed, one first device T of the K +1 devices is a target of detection and determination, and the remaining K second devices are data targets for acquiring a dedicated determination model for generating the first device T.
As a specific embodiment, the system shows 3 similar apparatuses as the second apparatus, and a new device T to be detected as the first apparatus, and the system further includes: and an individual device determination model generation unit 100, a total device determination model generation unit 200, a determination model storage unit 300, and an abnormality determination module 400, which are configured on the cloud platform 504.
The system of fig. 1 shows 3 similar devices, but this is only a specific exemplary embodiment, and the number of similar devices in the system is not limited.
As shown in fig. 1, the individual device determination model generation unit collects data of each sensor of the similar devices 1, 2, and 3 to generate an individual device-specific determination model.
As shown in fig. 2, the individual device determination model generation unit includes: a state data item classification unit 101, an individual device-specific prediction model construction unit 102, and a group prediction deviation degree calculation model construction unit 103.
The processing flow of the individual device determination model generation unit 100 is shown in fig. 3, and the specific steps are as follows:
s101: multiple sensors (S) for collecting processing object device1～Sn) Number of state measurementsAccording to the DS, and stores it.
S102: the state data item classification unit 101 classifies the data items of the state measurement data DS into a target variable (Y), a collinearity explanatory variable (XC), and an independence explanatory variable (XD).
S103: the individual device-specific prediction model constructing unit 102 combines the collinear explanatory variable (XC) and the independent explanatory variable (XD), and generates an individual device-specific prediction model for the target variable (Y) for each collinear explanatory variable (XC).
An example in which the individual-apparatus-specific prediction model is expressed in a linear regression form can be expressed by formula (1), in which YpredFor the predicted value of the objective variable Y, XcIs an actual measured value, X, of a co-linear explanatory variable (XC)DIs an actual measured value of an independent explanatory variable (XD), acIs A Coefficient (AC), a, for a collinearity explanatory variable (XC)DB is the intercept (B) of the prediction formula for the target variable (Y), t is the data collection time (sample number), p is the identification symbol of the collinear explanatory variable data item (XC), q is the identification symbol of the independent explanatory variable data item (XD), and m is the identification symbol of the device 1.
In S104, the group prediction deviation degree calculation model for calculating the group prediction deviation degree (EG) which is a combination of the deviation degrees (prediction deviation degrees: EY) of the respective predicted values (PdY) and the actual measurement values (Y) of the plurality of prediction models is constructed by the group prediction deviation degree calculation model construction unit 103.
The prediction deviation degree (EY) can be expressed by formula (2). Where EY is the deviation value (predicted deviation degree) of the objective variable (Y), YmeasMeasured value of the object variable (Y), YpredThe predicted value of the objective variable (Y).
EY(t,p,m)＝Ypred(t,p,m)-Ymeans(t,m) (2)
The group prediction deviation degree calculation Model (MG) when the group prediction deviation degree (EG) is taken as the sum of the standard values of the respective prediction deviation degrees (EY) can be expressed by formula 3,
EG(t,m)＝∫{EY(t,p,m)} (3)
in S105, the constructed group prediction deviation degree calculation Model (MG) is stored in the determination model storage unit 300 together with the data of the individual device-specific prediction model as data of an individual device-specific determination model.
The overall device determination model generation unit 200 has a process flow as shown in fig. 4, and includes the following specific steps:
s201: a plurality of individual device-specific determination models including a prediction model for a collinearity explanatory variable are collected from the determination model storage unit 300.
S202: data (DD) representing differences in characteristics or configurations, etc. between similar devices is collected for each device.
S203: a common prediction model (MM) for similar devices is generated, and prediction model parameters included in the collected individual device-specific judgment models are predicted using feature configuration data (DD) between similar devices.
