CN104880217B - A kind of fault sensor signal reconstruct method based on the measured value degree of association - Google Patents
A kind of fault sensor signal reconstruct method based on the measured value degree of association Download PDFInfo
- Publication number
- CN104880217B CN104880217B CN201510338144.2A CN201510338144A CN104880217B CN 104880217 B CN104880217 B CN 104880217B CN 201510338144 A CN201510338144 A CN 201510338144A CN 104880217 B CN104880217 B CN 104880217B
- Authority
- CN
- China
- Prior art keywords
- response
- measuring point
- variable
- reconstruct
- formula
- 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
Landscapes
- Testing Or Calibration Of Command Recording Devices (AREA)
Abstract
The present invention provides a kind of fault sensor signal reconstruct method this method based on the measured value degree of association and comprised the following steps：S1, the degree of association responded for fault sensor, measuring point response where when calculating the normal operation of sensor with remaining measuring point；S2, by contrasting degree of association size, determine correlation model set up needed for response variable；S3 and then utilization PLS foundation reconstruct variable and the reconstruction model of response variable, with monitoring structural health conditions measured data, fault sensor response message reconstruct is carried out to fault sensor.The method that the present invention is provided is fine for fault sensor signal reconstruct effect, it will be apparent that reduce reconstructed error, while ensure that the variation tendency of reconstruction value and the variation tendency of actual value are also consistent, the reliability of significant analysis of Integral Structure.
Description
Technical field
The invention belongs to civil engineering works structure health monitoring, field of information processing, and in particular to one kind is closed based on measured value
The fault sensor signal reconstruct method of connection degree.
Background technology
Data acquisition is very important link in structural healthy monitoring system, and the analysis of structural safety performance depends on number
According to the whether reliable of collection.During the actual operation of civil structure health monitoring, because the factors such as ageing equipment can not be kept away
Exempt from there is local sensor failure to cause data distortion, simultaneously because some sensors are difficult to change, in turn result in data
Missing, reduces the reliability of overall data, it is difficult to realizes the global analysis to structure, causes incorrect decision, loses structure prison
The meaning of survey；Structure monitoring can produce substantial amounts of data simultaneously, and useful information is chosen from mass data to fault sensor
It is reconstructed very difficult, and the response of each measuring point has relevance, it is necessary to which the foundation for the response of different measuring points has
Fault message is reconstructed the correlation model of effect.
Recovery of the domestic and foreign scholars to fault data has carried out the research of correlation, and some scholars are the number for solving bridge monitoring
According to problem of dtmf distortion DTMF, the relation of each measuring point Monitoring Data is analyzed by Granger causality, extreme learning machine input variable is chosen
Responded with the big measuring point of distortion data causality, realize the recovery of distortion data, but the prediction of some individuals of this method
Larger error occurs, the variation tendency for data also fails to be consistent, and training result is unstable, and its training result needs
Further improve.Some scholars propose the corrupt data restoration methods based on neutral net, set up the RBF nerve nets of data
Network model, recovers to deflection data, and this method needs mass data to be modeled, while being still easily trapped into local optimum.
Some scholars propose the response prediction method based on Kalman filtering, are set up and predicted by structural model parameter and external drive
Model, by known acceleration measurement predictive displacement value and velocity amplitude, and is verified by frame model to method, but
When this method is directed to the structure of unknown parameter, it is impossible to effectively set up forecast model.Some scholars are directed to the number of wireless senser
The method that data compression sampling recovers according to Research on Problems, generates new data by data linear projections by random matrix and goes forward side by side
Row transmission, the method recovered of being sampled in base station by data compression carries out data recovery, but when initial data is openness poor,
The recovery precision reduction of acceleration information.Some scholars propose the data recovery side based on least square method supporting vector machine time span of forecast
Method, the forecast model based on this method is set up using the time course data of sensor, and extensive by on-line study diagnosis progress signal
It is multiple.But this method is time-consuming few, and due to neutral net restoration methods, but mainly for sensor failure short time sensor
Data reconstruction, for the high sensor of sample rate and realizing that chromic trouble prediction effect is not obvious.
