CN109376478A - Bridge health monitoring fault data restorative procedure and system - Google Patents
Bridge health monitoring fault data restorative procedure and system Download PDFInfo
- Publication number
- CN109376478A CN109376478A CN201811448903.0A CN201811448903A CN109376478A CN 109376478 A CN109376478 A CN 109376478A CN 201811448903 A CN201811448903 A CN 201811448903A CN 109376478 A CN109376478 A CN 109376478A
- Authority
- CN
- China
- Prior art keywords
- data
- missing
- type
- point
- health monitoring
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 64
- 238000012544 monitoring process Methods 0.000 title claims abstract description 46
- 230000036541 health Effects 0.000 title claims abstract description 45
- 230000008439 repair process Effects 0.000 claims abstract description 22
- 238000012217 deletion Methods 0.000 claims description 54
- 230000037430 deletion Effects 0.000 claims description 54
- 241000728173 Sarima Species 0.000 claims description 28
- 238000012549 training Methods 0.000 claims description 26
- 230000001932 seasonal effect Effects 0.000 claims description 14
- 238000000714 time series forecasting Methods 0.000 claims description 14
- 241001269238 Data Species 0.000 claims description 6
- 238000012545 processing Methods 0.000 abstract description 30
- 238000001514 detection method Methods 0.000 abstract description 7
- 230000008569 process Effects 0.000 description 15
- 238000006467 substitution reaction Methods 0.000 description 4
- 238000007476 Maximum Likelihood Methods 0.000 description 2
- 238000000546 chi-square test Methods 0.000 description 2
- 230000007812 deficiency Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000001744 unit root test Methods 0.000 description 2
- 238000013480 data collection Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000008030 elimination Effects 0.000 description 1
- 238000003379 elimination reaction Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 210000003739 neck Anatomy 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
- 238000000611 regression analysis Methods 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/10—Geometric CAD
- G06F30/13—Architectural design, e.g. computer-aided architectural design [CAAD] related to design of buildings, bridges, landscapes, production plants or roads
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Geometry (AREA)
- General Physics & Mathematics (AREA)
- Computer Hardware Design (AREA)
- Theoretical Computer Science (AREA)
- Civil Engineering (AREA)
- Structural Engineering (AREA)
- Computational Mathematics (AREA)
- Architecture (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a kind of methods of bridge health monitoring fault data reparation, are related to technical field of data processing, the missing data type including judging initial data；Single-point missing data is repaired according to the initial data of judgement；Continuity point missing data is repaired according to the single-point missing data of reparation.The missing data type of initial data is judged using consecutive points detection method；Single-point missing data is repaired using interpolation method, continuity point missing data is repaired using time series method.The invention also discloses a kind of systems of bridge health monitoring fault data reparation.The present invention is based on the realizations of the raw monitored fault data of small sample to repair to missing data, and it is fast to repair speed.
Description
Technical field
The present invention relates to bridge health monitoring data recovery technique fields, and in particular to a kind of bridge health monitoring number of faults
According to restorative procedure and system.
Background technique
Currently, the evaluation of structure safety is almost all based on complete data collection, with regard to currently used several processing missing numbers
According to method in, direct elimination method is a kind of most common method, i.e., directly by the relevant information for missing data occur delete.This
That sample is done the result is that making that treated, data are true data, but some valuable information can also be made to be ignored,
To cause greatly to waste.And the incomplete result to final health assessment system of data of bridge health monitoring system acquisition
It has a huge impact, gently then makes assessment result and virtual condition there is deviation, generation that is heavy then leading to false alarm phenomenon,
So that assessment result is completely opposite with virtual condition.So the presence of deficiency of data is considered in bridge health monitoring system,
And actively take measures make up reparation with very big practical application value.
It is established on the basis of large sample mostly in the restorative procedure of existing common missing data, such as EM
(Expectation Maximization) algorithm and regression analysis etc..Although there is researcher using these methods in some necks
Domain realizes the reparation of missing data, and some methods have also achieved good effect, but the data in these methods are big
It is small often to reach 100M or more.The integrality that not can guarantee huge historical data in a practical situation is gone through using huge
History data are repaired, time-consuming and laborious.
Summary of the invention
In view of the deficiencies in the prior art, the purpose of the present invention is to provide a kind of bridge health monitoring fault datas
Restorative procedure and system, the raw monitored fault data based on small sample realize the reparation to missing data, and it is fast to repair speed.