At least one group of normalization parameters exist between the equipment T to be detected and a plurality of similar devices, the normalization coefficient of the parameters is solved through the K regression models, the normalization coefficient is brought into the special prediction model for generating the equipment T, and the specific steps for generating the special prediction model for the equipment T are as follows:
for similar devices with similar characteristics and structures, since the power and each parameter generally have a direct or inverse relationship, the normalized coefficient w is measured for each device under the condition of the same power as the deviceMeasuringThe value of (a) is 1, and a under each power condition can be obtained by early data learning for machines with different powersc、aDB measured normalization coefficient w of the global coefficient1 measurement、w2 measurement of、w3 measurement of。
Therefore, by the preceding data learning, a large number of prediction model parameters included in the individual device-specific determination model are collected, and the measurement normalization coefficient w is obtained from the formula (4)Measuring。
A in formula (4)c1、ac2Coefficients of collinearity explanatory variables calculated for two different power states, aD1、aD2Coefficients of independent explanatory variables calculated for two different power states, b1、b2The intercept of the formula is predicted for the target variable calculated at two different power states.
Through least square method or other optimization algorithm, a plurality of groups w are calculated according to a large amount of early data learning1 measurement、w2 measurement of、w3 measurement ofTo w1 measurement、w2 measurement of、w3 measurement ofThe value of (2) is processed by a formula (5) to obtain w under the difference multiple of unit powerMeasuringIncrease value V ofiGet ViMean value of
wi side test/(P1/P2)＝vi (5)
In building the dedicated predictive model for the device T,
multiplied by saidAnd obtaining the prediction normalization coefficient w by the unit power multiple of the equipment T, and finally generating a common overall prediction model of the similar device according to each obtained prediction normalization coefficient w.
A similar device-shared overall prediction model, which represents the coefficients (AC) of the co-linear explanatory variable (XC) for the parameters of the prediction model of equation (1), i.e., the first term, can be represented by equation (6).
A similar device-shared overall prediction model in which the coefficient (AD) of the independent explanatory variable (XD) for the second term is expressed by a linear regression expression can be expressed by formula (7).
A similar device-shared overall prediction model, in which the intercept (B) for the third term is expressed in a linear regression form, can be expressed by equation (8).
The above formulas (6) to (8) are shown, wherein αC、αD、α0Is ac、aDB overall prediction coefficient, betaC、βD、β0Is ac、aDB, and the total predicted intercept, φ is the value of the characteristic item variable of the device (representing whether the device is the same type, and is expressed by 0 and 1, that is, used to determine whether the device belongs to the unified type of device data).
S204: the generated similar devices share the overall prediction model and are stored in the judgment model storage unit.
The determination model storage 300 selects a cloud storage partition for storing the individual device-specific prediction model and the similar device-common overall prediction model, based on the individual device-specific prediction model and the similar device-common overall prediction model.
The abnormality determination module is configured to detect whether the device T to be detected is abnormal, and a specific detection flow is shown in fig. 5, including the steps of:
s401: and uploading the sensor data of the equipment T to be detected to the cloud platform.
S402: reading the common global prediction model of the similar devices from the judgment model storage part, a to be readc、aDB into a decision model specific to the individual device, i.e. ac、aDB substituting the formula (1) to generate a prediction model specific to the individual device T for the plant T, and obtaining a group prediction deviation degree calculation model by using the formula (3), wherein the prediction model specific to the individual device T and the group prediction deviation degree calculation model are used as a judgment model specific to the individual device T.
S403: and comparing the group of prediction deviation degrees with a threshold value, performing abnormity judgment according to a comparison result, and outputting a maintenance instruction if the abnormity exists.
The system can also realize the diagnosis of the abnormal reason, and the realized simple flow is as follows:
s501: when the anomaly detection module detects system anomaly, the anomaly diagnosis module starts to collect data of each sensor uploaded to the cloud end by the anomaly time period device, and establishes an anomaly map model according to the obtained data;
s502: extracting normal graph models and abnormal models stored in other devices at the cloud end for differential processing;
s503: by comparing with the abnormal model stored in the cloud, if the similar device has the abnormality, the abnormal model can be directly compared with cloud data, and if the similar device has no consistent model, the reason of the abnormality needs to be manually determined, and the diagnosis is uploaded to the cloud for learning. In the process of establishing the cloud anomaly judgment model, the similar equipment analyzes the anomaly through the method, gradually learns and perfects various anomaly modes, and obtains a more complete diagnosis model.