By analyzing domestic and foreign scholars to data recovery and the research of prediction, find current method to setting up Stability Analysis of Structures
Correlation model on still have deficiency, for long-term failure reconfiguration DeGrain, it is necessary to research set up be based on Monitoring Data
The method of correlation model carries out fault data reconstruct.
The content of the invention
In view of this, it is based on the measured value degree of association it is an object of the invention to overcome the deficiencies of the prior art and provide one kind
Fault sensor signal reconstruct method.
To realize object above, the present invention is adopted the following technical scheme that：A kind of failure sensing based on the measured value degree of association
Device signal reconstruct method, this method comprises the following steps：
S1, for fault sensor, measuring point response where when calculating the normal operation of sensor is responded with remaining measuring point
The degree of association；
S2, by contrasting degree of association size, determine correlation model set up needed for response variable；
S3 and then utilization PLS foundation reconstruct variable and the reconstruction model of response variable, with structural health
Measured data is monitored, fault sensor response message reconstruct is carried out to fault sensor.
Further, the association journey between different type response is weighed using Pearson correlation coefficient in the step S1
Degree, it is specific as follows：
If X is structural response matrix of the m measuring point under the load action of n time step；
xi=[xi1 xi2 … xin]T (2)
xiRepresentative sensor measuring point i response；
Note y is response when fault sensor is normally run；xiWith y correlation coefficient ρiFor：
σ in formulaxi, σyFor measuring point i and fault sensor measuring point response variance, Cov (xi, y) it is its covariance, xi(k) and
Y (k) is respectively the response of measuring point i and fault sensor measuring point in kth time step, μxiAnd μyFor measuring point i and fault sensor
Average value of the measuring point under the load action of n time step.
Further, the method that the required response variable of correlation model foundation is determined in described step S2 is specific as follows：
The degree of association matrix P that failure measuring point is responded with remaining measuring point is
P=[ρ1 ρ2 … ρm] (7)
By correlation coefficient ρ in PiAbsolute value sort from big to small, be designated as DP
DP={ (ρ)1 …(ρ)i … (ρ)m} (8)
(ρ)iThe coefficient correlation for being ordered as i-th bit is represented, corresponding measuring point response is (x)i；Then according to coefficient correlation size
Structural response matrix D after sequenceXFor
DX={ (x)1 …(x)i … (x)m} (9)
P measuring point monitoring information of selection is used as response variable XP, regard fault sensor reconfiguration information as reconstruct variable YP,
I.e.
XP={ (x)1 (x)2 … (x)p} (10)
YP={ y } (11)
Further, the specific method for the reconstruction model for reconstructing variable and response variable is set up in described step S3 such as
Under：
Response variable and reconstruct variable are done into standardization,
Response variable XPWith reconstruct variable YPIt is after normalized processing
F0={ y*} (15)
With principal component analytical method, E is obtained0And F0Principal component set T and U be
T={ t1 t2 ... th ... tA} (16)
U={ u1 u2 ... uh ... uA} (17)
With PLS and first principal component, E0, F0Regression equation be
E0=t1p1 T+E1 (18)
F0=t1r1 T+F1 (19)
Regression coefficient vector is：
Take h-th of principal component, Eh-1Regression equation be
Eh-1=thph T+Eh (22)
Fh-1=thrh T+Fh (23)
Wherein
E0H-th of principal component thAnd E0H-th of axle whRelation be
th=Eh-1wh (26)
Wherein, whIt is matrix Eh-1 TFh-1Fh-1 TEh-1Eigenvalue of maximum θh 2Unit character vector.
Analysis mode (18), (19), (22), (43) can be obtained
E0=t1p1 T+t2p2 T+…+thph T+Eh (27)
F0=t1r1 T+t2r2 T+…+thrh T+Fh (28)
Formula (26) substitutes into formula (22) and obtained
Eh=Eh-1(I-whph T) (29)
Formula (26) and (29) substitute into formula (27) and obtained
Formula (30) substitutes into formula (26) and obtained
Formula (31) substitutes into formula (28) and obtained
F0=E0w1 *r1 T+…+E0wh *rh T+Fh (33)
Number h when taking the principal component to meet Cross gain modulation principle, now FhLevel off to 0, can be neglected.