To achieve the above objectives, the method for bridge health monitoring fault data reparation of the present invention, comprising the following steps:
During acquiring bridge health monitoring fault data, missing data is assigned to null value；
When determining that the type of missing data lacks for single-point, then according to multiple no missing datas before and after missing data, adopt
The data of single-point deletion type are repaired with interpolation method；
It is when determining that the type of missing data lacks for continuity point, then pre- using time series according to one section without missing data
The data of survey method reparation continuity point deletion type.
On the basis of above scheme, the type for determining missing data is single-point missing, comprising the following steps: if certain
The adjacent data of one missing data is no missing data, then determines the type of the missing data for single-point missing.
On the basis of above scheme, the type for determining missing data is continuity point missing, comprising the following steps: if
Continuous multiple data are missing data, then determine the type of these continuous multiple data for continuity point missing.
On the basis of above scheme, first the data of single-point deletion type are repaired, then to continuity point deletion type
Data repaired.
On the basis of above scheme, the data that continuity point deletion type is repaired using time series forecasting, packet
Include following steps:
In the bridge health monitoring fault data after the data of single-point deletion type are repaired, one section is chosen without missing
Data, as training data sequence；
SARIMA model is established according to training data sequence, forward prediction is carried out, obtains past since training data sequence
The predicted value of the data of continuity point deletion type afterwards, and corresponding null value is replaced with predicted value, obtain partially complete data sequence
Column；
SARIMA model is established again according to partially complete data sequence, is carried out back forecast, is obtained training data sequence
Start the predicted value of the data of continuity point deletion type forward, and replace corresponding null value with predicted value, obtains partial data
Sequence.
It is an object of the invention to also provide a kind of bridge health monitoring fault data repair system, comprising:
Acquisition unit is used for: during acquiring bridge health monitoring fault data, missing data being assigned empty
Value；
Judging unit is used for: judging the type of missing data for single-point missing or continuity point missing；
Unit is repaired, is used for: when the type of missing data is single-point missing, according to multiple intact before and after missing data
Data are lost, the data of single-point deletion type are repaired using interpolation method；When the type of missing data is that continuity point lacks, according to one section
Without missing data, the data of continuity point deletion type are repaired using time series forecasting.
On the basis of above scheme, the judging unit judges that the adjacent data of a certain missing data is no missing number
According to when, then determine the type of the missing data for single-point missing.
On the basis of above scheme, the judging unit judges that continuous multiple data are missing data, described to sentence
Disconnected unit then determines the type of these continuous multiple data for continuity point missing.
On the basis of above scheme, the reparation unit first repairs the data of single-point deletion type, then to even
The data of continuous point deletion type are repaired.
On the basis of above scheme, the unit of repairing is using time series forecasting reparation continuity point deletion type
Data, specific steps are as follows:
The reparation unit selects in the bridge health monitoring fault data after the data of single-point deletion type are repaired
One section is taken without missing data, as training data sequence；
The reparation unit establishes seasonal difference autoregressive moving-average model SARIMA according to training data sequence, into
Row forward prediction, obtains the predicted value of the data of the continuity point deletion type since training data sequence backward, and with prediction
Value replaces corresponding null value, obtains partially complete data sequence；
The reparation unit establishes seasonal difference autoregressive moving-average model according to partially complete data sequence again
SARIMA carries out back forecast, obtains the predicted value that training data sequence starts the data of continuity point deletion type forward, and
Corresponding null value is replaced with predicted value, obtains partial data sequence.
Compared with the prior art, the advantages of the present invention are as follows:
(1) method of bridge health monitoring fault data reparation of the invention, does not need huge historical data, based on small
The raw monitored fault data of sample is repaired the data of single-point deletion type using interpolation method, is repaired using time series forecasting
The data of multiple continuity point deletion type, it is fast to repair speed.
(2) system of bridge health monitoring fault data reparation of the invention does not need to handle huge historical data, right
The requirement of hardware device is low.
Detailed description of the invention
Fig. 1 is one embodiment process signal of the method for Bridge of embodiment of the present invention health monitoring fault data reparation
Figure；
Fig. 2 is one embodiment flow diagram that missing data type is judged in the embodiment of the present invention；
Fig. 3 is one embodiment procedure Procedure figure repaired in the embodiment of the present invention to single-point missing data type；
Fig. 4 is one embodiment procedure Procedure repaired in the embodiment of the present invention to continuity point missing data type
Figure；
Fig. 5 is a structure flow chart of the system of Bridge of embodiment of the present invention health monitoring fault data reparation.