The diagnosis method of the cause of abnormality described above has been disclosed in the prior art (patent document CN102460529B), and thus the present invention will not be described in detail, but those of ordinary skill in the art will recognize that the diagnosis method of the cause of abnormality described in the present invention is achievable.
In addition, the anomaly detection system provided by the invention is based on the cloud platform, the range of available data is expanded by utilizing the advantages of the cloud platform, and the regional limitation when local data is utilized is removed, so that the problem of data shortage is further solved.
As shown in FIG. 6, a cloud computing profile is provided as an exemplary architecture in which a large number of industrial assets are located on a plant network 524 in a manufacturing environment. These assets can include industrial controllers 510 and 520 that monitor and control I/ O devices 512 and 522, a data server 514, a motor drive 516, and a remote I/O interface 518 that remotely connects a group I/O device 526 to one or more of the industrial controllers 510 or 520.
Also located at the plant facility is a cloud agent 508 that provides for the collection, packaging of the underlying data and transmission of industrial data generated by the industrial assets, the cloud agent 508 acting as a common gateway for collecting data items from various industrial assets on the plant network 524, the cloud agent including a collection component, a communication component, and a notification component, the cloud agent packaging the collected data via a universal unified data packaging model that moves the underlying data to the cloud platform 504 via the internet 506. Once the packaged data has been provided to cloud platform 504, the data may be retrieved or viewed from remote monitoring center 502, and notifications may also be sent to the client. Cloud agent 508 provides a software mechanism to dynamically link the underlying cloud gateways so that the anomaly detection system can arbitrarily add devices to be detected without making changes to the system.
Although the preferred embodiments of the present invention have been disclosed for illustrative purposes, those skilled in the art will appreciate that various modifications, additions and substitutions are possible, without departing from the scope and spirit of the invention as disclosed in the accompanying claims.
Claims (10)
1. An apparatus abnormality detection method for performing abnormality processing for detecting and determining a state of at least one apparatus among K +1 apparatuses, based on a plurality of data items of the respective apparatuses obtained by measuring states of the K +1 apparatuses using sensors, with similar K +1 apparatuses as processing targets, characterized in that:
one first device T of the K +1 devices is a target of detection and determination, and the remaining K second devices are targets of data for acquiring a dedicated determination model for generating the first device T, the dedicated determination model for the first device T being used for monitoring and determination of the first device T, data generated by the K +1 devices being all transmitted to a cloud platform, the cloud platform processing the data in accordance with a device abnormality detection method;
the device abnormality detection method includes the steps of:
a preprocessing step of normalizing the data items generated by at least one of the K second devices;
a first step of generating K regression models of the K second devices as a common global prediction model for the individual device-specific prediction models of the respective second devices from a plurality of data items including normalized data items when the K second devices are normal, and generating a prediction model specific to the first device; and
a second step of inputting a plurality of data items from the first device in units of a predetermined time, monitoring and determining an abnormality in the state of the first device using a prediction model dedicated to the first device, and outputting detection information when the abnormality is detected;
wherein K is a natural number greater than 2.
2. The apparatus abnormality detection method according to claim 1, characterized in that: the normalization process includes the steps of:
solving coefficients and intercepts of regression expressions of the K regression models according to historical data of the K second devices;
and solving a plurality of groups of measurement normalization coefficients according to the coefficients and the intercept, and carrying out normalization processing on the measurement normalization coefficients to obtain a prediction normalization coefficient.
3. The apparatus abnormality detection method according to claim 2, characterized in that: generating the first device-specific predictive model comprises the steps of: and generating a similar device common overall prediction model according to the feature item value or the setting environment measured value of each of the K second devices, substituting the prediction normalization coefficient into the similar device common overall prediction model, and acquiring the coefficient and intercept of the regression expression of the regression model of the first device, thereby generating the first device-specific prediction model.
4. The apparatus abnormality detection method according to claim 1, characterized in that: the preprocessing step includes a step of classifying the explanatory variables into collinear explanatory variables having a large cross-correlation strength and independent explanatory variables other than the collinear explanatory variables; in the step of generating the K regression models, a set of regression models is generated for each of the collinearity explanatory variables, and the target variable is predicted from the collinearity explanatory variable and the other independent explanatory variables.