Then
F0=E0(w1 *r1 T+…+wh *rh T) (34)
Formula (14), (15) substitute into formula (34) and obtained
If
Then y*=α1 *(x)1 *+…+αp *(x)p * (37)
Formula (12), (13) substitution formula (37) must reconstruct variable and the reconstruct equation of response variable is
Y=α1(x)1+…+αp(x)p+α0 (38)。
Heretofore described sensor is that stress strain gauge, acceleration transducer, displacement transducer or blast are passed
Sensor, anemobiagraph etc. monitor combination more than one or both of civil engineering structure load and the sensor of response.
The present invention uses above technical scheme, for the sensor of failure, calculates normal operation of sensor when institute
The degree of association responded with remaining measuring point is responded in measuring point, that is, reconstructs the degree of association of variable and response variable.By contrasting the degree of association
Size, determines that correlation model sets up required response variable；And then become with PLS foundation reconstruct variable and response
The reconstruction model of amount, with monitoring structural health conditions measured data, fault sensor response message reconstruct is carried out to fault sensor.
The method that the present invention is provided is fine for fault sensor signal reconstruct effect, it will be apparent that reduce reconstructed error, ensures simultaneously
The variation tendency of reconstruction value and the variation tendency of actual value are also consistent, the reliability of significant analysis of Integral Structure.
Brief description of the drawings
Fig. 1 is the fault sensor signal reconstruct method flow diagram of the invention based on the measured value degree of association；
Embodiment
Below in conjunction with accompanying drawing, clear, complete description is carried out to the technical scheme in the embodiment of the present invention.
As shown in figure 1, the invention provides a kind of fault sensor signal reconstruct method based on the measured value degree of association, should
Method comprises the following steps：
S1, for fault sensor, measuring point response where when calculating the normal operation of sensor is responded with remaining measuring point
The degree of association；
S2, by contrasting degree of association size, determine correlation model set up needed for response variable；
S3 and then utilization PLS foundation reconstruct variable and the reconstruction model of response variable, with structural health
Measured data is monitored, fault sensor response message reconstruct is carried out to fault sensor.
Relevance is there is between structural response, Pearson correlation coefficient is the finger for reflecting structural response linear correlation degree
The correlation degree between different type response is weighed using Pearson correlation coefficient in step S1 in mark, the present embodiment, specifically
It is as follows：
If X is structural response matrix of the m measuring point under the load action of n time step；
xi=[xi1 xi2 … xin]T (2)
xiRepresentative sensor measuring point i response；
Note y is response when fault sensor is normally run；xiWith y correlation coefficient ρiFor：
σ in formulaxi, σyFor measuring point i and fault sensor measuring point response variance, Cov (xi, y) it is its covariance, xi(k) and
Y (k) is respectively the response of measuring point i and fault sensor measuring point in kth time step, μxiAnd μyFor measuring point i and fault sensor
Average value of the measuring point under the load action of n time step.
As a preferred embodiment, determining that correlation model sets up required response variable in described step S2
Method is specific as follows：
The degree of association matrix P that failure measuring point is responded with remaining measuring point is
P=[ρ1 ρ2 … ρm] (7)
By correlation coefficient ρ in PiAbsolute value sort from big to small, be designated as DP
DP={ (ρ)1 …(ρ)i … (ρ)m} (8)
(ρ)iThe coefficient correlation for being ordered as i-th bit is represented, corresponding measuring point response is (x)i；Then according to coefficient correlation size
Structural response matrix D after sequenceXFor
DX={ (x)1 …(x)i …(x)m} (9)
P measuring point monitoring information of selection is used as response variable XP, regard fault sensor reconfiguration information as reconstruct variable YP,
I.e.
XP={ (x)1 (x)2 … (x)p} (10)
YP={ y } (11)
Now step S3 is described in further detail, correlation model is to set up reconstruct variable and the relational model of response variable.Partially
Least square method is a kind of multivariate statistics data analysing method, is to reconstruct variable to the modeling method [9] of polyphony dependent variable, can be with
For handling with the insurmountable problem of common multiple regression.In multiple regression, if there is multiplephase in the response variable selected
Guan Xing, the accuracy of parameter Estimation can be reduced using common least square method, increase model error；PLS is utilized
The mode that the principle of multiple linear regression analysis, principal component analysis and canonical correlation analysis is decomposed and screened to data,
Extract to the explanatory most strong generalized variable of reconstruct variable, obtain regression model of the reconstruct variable to the response variable of selection.