Specific embodiment
Invention is further described in detail with reference to the accompanying drawings and embodiments.
Embodiment one
Shown in Figure 1, the embodiment of the invention provides a kind of methods of bridge health monitoring fault data reparation, including
Following steps:
Step S1 assigns missing data to null value during acquiring bridge health monitoring fault data.
Step S2 judges the type of missing data for single-point missing or continuity point missing.
Step S3, when the type of missing data is that single-point lacks, then according to multiple no missing datas before and after missing data,
The data of single-point deletion type are repaired using interpolation method.
Step S4, when the type of missing data is that continuity point lacks, then according to one section without missing data, using time series
The data of predicted method reparation continuity point deletion type.The method root of Bridge health monitoring fault data reparation of the embodiment of the present invention
According to one section without missing data, before the data that continuity point deletion type is repaired using time series forecasting, first single-point is lacked
The data of type are repaired, and the precision of repair data is helped to improve.
Compared with prior art, the method for the bridge health monitoring fault data reparation of the embodiment of the present invention, does not need Pang
The type of missing data is divided into single-point missing and continuous by big historical data, the raw monitored fault data based on small sample
Point missing, the data of single-point deletion type are repaired using interpolation method, repair continuity point deletion type using time series forecasting
Data, repair speed it is fast, can also guarantee repair precision.
Embodiment two
It is on the basis of example 1, shown in Figure 2 as preferred embodiment, in step S2, if a certain missing
The adjacent data of data is no missing data, then determines the type of the missing data for single-point missing；If continuous multiple numbers
According to being missing data, then determine the type of these continuous multiple data for continuity point missing.
The missing data type for judging initial data in the embodiment of the present invention using consecutive points detection method, by current number
The value of the initial data of the value of initial data and next number is compared to judge the missing data type of initial data.In conjunction with
Fig. 2 judges that the process of the missing data type of initial data is done as described below using consecutive points detection method.
Step S201, the process flow start.
Step S202, current number are initialized as 0.
Step S203 reads the value of the initial data of current number.
Step S204, judges whether the value of the initial data of current number is equal to null value, if step is no, processing entrance
Step S205；If it is, processing enters step S206.
Current number is added up 1, the processing returns to step S203 by step S205.
Step S206 judges whether the value of the initial data of current number is equal with the value of the initial data of next number,
If NO, then processing enters step S207；If YES, then processing enters step S208.
Step S207, marking the initial data of current number is single-point missing data, and step S209 is jumped in processing.
Step S208, marking the initial data of current number is continuity point missing data, goes to step S209.
Step S209, judges whether current number is greater than or equal to scheduled data amount check, and if NO, then processing enters
Step S210；If YES, then processing terminates in step S211.
Current number is added up 1, the processing returns to step S203 by step S210.
Embodiment three
On the basis of example 2, shown in Figure 3 as preferred embodiment, in step S3, the present invention is implemented
According to multiple no missing datas before and after missing data in example, the data type of single-point missing is repaired using interpolation method.In conjunction with figure
3, the process of the data type reparation of single-point missing is done as described below.
Step S301, the process flow start.
Step S302, current number are initialized as 0.
Step S303, the initial data of the current number after reading detection.
Step S304, judges whether the initial data of current number is the single-point missing data of label, if it is not, then handling
Enter step S305；If it is, processing enters step S306.
Current number is added up 1, return step S303 by step S305.
Step S306 constructs cubic spline functions and calculates interpolation, interpolation is replaced to the value of single-point missing data, turns
To step S307.
As described in step S306, the interpolation method used in the embodiment of the present invention is cubic spline interpolation, the present invention
Technical solution be not intended to limit and specifically which kind of prior art be interpolation is calculated using, interpolation is replaced to the value of single-point missing data.
Step S307, judges whether current number is greater than or equal to scheduled data amount check, and if NO, then processing enters
Step S308；If YES, then processing terminates in step S309.
Current number is added up 1, the processing returns to step S303 by step S308.
Example IV
It is shown in Figure 4 on the basis of embodiment three as preferred embodiment, in step S4, missing data
When type is that continuity point lacks, then according to one section without missing data, continuity point deletion type is repaired using time series forecasting
Data.As shown in connection with fig. 4, the process of the data reparation of continuity point deletion type is done as described below.