5. The apparatus abnormality detection method according to claim 1, characterized in that: the second step includes a step of calculating a group deviation degree in which one or more deviation degrees obtained for each of the collinear explanatory variables are combined, the step of calculating the predicted value of the target variable includes calculating a group of the predicted values of the target variable by inputting the collinear explanatory variable and other independent explanatory variables to the first device-specific prediction model, the step of calculating the deviation degree includes calculating a group of deviation degrees for the group of the predicted values, and the step of detecting an abnormality of the first device includes comparing a group deviation degree in which one or more deviation degrees included in the group of deviation degrees are combined with a threshold value to detect an abnormality of the first device.
6. A system for device anomaly detection, comprising:
similar K +1 devices as processing targets, an abnormality process of detecting and determining a state of at least one device among the K +1 devices from a plurality of data items of the respective devices obtained by measuring states of the K +1 devices using sensors;
a collection component configured to collect data from the K +1 similar devices;
a communication component configured to send the data to a cloud platform;
a cloud platform for receiving the data and processing the data to implement device anomaly detection, the cloud platform having:
an individual device-specific determination model generation unit that performs the following processing: generating K regression models of K second devices as individual device-specific prediction models of the respective second devices based on a plurality of data items when the respective second devices of the K second devices are normal;
an overall device determination model generation unit that performs the following processing: a group of parameters with normalization exists between the first device and the plurality of second devices, normalized coefficients of the parameters are solved through the K regression models, and a common overall prediction model of the similar devices for predicting the coefficient and intercept of the regression model of each co-linear explanatory variable of the group of regression models is generated according to the characteristic item values or environment measurement values of the K second devices and the normalized coefficients;
a determination model storage unit that selects a cloud storage partition for storing the individual device-specific prediction model and the similar device-shared global prediction model, based on the K individual device-specific prediction models and the similar device-shared global prediction model;
an abnormality determination module that performs the following processing: a plurality of data items from the first device are input in units of a predetermined time, an abnormality in the state of the first device is monitored and determined using a prediction model dedicated to the first device, and detection information is output when the abnormality is detected.
7. The system for device anomaly detection according to claim 6, wherein: the abnormality detection system further includes an abnormality cause diagnosis including the steps of:
s501: when the anomaly detection module detects system anomaly, the anomaly diagnosis module starts to collect data of each sensor uploaded to the cloud end by the anomaly time period device, and establishes an anomaly map model according to the obtained data;
s502: extracting a normal graph model and an abnormal model stored in the second device from the cloud platform for differential processing;
s503: by comparing with the abnormal model stored in the cloud platform, if the similar device has the abnormality, the abnormal model can be directly compared with the cloud data, and if the similar device has no consistent model, the abnormal reason needs to be manually determined, and the diagnosis is uploaded to the cloud platform for learning.
8. The system for device anomaly detection according to claim 6, wherein: the communication component provides a software mechanism to dynamically link the underlying cloud gateway.
9. A system for device anomaly detection according to claim 7, characterized in that: the system also includes a notification component configured to send a notification of the system's anomaly detection results and anomaly cause diagnostic results to a client device.