So setting up the specific method of reconstruct variable and the reconstruction model of response variable in step S3 in the present embodiment such as
Under：
Response variable and reconstruct variable are done into standardization,
Response variable XPWith reconstruct variable YPIt is after normalized processing
F0={ y*} (15)
With principal component analytical method, E is obtained0And F0Principal component set T and U be
T={ t1 t2 ...th ... tA} (16)
U={ u1 u2 ... uh ... uA} (17)
With PLS and first principal component, E0, F0Regression equation be
E0=t1p1 T+E1 (18)
F0=t1r1 T+F1 (19)
Regression coefficient vector is：
Take h-th of principal component, Eh-1Regression equation be
Eh-1=thph T+Eh (22)
Fh-1=thrh T+Fh (23)
Wherein
E0H-th of principal component thAnd E0H-th of axle whRelation be
th=Eh-1wh (26)
Wherein,whIt is matrix Eh-1 TFh-1Fh-1 TEh-1Eigenvalue of maximum θh 2Unit character vector.
Analysis mode (18), (19), (22), (43) can be obtained
E0=t1p1 T+t2p2 T+…+thph T+Eh (27)
F0=t1r1 T+t2r2 T+…+thrh T+Fh (28)
Formula (26) substitutes into formula (22) and obtained
Eh=Eh-1(I-whph T) (29)
Formula (26) and (29) substitute into formula (27) and obtained
Formula (30) substitutes into formula (26) and obtained
Formula (31) substitutes into formula (28) and obtained
F0=E0w1 *r1 T+…+E0wh *rh T+Fh (33)
Number h when taking the principal component to meet Cross gain modulation principle, now FhLevel off to 0, can be neglected.Then
F0=E0(w1 *r1 T+…+wh *rh T) (34)
Formula (14), (15) substitute into formula (34) and obtained
If
Then y*=α1 *(x)1 *+…+αp *(x)p * (37)
Formula (12), (13) substitution formula (37) must reconstruct variable and the reconstruct equation of response variable is
Y=α1(x)1+…+αp(x)p+α0 (38)。
It should be noted that：Heretofore described sensor is stress strain gauge, acceleration transducer, displacement biography
Sensor or wind pressure sensor, anemobiagraph etc. are monitored more than one or both of civil engineering structure load and the sensor of response
Combination.
The present invention is not limited to above-mentioned preferred forms, and anyone can show that other are various under the enlightenment of the present invention
The product of form, however, make any change in its shape or structure, it is every that there is skill identical or similar to the present application
Art scheme, is within the scope of the present invention.
Claims (3)
1. a kind of fault sensor signal reconstruct method based on the measured value degree of association, it is characterised in that：This method includes following
Step：
S1, for fault sensor, measuring point response where when calculating the normal operation of sensor is associated with the response of remaining measuring point
Degree；
S2, by contrasting degree of association size, determine correlation model set up needed for response variable；
S3 and then utilization PLS foundation reconstruct variable and the reconstruction model of response variable, with monitoring structural health conditions
Measured data, fault sensor response message reconstruct is carried out to fault sensor；
The correlation degree between different type response is weighed using Pearson correlation coefficient in the step S1, it is specific as follows：
If X is structural response matrix of the m measuring point under the load action of n time step；
xi=[xi1 xi2 … xin]T (2)
xiRepresentative sensor measuring point i response；
Note y is response when fault sensor is normally run；xiWith y correlation coefficient ρiFor：
σ in formulaxi, σyFor measuring point i and fault sensor measuring point response variance, Cov (xi, y) it is its covariance, xi(k) with y (k)
The response of respectively measuring point i and fault sensor measuring point in kth time step, μxiAnd μyFor measuring point i and fault sensor measuring point
Average value under the load action of n time step；
The method that the required response variable of correlation model foundation is determined in described step S2 is specific as follows：
The degree of association matrix P that failure measuring point is responded with remaining measuring point is
P=[ρ1 ρ2 … ρm] (7)
By correlation coefficient ρ in PiAbsolute value sort from big to small, be designated as DP
DP={ (ρ)1 … (ρ)i … (ρ)m} (8)