Step S401, the process flow start.
Step S402 chooses one section without missing data.Specifically, the bridge after the data of single-point deletion type are repaired
In beam health monitoring fault data, one section is chosen without missing data, as training data sequence；
Step S403, sequence forward prediction, predicted value substitution are vacant.Specifically, season is established according to training data sequence
Property difference autoregressive moving-average model SARIMA, carry out forward prediction, obtain since training data sequence backward continuous
The predicted value of the data of point deletion type, predicted value replace corresponding null value, obtain partially complete data sequence.
Step S404, sequence back forecast, predicted value substitution are vacant.Season is established again according to partially complete data sequence
Property difference autoregressive moving-average model SARIMA, carry out back forecast, obtain training data sequence and start continuity point forward
The predicted value of the data of deletion type, and corresponding null value is replaced with predicted value, obtain perfect mistake data sequence.
Step S405 exports final perfect mistake data sequence.
Step S406, processing terminate.
Embodiment five
As preferred embodiment, on the basis of example IV, in step S403 and step S404, SARIMA is established
Model is represented by SARIMA (p, d, q) (P, D, Q) s, by the model is defined as:
φ(B)Φ(Bs)(1-Bs)D(1-B)dxt=θ (B) Θ (Bs)εt；
Wherein: t indicates time index；
{xtIndicate monitoring data sequence；
{εtIndicate mean value be zero, the independent identically distributed random sequence that variance is definite value；
P indicates the polynomial order of autoregression in short-term；
Q indicates the polynomial order of sliding average in short-term；
D indicates the order of difference in short-term；
S indicates seasonal；
P indicates the polynomial order of season autoregression；
Q indicates the polynomial order of season sliding average；
The order of D expression seasonal difference；
B indicates delay operator, makes Bxt=xt-1；
(1-Bs)DIndicate seasonal difference；
(1-B)dIndicate difference in short-term；
φ (B)=1- φ1B-φ2B2-···-φpBpIndicate autoregression multinomial in short-term；
θ (B)=1- θ1B-θ2B2-···-θqBqIndicate sliding average multinomial in short-term；
Φ(Bs)=1- Φ1(Bs)-Φ2(Bs)2-···-Φp(Bs)pIndicate season autoregression multinomial；
Θ(Bs)=1- Θ1(Bs)-Θ2(Bs)2-···-Θp(Bs)pIndicate season sliding average multinomial.It establishes
SARIMA (p, d, q) (P, D, Q) s model process is as follows:
Step A1 confirms whether training data sequence is stationary sequence by unit root test method, if then going to step
A2；Otherwise to no missing data carry out difference, until differentiated data be stationary sequence, record seasonal difference order D and
The order d of difference in short-term, goes to step A2；
Step A2 determines the p value of SARIMA model, q value, P value, Q value and season using red pond information content AIC criterion
Property s value, goes to step A3；
Step A3, the polynomial coefficient φ of estimation that model is obtained using maximum likelihood estimate1,φ2,···,φp,
θ1,θ2,···,θq, Φ1,Φ2,···,Φp, Θ1,Θ2,···,ΘqValue；
Step A4 diagnoses { ε using Chi-square TesttWhether sequence is white noise sequence, if so, SARIMA model is can
Receive；Conversely, then SARIMA model be it is unacceptable, modify SARIMA model, the step of repeating A1~A4 until
SARIMA model is acceptable.
Embodiment six
Shown in Figure 5, the embodiment of the invention provides also a kind of system of bridge health monitoring fault data reparation, packets
It includes:
Acquisition unit 501, is used for: during acquiring bridge health monitoring fault data, missing data being assigned
Null value；
Judging unit 502, is used for: judging the type of missing data for single-point missing or continuity point missing；
Unit 503 is repaired, is used for: when the type of missing data is single-point missing, according to multiple before and after missing data
Without missing data, the data of single-point deletion type are repaired using interpolation method；When the type of missing data is that continuity point lacks, according to
One section, without missing data, the data of continuity point deletion type is repaired using time series forecasting.Bridge of the embodiment of the present invention
The system of health monitoring fault data reparation, without missing data, repairs continuity point missing using time series forecasting according to one section
Before the data of type, repairs unit 503 and first the data of single-point deletion type are repaired, help to improve repair data
Precision.