10. The system for device anomaly detection according to claim 6, wherein: the cloud platform processing data comprises: the plurality of data items of each of the K second devices are classified into a target variable and an explanatory variable other than the target variable in the regression analysis.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011505191.9A CN112654060B (en) | 2020-12-18 | 2020-12-18 | Device abnormity detection method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011505191.9A CN112654060B (en) | 2020-12-18 | 2020-12-18 | Device abnormity detection method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112654060A true CN112654060A (en) | 2021-04-13 |
| CN112654060B CN112654060B (en) | 2023-03-24 |
Family
ID=75354992
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011505191.9A Active CN112654060B (en) | 2020-12-18 | 2020-12-18 | Device abnormity detection method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112654060B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102460529A (en) * | 2009-06-11 | 2012-05-16 | 株式会社日立制作所 | Device abnormality monitoring method and system |
| CN105071983A (en) * | 2015-07-16 | 2015-11-18 | 清华大学 | Abnormal load detection method for cloud calculation on-line business |
| KR20160069444A (en) * | 2014-12-08 | 2016-06-16 | 엔트릭스 주식회사 | Quality of service monitoring system and method for cloud streaming service, and computer readable medium having computer program recorded |
| CN108604360A (en) * | 2016-05-04 | 2018-09-28 | 斗山重工业建设有限公司 | Facility method for monitoring abnormality and its system |
| CN110353725A (en) * | 2019-07-10 | 2019-10-22 | 东南大学 | A kind of heart sound acquisition and analysis system and method based on cloud framework |
| CN111914873A (en) * | 2020-06-05 | 2020-11-10 | 华南理工大学 | Two-stage cloud server unsupervised anomaly prediction method |
-
2020
- 2020-12-18 CN CN202011505191.9A patent/CN112654060B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102460529A (en) * | 2009-06-11 | 2012-05-16 | 株式会社日立制作所 | Device abnormality monitoring method and system |
| KR20160069444A (en) * | 2014-12-08 | 2016-06-16 | 엔트릭스 주식회사 | Quality of service monitoring system and method for cloud streaming service, and computer readable medium having computer program recorded |
| CN105071983A (en) * | 2015-07-16 | 2015-11-18 | 清华大学 | Abnormal load detection method for cloud calculation on-line business |
| CN108604360A (en) * | 2016-05-04 | 2018-09-28 | 斗山重工业建设有限公司 | Facility method for monitoring abnormality and its system |
| CN110353725A (en) * | 2019-07-10 | 2019-10-22 | 东南大学 | A kind of heart sound acquisition and analysis system and method based on cloud framework |
| CN111914873A (en) * | 2020-06-05 | 2020-11-10 | 华南理工大学 | Two-stage cloud server unsupervised anomaly prediction method |
Non-Patent Citations (2)
| Title |
|---|
| 徐静: "基于云计算的数据异常智能检测技术研究", 《自动化与仪器仪表》 * |
| 曾志华等: "云计算环境下频繁出现异常数据挖掘方法研究", 《计算机仿真》 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112654060B (en) | 2023-03-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Rezamand et al. | Critical wind turbine components prognostics: A comprehensive review | |
| JP7017861B2 (en) | Anomaly detection system and anomaly detection method | |
| CN102460529B (en) | Device abnormality monitoring method and system | |
| CN104699037B (en) | Plant control system, controller, manager and plant information processing method | |
| US20110020122A1 (en) | Integrated condition based maintenance system for wind turbines | |
| Udo et al. | Data-driven predictive maintenance of wind turbine based on SCADA data | |
| JP2022524244A (en) | Predictive classification of future behavior | |
| JP4276623B2 (en) | Technical equipment monitoring apparatus and method | |
| EP3638900A1 (en) | Independent monitoring system for a wind turbine | |
| JP2009053938A (en) | Equipment diagnosing system and equipment-diagnosing method on the basis of multiple model | |
| Mosallam et al. | Component based data-driven prognostics for complex systems: Methodology and applications | |
| Xu et al. | Quantile regression neural network‐based fault detection scheme for wind turbines with application to monitoring a bearing | |
| CN111103137A (en) | Wind turbine gearbox fault diagnosis method based on deep neural network | |
| TWI780434B (en) | Abnormal diagnosis device and method | |
| Mohammed et al. | An IoT and machine learning-based predictive maintenance system for electrical motors | |
| CN112654060B (en) | Device abnormity detection method and system | |
| CN116880402A (en) | Intelligent factory cooperative control system and method thereof | |
| CN116755411A (en) | Industrial equipment data acquisition method and device | |
| Imamverdiyev et al. | Condition monitoring of equipment in oil wells using deep learning | |
| JP2022184619A (en) | Measuring device, measuring method, and monitoring system | |
| Johanson et al. | Condition monitoring for wave energy converters | |
| CN110709789B (en) | Method and apparatus for monitoring the condition of subsystems within a renewable power generation device or microgrid | |
| CN117589444B (en) | Wind driven generator gear box fault diagnosis method based on federal learning | |
| CN117786373B (en) | Equipment operation diagnosis system based on big data corrugated paper processing | |
| KR102420120B1 (en) | Electronic meter measuring device of abnormal condition monitoring system |
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 |