(ρ)iThe coefficient correlation for being ordered as i-th bit is represented, corresponding measuring point response is (x)i；Then sorted according to coefficient correlation size
Structural response matrix D afterwardsXFor
DX={ (x)1 … (x)i … (x)m} (9)
P measuring point monitoring information of selection is used as response variable XP, regard fault sensor reconfiguration information as reconstruct variable YP, i.e.,
XP={ (x)1 (x)2 … (x)p} (10)
YP={ y } (11)
The specific method that reconstruct variable and the reconstruction model of response variable are set up in described step S3 is as follows：
Response variable and reconstruct variable are done into standardization,
Response variable XPWith reconstruct variable YPIt is after normalized processing
F0={ y*} (15)
With principal component analytical method, E is obtained0And F0Principal component set T and U be
T={ t1 t2 ... th ... tA} (16)
U={ u1 u2 ... uh ... uA} (17)
With PLS and first principal component, E0, F0Regression equation be
E0=t1p1 T+E1 (18)
F0=t1r1 T+F1 (19)
Regression coefficient vector is
Take h-th of principal component, Eh-1Regression equation be
Eh-1=thph T+Eh (22)
Fh-1=thrh T+Fh (23)
Wherein
E0H-th of principal component thAnd E0H-th of axle whRelation be
th=Eh-1wh (26)
Wherein, whIt is matrix Eh-1 TFh-1Fh-1 TEh-1Eigenvalue of maximum θh 2Unit character vector, analysis mode (18), (19),
(22), (23) can be obtained
E0=t1p1 T+t2p2 T+…+thph T+Eh (27)
F0=t1r1 T+t2r2 T+…+thrh T+Fh (28)
Formula (26) substitutes into formula (22) and obtained
Eh=Eh-1(I-whph T) (29)
Formula (26) and (29) substitute into formula (27) and obtained
Formula (30) substitutes into formula (26) and obtained
Formula (31) substitutes into formula (28) and obtained
F0=E0w1 *r1 T+…+E0wh *rh T+Fh
(33)
Number h when taking the principal component to meet Cross gain modulation principle, now FhLevel off to 0, can be neglected, then
F0=E0(w1 *r1 T+…+wh *rh T) (34)
Formula (14), (15) substitute into formula (34) and obtained
If
Then y*=α1 *(x)1 *+…+αp *(x)p * (37)
Formula (12), (13) substitution formula (37) must reconstruct variable and the reconstruct equation of response variable is
Y=α1(x)1+…+αp(x)p+α0 (38)。
2. a kind of fault sensor signal reconstruct method based on the measured value degree of association according to claim 1, its feature
It is：Described sensor is stress strain gauge, acceleration transducer, displacement transducer, wind pressure sensor or anemobiagraph
One or both of more than combination.
3. a kind of fault sensor signal reconstruct method based on the measured value degree of association according to claim 1, its feature
It is：Described sensor is monitoring civil engineering structure load and the sensor of response.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510338144.2A CN104880217B (en) | 2015-06-17 | 2015-06-17 | A kind of fault sensor signal reconstruct method based on the measured value degree of association |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510338144.2A CN104880217B (en) | 2015-06-17 | 2015-06-17 | A kind of fault sensor signal reconstruct method based on the measured value degree of association |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104880217A CN104880217A (en) | 2015-09-02 |
| CN104880217B true CN104880217B (en) | 2017-08-15 |
Family
ID=53947803
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510338144.2A Active CN104880217B (en) | 2015-06-17 | 2015-06-17 | A kind of fault sensor signal reconstruct method based on the measured value degree of association |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104880217B (en) |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106372426A (en) * | 2016-08-31 | 2017-02-01 | 郑州航空工业管理学院 | Multi-response parameter optimization method based on principal component analysis and neural network |
| CN108416309B (en) * | 2018-03-14 | 2022-02-18 | 揭阳职业技术学院 | Multi-fault sensing signal reconstruction method for intelligent sensor |
| CN109738220B (en) * | 2019-01-07 | 2021-01-08 | 哈尔滨工业大学（深圳） | Sensor optimal arrangement method based on multi-load working condition structure response correlation |
| CN112179655B (en) * | 2020-08-17 | 2021-07-09 | 中国农业大学 | Turbo generator fault early warning method based on threshold classification |
| CN111982256B (en) * | 2020-08-18 | 2021-11-16 | 上海陆杰电子科技有限公司 | Automobile balance pressure sensor fault early warning method based on multichannel CNN model |