Compared with prior art, the system of the bridge health monitoring fault data reparation of the embodiment of the present invention, does not need Pang
The type of missing data is divided into single-point missing and continuous by big historical data, the raw monitored fault data based on small sample
Point missing, the data of single-point deletion type are repaired using interpolation method, repair continuity point deletion type using time series forecasting
Data, repair speed it is fast, can also guarantee repair precision.The system of bridge health monitoring fault data reparation of the invention,
It does not need to handle huge historical data, the requirement to hardware device is low.
Embodiment seven
Shown in Figure 2 on the basis of embodiment six as preferred embodiment, judging unit 502 judges a certain
The adjacent data of missing data is no missing data, then determines the type of the missing data for single-point missing；Judging unit is sentenced
Disconnected 502 continuous multiple data are missing data, then determine the type of these continuous multiple data for continuity point missing.
The missing data type for judging initial data in the embodiment of the present invention using consecutive points detection method, by current number
The value of the initial data of the value of initial data and next number is compared to judge the missing data type of initial data.In conjunction with
Fig. 2 judges that the process of the missing data type of initial data is done as described below using consecutive points detection method.
Step S201, the process flow start.
Step S202, current number are initialized as 0.
Step S203 reads the value of the initial data of current number.
Step S204, judges whether the value of the initial data of current number is equal to null value, and if NO, then processing enters step
Rapid S205；If YES, then processing enters step S206.
Current number is added up 1, the processing returns to step S203 by step S205.
Step S206 judges whether the value of the initial data of current number is equal with the value of the initial data of next number,
If NO, then processing enters step S207；If YES, then processing enters step S208.
Step S207, marking the initial data of current number is single-point missing data, and step S209 is jumped in processing.
Step S208, marking the initial data of current number is continuity point missing data, goes to step S209.
Step S209, judges whether current number is greater than or equal to scheduled data amount check, and if NO, then processing enters
Step S210；If YES, then processing terminates in step S211.
Current number is added up 1, the processing returns to step S203 by step S210.
Embodiment eight
Shown in Figure 3 on the basis of embodiment seven as preferred embodiment, the type of missing data is single
When point missing, unit 503 is repaired according to multiple no missing datas before and after missing data, single-point is repaired using interpolation method and lacks class
The data of type.In conjunction with Fig. 3, the process of the data reparation of single-point deletion type is done as described below.
Step S301, the process flow start.
Step S302, current number are initialized as 0.
Step S303, the initial data of the current number after reading detection.
Step S304, judges whether the initial data of current number is the single-point missing data of label, if it is not, then handling
Enter step S305；If it is, processing enters step S306.
Current number is added up 1, return step S303 by step S305.
Step S306 constructs cubic spline functions and calculates interpolation, interpolation is replaced to the value of single-point missing data, turns
To step S307.
As described in step S306, the interpolation method used in the embodiment of the present invention is cubic spline interpolation, the present invention
Technical solution be not intended to limit and specifically which kind of prior art be interpolation is calculated using, interpolation is replaced to the value of single-point missing data.
Step S307, judges whether current number is greater than or equal to scheduled data amount check, and if NO, then processing enters
Step S308；If YES, then processing terminates in step S309.
Current number is added up 1, the processing returns to step S303 by step S308.
Embodiment nine
Shown in Figure 4 on the basis of embodiment eight as preferred embodiment, the type of missing data is to connect
When continuous point missing, unit 503 is repaired according to one section without missing data, continuity point deletion type is repaired using time series forecasting
Data.As shown in connection with fig. 4, the process of the data reparation of continuity point deletion type is done as described below.
Step S401, the process flow start.
Step S402 chooses one section without missing data.Specifically, the bridge after the data of single-point deletion type are repaired
In beam health monitoring fault data, one section is chosen without missing data, as training data sequence；
Step S403, sequence forward prediction, predicted value substitution are vacant.Specifically, season is established according to training data sequence
Property difference autoregressive moving-average model SARIMA, carry out forward prediction, obtain since training data sequence backward continuous
The predicted value of the data of point deletion type, predicted value replace corresponding null value, obtain partially complete data sequence.
Step S404, sequence back forecast, predicted value substitution are vacant.Season is established again according to partially complete data sequence
Property difference autoregressive moving-average model SARIMA, carry out back forecast, obtain training data sequence and start continuity point forward
The predicted value of the data of deletion type, and corresponding null value is replaced with predicted value, obtain perfect mistake data sequence.
Step S405 exports final perfect mistake data sequence.
Step S406, processing terminate.
Embodiment ten
As preferred embodiment, on the basis of embodiment nine, in step S403 and step S404, SARIMA is established
Model is represented by SARIMA (p, d, q) (P, D, Q) s, by the model is defined as:
φ(B)Φ(Bs)(1-Bs)D(1-B)dxt=θ (B) Θ (Bs)εt；
Wherein: t indicates time index；
{xtIndicate monitoring data sequence；
{εtIndicate mean value be zero, the independent identically distributed random sequence that variance is definite value；
P indicates the polynomial order of autoregression in short-term；
Q indicates the polynomial order of sliding average in short-term；
D indicates the order of difference in short-term；
S indicates seasonal；
P indicates the polynomial order of season autoregression；
Q indicates the polynomial order of season sliding average；
The order of D expression seasonal difference；
B indicates delay operator, makes Bxt=xt-1；
(1-Bs)DIndicate seasonal difference；
(1-B)dIndicate difference in short-term；
φ (B)=1- φ1B-φ2B2-···-φpBpIndicate autoregression multinomial in short-term；
θ (B)=1- θ1B-θ2B2-···-θqBqIndicate sliding average multinomial in short-term；
Φ(Bs)=1- Φ1(Bs)-Φ2(Bs)2-···-Φp(Bs)pIndicate season autoregression multinomial；
Θ(Bs)=1- Θ1(Bs)-Θ2(Bs)2-···-Θp(Bs)pIndicate season sliding average multinomial.It establishes
SARIMA (p, d, q) (P, D, Q) s model process is as follows:
Step A1 confirms whether training data sequence is stationary sequence by unit root test method, if then going to step
A2；Otherwise to no missing data carry out difference, until differentiated data be stationary sequence, record seasonal difference order D and
The order d of difference in short-term, goes to step A2；
Step A2 determines the p value of SARIMA model, q value, P value, Q value and season using red pond information content AIC criterion
Property s value, goes to step A3；
Step A3, the polynomial coefficient φ of estimation that model is obtained using maximum likelihood estimate1,φ2,···,φp,
θ1,θ2,···,θq, Φ1,Φ2,···,Φp, Θ1,Θ2,···,ΘqValue；
Step A4 diagnoses { ε using Chi-square TesttWhether sequence is white noise sequence, if so, SARIMA model is can
Receive；Conversely, then SARIMA model be it is unacceptable, modify SARIMA model, the step of repeating A1~A4 until
SARIMA model is acceptable.
The present invention is not limited to the above-described embodiments, for those skilled in the art, is not departing from
Under the premise of the principle of the invention, several improvements and modifications can also be made, these improvements and modifications are also considered as protection of the invention
Within the scope of.The content being not described in detail in this specification belongs to the prior art well known to professional and technical personnel in the field.
Claims (10)
1. a kind of bridge health monitoring fault data restorative procedure, which comprises the following steps:
During acquiring bridge health monitoring fault data, missing data is assigned to null value；
When determining that the type of missing data lacks for single-point, then according to multiple no missing datas before and after missing data, using slotting
The data of value method reparation single-point deletion type；
When determining that the type of missing data lacks for continuity point, then according to one section without missing data, using time series forecasting
Repair the data of continuity point deletion type.
2. bridge health monitoring fault data restorative procedure as described in claim 1, which is characterized in that the judgement missing number
According to type be single-point missing, comprising the following steps: if the adjacent data of a certain missing data is no missing data, determine
The type of the missing data is single-point missing.
3. bridge health monitoring fault data restorative procedure as described in claim 1, which is characterized in that the judgement missing number
According to type be continuity point missing, comprising the following steps: if continuous multiple data are missing data, determine that these are continuous
Multiple data type be continuity point missing.
4. bridge health monitoring fault data restorative procedure as described in claim 1, it is characterised in that: first lack class to single-point
The data of type are repaired, then are repaired to the data of continuity point deletion type.
5. bridge health monitoring fault data restorative procedure as claimed in claim 4, which is characterized in that described to use time sequence
The data of column predicted method reparation continuity point deletion type, comprising the following steps:
In the bridge health monitoring fault data after the data of single-point deletion type are repaired, one section is chosen without missing number
According to as training data sequence；
Seasonal difference autoregressive moving-average model SARIMA is established according to training data sequence, forward prediction is carried out, obtains
The predicted value of the data of continuity point deletion type since training data sequence backward, and corresponding sky is replaced with predicted value
Value, obtains partially complete data sequence；
Seasonal difference autoregressive moving-average model SARIMA is established again according to partially complete data sequence, to pre- after progress
It surveys, obtains the predicted value that training data sequence starts the data of continuity point deletion type forward, and replaced with predicted value corresponding
Null value, obtain partial data sequence.
6. a kind of bridge health monitoring fault data repair system characterized by comprising
Acquisition unit is used for: during acquiring bridge health monitoring fault data, assigning missing data to null value；
Judging unit is used for: judging the type of missing data for single-point missing or continuity point missing；
Unit is repaired, is used for: when the type of missing data is single-point missing, according to multiple no missing numbers before and after missing data
According to using the data of interpolation method reparation single-point deletion type；It is intact according to one section when the type of missing data is that continuity point lacks
Data are lost, the data of continuity point deletion type are repaired using time series forecasting.
7. bridge health monitoring fault data repair system as claimed in claim 6, it is characterised in that: the judging unit is sentenced
When the adjacent data of a certain missing data of breaking is no missing data, then determine the type of the missing data for single-point missing.
8. bridge health monitoring fault data repair system as claimed in claim 6, it is characterised in that: the judging unit is sentenced
Disconnected continuous multiple data are missing data, and the judging unit then determines that the type of these continuous multiple data is continuous
Point missing.
9. bridge health monitoring fault data repair system as claimed in claim 6, it is characterised in that: the reparation unit is first
The data of single-point deletion type are repaired, then the data of continuity point deletion type are repaired.
10. bridge health monitoring fault data repair system as claimed in claim 9, it is characterised in that: the reparation unit
The data of continuity point deletion type, specific steps are repaired using time series forecasting are as follows:
The reparation unit chooses one in the bridge health monitoring fault data after the data of single-point deletion type are repaired
Section is without missing data, as training data sequence；
The reparation unit establishes seasonal difference autoregressive moving-average model SARIMA according to training data sequence, before progress
To prediction, the predicted value of the data of the continuity point deletion type since training data sequence backward is obtained, and replaced with predicted value
Corresponding null value is changed, partially complete data sequence is obtained；
The reparation unit establishes seasonal difference autoregressive moving-average model according to partially complete data sequence again
SARIMA carries out back forecast, obtains the predicted value that training data sequence starts the data of continuity point deletion type forward, and
Corresponding null value is replaced with predicted value, obtains partial data sequence.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811448903.0A CN109376478A (en) | 2018-11-28 | 2018-11-28 | Bridge health monitoring fault data restorative procedure and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811448903.0A CN109376478A (en) | 2018-11-28 | 2018-11-28 | Bridge health monitoring fault data restorative procedure and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109376478A true CN109376478A (en) | 2019-02-22 |
Family
ID=65375044
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811448903.0A Pending CN109376478A (en) | 2018-11-28 | 2018-11-28 | Bridge health monitoring fault data restorative procedure and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109376478A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112926442A (en) * | 2021-02-24 | 2021-06-08 | 杭州电子科技大学 | Construction method for balanced and complete image target data set |
| CN113157769A (en) * | 2021-04-13 | 2021-07-23 | 成都江泰讯安信息技术有限公司 | Hydrological signaling defect report retrieval method |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6928398B1 (en) * | 2000-11-09 | 2005-08-09 | Spss, Inc. | System and method for building a time series model |
| CN102025531A (en) * | 2010-08-16 | 2011-04-20 | 北京亿阳信通软件研究院有限公司 | Filling method and device thereof for performance data |
| CN105046321A (en) * | 2015-06-25 | 2015-11-11 | 河海大学 | Combined model water level prediction method based on similarity search |
| CN107895014A (en) * | 2017-11-14 | 2018-04-10 | 辽宁工业大学 | A kind of time series bridge monitoring data analysing method based on MapReduce frameworks |
| CN107944464A (en) * | 2017-10-12 | 2018-04-20 | 华南理工大学 | A kind of office building by when energy consumption abnormal data online recognition and complementing method |
| CN108228862A (en) * | 2018-01-12 | 2018-06-29 | 上海市建筑科学研究院 | A kind of method for processing abnormal data and system of energy consumption monitoring platform |
-
2018
- 2018-11-28 CN CN201811448903.0A patent/CN109376478A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6928398B1 (en) * | 2000-11-09 | 2005-08-09 | Spss, Inc. | System and method for building a time series model |
| CN102025531A (en) * | 2010-08-16 | 2011-04-20 | 北京亿阳信通软件研究院有限公司 | Filling method and device thereof for performance data |
| CN105046321A (en) * | 2015-06-25 | 2015-11-11 | 河海大学 | Combined model water level prediction method based on similarity search |
| CN107944464A (en) * | 2017-10-12 | 2018-04-20 | 华南理工大学 | A kind of office building by when energy consumption abnormal data online recognition and complementing method |
| CN107895014A (en) * | 2017-11-14 | 2018-04-10 | 辽宁工业大学 | A kind of time series bridge monitoring data analysing method based on MapReduce frameworks |
| CN108228862A (en) * | 2018-01-12 | 2018-06-29 | 上海市建筑科学研究院 | A kind of method for processing abnormal data and system of energy consumption monitoring platform |
Non-Patent Citations (5)
| Title |
|---|
| 刘小玲 等: "带反馈机制的桥梁监测数据处理流程框架" * |
| 刘小玲 等: "带反馈机制的桥梁监测数据处理流程框架", 《公路》 * |
| 张广兴 等: "基于B-样条自由节点的外测缺失数据重构方法" * |
| 张广兴 等: "基于B-样条自由节点的外测缺失数据重构方法", 《测控技术》 * |
| 李文峰 等: "苏州市轨道交通典型站点客流离散特性对比分析" * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112926442A (en) * | 2021-02-24 | 2021-06-08 | 杭州电子科技大学 | Construction method for balanced and complete image target data set |
| CN112926442B (en) * | 2021-02-24 | 2024-03-29 | 杭州电子科技大学 | Construction method for image target data set balance completion |
| CN113157769A (en) * | 2021-04-13 | 2021-07-23 | 成都江泰讯安信息技术有限公司 | Hydrological signaling defect report retrieval method |
| CN113157769B (en) * | 2021-04-13 | 2023-05-02 | 成都江泰讯安信息技术有限公司 | Method for checking hydrological message lack of report |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109359698B (en) | Leakage identification method based on long-time memory neural network model | |
| CN115578015B (en) | Sewage treatment whole process supervision method, system and storage medium based on Internet of things | |
| CN107733357A (en) | The fault detection algorithm of battery panel in a kind of large-sized photovoltaic power station | |
| CN104077445B (en) | Accelerated life test statistical analysis technique based on fuzzy theory | |
| CN105116870B (en) | Fault analysis method, device and system of air conditioning unit | |
| CN106198749B (en) | A kind of data fusion method of the multiple sensor based on Metal Crack monitoring | |
| CN109376478A (en) | Bridge health monitoring fault data restorative procedure and system | |
| CN108896804A (en) | Stealing detection method, device and electronic equipment | |
| CN112270335A (en) | Method and system for predicting welding quality defects of lap joint and computer readable storage medium | |
| CN109784668A (en) | A kind of sample characteristics dimension-reduction treatment method for electric power monitoring system unusual checking | |
| CN112085621A (en) | Distributed photovoltaic power station fault early warning algorithm based on K-Means-HMM model | |
| CN103197168A (en) | Fault diagnosis control method in power system based on event set causal chain | |
| CN104122487B (en) | Method and device for identifying cable overcurrent causes | |
| CN110738415A (en) | Electricity stealing user analysis method based on electricity utilization acquisition system and outlier algorithm | |
| CN114765574B (en) | Network anomaly delimitation positioning method and device | |
| CN110597792A (en) | Multistage redundant data fusion method and device based on synchronous line loss data fusion | |
| CN116862209B (en) | New energy automobile charging facility management method and system | |
| CN105634781B (en) | Multi-fault data decoupling method and device | |
| CN116070106B (en) | Behavior sequence-based data interaction anomaly detection feature extraction method and device | |
| CN116720983A (en) | Power supply equipment abnormality detection method and system based on big data analysis | |
| CN107064745A (en) | Stagewise method for diagnosing faults based on transient current information and Wavelet Entropy | |
| CN111934903A (en) | Docker container fault intelligent prediction method based on time sequence evolution genes | |
| CN112507290B (en) | Power distribution equipment fault probability pre-judging method, device and storage medium | |
| JP2014530629A5 (en) | ||
| Che et al. | An efficient intrusion detection approach based on hidden markov model and rough set |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190222 |