| CN116484306B (en) * | 2023-06-20 | 2023-09-26 | 蘑菇物联技术（深圳）有限公司 | Positioning method and device of abnormal sensor, computer equipment and storage medium |
| CN116698323B (en) * | 2023-08-07 | 2023-10-13 | 四川华腾公路试验检测有限责任公司 | Bridge health monitoring method and system based on PCA and extended Kalman filtering |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101382439B (en) * | 2008-10-27 | 2010-12-29 | 哈尔滨工业大学 | Multi-parameter self-confirming sensor and state self-confirming method thereof |
| CN101477375B (en) * | 2009-01-05 | 2012-01-04 | 东南大学 | Sensor data verification method based on matrix singular values association rules mining |
| CN103198638B (en) * | 2013-04-27 | 2015-11-25 | 上海交通大学 | Based on the Railway Bridge Structural Health Monitoring system of wireless sensor and actor network |
| CN104462770B (en) * | 2014-11-10 | 2017-04-12 | 福州大学 | Signal reconstruction method taking sensor performance degradation into consideration |
| CN105318905B (en) * | 2014-12-30 | 2018-05-25 | 国家电网公司 | A kind of Thermal System of Fossil Fuel-Fired Power Station sensor fault diagnosis method |
-
2015
- 2015-06-17 CN CN201510338144.2A patent/CN104880217B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN104880217A (en) | 2015-09-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104880217B (en) | A kind of fault sensor signal reconstruct method based on the measured value degree of association | |
| CN112033463B (en) | Nuclear power equipment state evaluation and prediction integrated method and system | |
| CN103776480B (en) | Small fault detection method based on repeatedly rolling average and device | |
| CN109489977B (en) | KNN-AdaBoost-based bearing fault diagnosis method | |
| CN111737909B (en) | Structural health monitoring data anomaly identification method based on space-time graph convolutional network | |
| CN109034191B (en) | ELM-based one-dimensional telemetry data abnormal interpretation method | |
| CN108873859B (en) | Bridge type grab ship unloader fault prediction model method based on improved association rule | |
| CN105629958B (en) | A kind of batch process method for diagnosing faults based on sub-period MPCA SVM | |
| CN111946559B (en) | Method for detecting structures of wind turbine foundation and tower | |
| CN105241680A (en) | Health state assessment method for rotary machine based on probability density function | |
| CN105716814A (en) | Real-time monitoring system and method for assessing damage of truss structure | |
| KR100867938B1 (en) | Prediction method for watching performance of power plant measuring instrument by dependent variable similarity and kernel feedback | |
| CN112132430B (en) | Reliability evaluation method and system for distributed state sensor of power distribution main equipment | |
| CN110348150A (en) | A kind of fault detection method based on dependent probability model | |
| CN103103570B (en) | Based on the aluminium cell condition diagnostic method of pivot similarity measure | |
| CN112036042A (en) | Power equipment abnormality detection method and system based on variational modal decomposition | |
| CN112395684A (en) | Intelligent fault diagnosis method for high-speed train running part system | |
| CN109543743A (en) | A kind of refrigeration unit multiple sensor faults diagnosis method based on reconstruction prediction residual | |
| CN115659583A (en) | Point switch fault diagnosis method | |
| Son et al. | Deep learning-based anomaly detection to classify inaccurate data and damaged condition of a cable-stayed bridge | |
| CN114091600A (en) | Data-driven satellite associated fault propagation path identification method and system | |
| CN115238753B (en) | Self-adaptive SHM data cleaning method based on local outlier factor | |
| CN116244596A (en) | Industrial time sequence data anomaly detection method based on TCN and attention mechanism | |
| CN112069621B (en) | Method for predicting residual service life of rolling bearing based on linear reliability index | |
| CN114266013A (en) | Deep learning virtual perception network-based transmission system vibration decoupling method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| EXSB | Decision made by sipo to initiate substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |