CN111125824B - Structural damage identification method based on deletion model - Google Patents
Structural damage identification method based on deletion model Download PDFInfo
- Publication number
- CN111125824B CN111125824B CN201911373177.5A CN201911373177A CN111125824B CN 111125824 B CN111125824 B CN 111125824B CN 201911373177 A CN201911373177 A CN 201911373177A CN 111125824 B CN111125824 B CN 111125824B
- Authority
- CN
- China
- Prior art keywords
- model
- curvature
- regression model
- regression
- deletion
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T90/00—Enabling technologies or technologies with a potential or indirect contribution to GHG emissions mitigation
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Mathematical Physics (AREA)
- Computational Mathematics (AREA)
- Mathematical Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Operations Research (AREA)
- Probability & Statistics with Applications (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Algebra (AREA)
- Evolutionary Biology (AREA)
- Databases & Information Systems (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The invention discloses a structural damage identification method based on a deletion model, which comprises the following steps: obtaining the curvature modal difference between the current state and the intact state of the structure; establishing a curvature modal difference regression model, and matrixing the regression model; performing parameter estimation on the matrixed regression model; deleting the curvature modal difference of an observation point, reestablishing a regression model, matrixing and carrying out parameter estimation; analyzing the difference of regression coefficients before and after the observation point is deleted by using WK statistic, and judging whether the point of the structure is damaged; and deleting the observation points in sequence, and judging whether all the observation points of the structure are damaged. The invention can solve the technical problem that the judgment of the damage position is not clear due to noise and measurement error in the conventional dynamic damage identification.
Description
Technical Field
The invention relates to the technical field of engineering detection, in particular to a structural damage identification method based on a deletion model.
The background art comprises the following steps:
the damage research method based on the dynamic fingerprint comprehensively utilizes interdisciplinary technologies such as a structure vibration theory, a vibration testing technology, a data processing technology and the like, and is considered to be the most promising structure nondestructive testing method at present. In the dynamic fingerprint damage diagnosis method, the curvature mode shows high sensitivity of local damage, and the method becomes a hotspot of research in the structural engineering field. But the curvature mode sensitivity is high, and the index is easily influenced by random errors and measurement errors, so that the robustness of the index is poor.
Disclosure of Invention
The invention aims to provide a structural damage identification method based on a deletion model, which aims to solve the defect that damage position judgment is not clear due to noise and measurement error in the prior art.
A structural damage identification method based on a deletion model comprises the following steps:
obtaining the curvature modal difference between the current state and the intact state of the structure;
constructing a regression model according to the curvature modal difference, and performing parameter estimation after matrixing the regression model;
deleting curvature modal differences of a plurality of observation points, rebuilding a regression model, matrixing the regression model and then performing parameter estimation;
and calculating the WK value before and after the observation point is deleted according to the parameter estimation result, and judging whether the point is the damage in the structure or not according to the WK value.
Further, the method for acquiring the curvature modal difference comprises the following steps:
obtaining an initial curvature mode of the structure through a vibration test;
regularly observing the curvature mode of the current state;
and comparing the current curvature mode with the curvature mode in the intact state to obtain the curvature mode difference.
Further, the method for constructing the regression model according to the curvature modal difference and matrixing the regression model comprises the following steps:
let observation data be { x i ,y i },i=1,2,…,n;
Wherein x i Numbering the measuring points, y i Is the curvature mode difference of the corresponding unit;
establishing a regression model:
The model can be written in matrix form:
Y=Xβ+ε;
wherein Y is (Y) 1 ,y 2 ,…,y n ) T ;ε=(ε 1 ,ε 2 ,…ε n ) T The random error vector satisfies the white noise condition; x is an n X4 order matrix with the ith action
Further, the method for performing parameter estimation on the matrixed regression model comprises the following steps:
finding an estimate of the parameter beta using least squares estimationThe sum of the squares of the errors of the following equations:
As the fitting value of Y, the residual value is recorded asNote that the sum of the squares of the residuals is:
Further, the method for deleting curvature modal differences of a plurality of observation points, reconstructing a regression model, matrixing the regression model and then performing parameter estimation comprises the following steps:
the model after the ith data point is deleted is recorded as a data deletion model CDM expression:
y j =X j T β+ε j ,j=1,2,…,n,j≠i;
wherein j ≠ i means that the model does not contain the ith data point;
the model matrix form is:
Y(i)=X(i)β+ε(i);
wherein Y (i), X (i), epsilon (i) represents a vector or a matrix of Y, X, epsilon and epsilon after deleting the corresponding ith component; the CDM model least squares regression coefficient at this time is:
regression residual sum of squares:
and random error variance estimation:
further, the method for statistically analyzing the difference between the regression coefficients before and after the observation point is deleted and determining whether the point is a damage in the structure includes the following steps:
the observation points were analyzed using the WK statistic (Welsch-Kul statics, Welsh-Kura statistic), defined as:
can also be written asWhereinp ii Is a hat matrix P diagonal elements, andobeying a t-distribution with a parameter of n-5;
according to the WK distribution, giving a significance level alpha to obtain the WK i The confidence interval with confidence 1- α is:
wherein t is α (n-5) is a t distribution upper side alpha quantile;
thus, the ith data point { x i ,y i The judgment standard for the corresponding unit as a damaged unit is
Calculate i data points { x i ,y i | WK of } i The value of | and the threshold valueMake a comparison ifThe ith cell is determined to be a defective cell.
The invention has the advantages that: the method provided by the invention can accurately find the structure damage position under the condition of larger environmental noise, has strong robustness and realizes data support for structure damage detection and health evaluation; the method can be applied to civil engineering such as bridges, high-rise buildings, hydraulic engineering and the like, and has wide application range; can be implemented in combination with programming, and the data processing is automatic and efficient.
Drawings
Fig. 1 is an aerial view of an experimental model of a high-piled wharf in the invention.
FIG. 2 is a cross-sectional view of an experimental model of a high-piled wharf according to the present invention.
Fig. 3 is a diagram of a sensor arrangement according to the present invention.
FIG. 4 is a graph of 10% damage behind the test point No. 5 in the present invention.
Fig. 5 is a diagram of frequency analysis of the wharf model of the present invention.
Fig. 6 is a curve of the difference in curvature modes in the present invention.
Fig. 7 shows the 10% damage identification process of the present invention.
Fig. 8 shows the 20% damage identification process of the present invention.
FIG. 9 is a table of experimental conditions in the present invention.
FIG. 10 is a flow chart of the method of the present invention.
Detailed Description
In order to make the technical means, the creation characteristics, the achievement purposes and the effects of the invention easy to understand, the invention is further described with the specific embodiments.
As shown in fig. 1 to 10, a structural damage recognition method based on a deletion model,
firstly, a high-pile wharf model is manufactured in a laboratory, the high-loading wharf model is shown in figures 1 and 2, 1 acceleration sensor is arranged on the second pile body on the left side of the front row from top to bottom every 0.01m, 10 acceleration sensors are arranged, the number of the acceleration sensors is 1,2, … … and 10 from top to bottom, the sensor arrangement diagram is shown in figure 3, and the vibration pickup direction of the sensors is vertical to the front edge direction of the wharf. The hammering time-course response before and after the pile body is damaged is collected to analyze the local vibration mode of the pile body, the damage is set to be that the section inertia moment of the cut pile is adopted to reduce the section behind the No. 5 measuring point to simulate the damage of the pile foundation, the damage length is 0.01m as shown in figure 4, and the damage width and the test working condition are shown in figure 9.
The identification method specifically comprises the following steps:
the method comprises the following steps: obtaining the curvature modal difference between the current state and the intact state of the structure:
the mode analysis is carried out by using mature commercial software DHDAS after the test data are collected, fig. 5 is a natural frequency analysis chart of the structure, the graph shows that the order frequency of the structure 1 is 7.422Hz, the second order frequency is 20.703Hz, and the amplitude of the second order frequency corresponding to the vibration mode in the vibration pickup direction of the sensor is the largest, so that the second order vibration mode test data collected by the sensor have good signal-to-noise ratio, and the damage identification is carried out by analyzing the second order local vibration mode of the pile body in the vibration pickup direction through the test data. A curvature mode difference curve is calculated from the mode shape as shown in fig. 6.
Step two: establishing a curvature mode difference polynomial regression model, and matrixing the regression model: let observation data be { x i , y i 1,2, …, n. where x is i Numbering the measuring points, y i Is the curvature modal difference of the corresponding cell. Establishing a polynomial regression model:
vector of independent variablesCoefficient vector β ═ β (β) 0 ,β 1 ,β 2 ,β 3 ) T . The model can be written in matrix form:
Y=Xβ+ε.
wherein Y is (Y) 1 ,y 2 ,…,y n ) T ;ε=(ε 1 ,ε 2 ,…ε n ) T Is a random error vector; x is an n X4 order matrix with the ith actionThe regression model satisfies the regression model assumption: epsilon 1 ,ε 2 ,…ε n Obedience mean 0 and variance σ 2 Normal distributions independent of each other.
Step three: performing parameter estimation on the matrixed regression model: and for the polynomial regression model, performing parameter estimation by adopting a least square estimation method. I.e. finding an estimate of the parameter betaSum of squares of errorsTo a minimum. WhereinCommonly referred to as P ═ X T X) -1 X T For hat matrix, noteAs the fitting value of Y, the residual value is recorded asThe sum of the squares of the residuals isVariance of random error σ 2 Is estimated as
Step four: and deleting the curvature modal difference of a certain observation point, reestablishing a polynomial regression model, matrixing and carrying out parameter estimation. The model after the Deletion of the ith data point is recorded as a data Deletion model cdm (case Deletion model), and the expression is as follows:
y j =X j T β+ε j ,j=1,2,…,n,j≠i.
wherein j ≠ i means that the model does not contain the ith data point matrix form as:
Y(i)=X(i)β+ε(i).
where Y (i), X (i), ε (i) represents the vector or matrix after the removal of the corresponding i-th component { Y }, [ X ], { ε }. The CDM model least squares regression coefficient at this time is:
regression residual sum of squares:
and random error variance estimation:
the parameter estimation method is the same as the third step.
Step five: and analyzing the difference of regression coefficients before and after the observation point is deleted by using the WK statistic, and judging whether the point of the structure is damaged or not. The WK statistic is defined as:can also be written asWhereinp ii Is a hat matrix P diagonal elements, andobeying a t-distribution with a parameter n-5. According to the WK distribution, giving a significance level alpha to obtain the WK i The confidence interval with confidence 1- α is:
wherein t is α/2 (n-5) is a t distribution upper side alpha/2 quantile point;
thus, the ith data point { x i ,y i Couple (c)The judgment standard for the unit to be damaged is
That is, i data points { x } are calculated i ,y i | WK of } i The value of | and the threshold valueMake a comparison ifThe ith cell is determined to be a defective cell.
Step six: judging whether the observation point is damaged: using the data deletion model for condition 1 (10% damage), the damage identification process is shown in fig. 7: and gradually deleting each unit to obtain MWK values of each unit after deletion, finding that the MWK value of the No. 5 unit is larger than the threshold value, and the MWK values of other units are smaller than the threshold value, so that the No. 5 unit can be determined as a damaged unit, which is consistent with the actual situation. Using a step-by-step data deletion model for condition 2 (20% damage), the damage identification process is shown in fig. 8: and gradually deleting each unit to obtain MWK values of each unit after deletion, finding that the MWK value of the No. 5 unit is larger than the threshold value, and the MWK values of other units are smaller than the threshold value, so that the No. 5 unit can be determined to be a damaged unit.
Based on the above, in a statistical sense, the damage identification is essentially to find out the abnormal points whose structural unit behaviors are different from the general rule. In actual engineering, a damaged area is local and small-range relative to the whole structure, so that the observed data features in the non-damaged area are dominant and dominant in the overall observed data, and the observed data features of the damaged area are different from those of the non-damaged observed area, so that the abnormal value which is found to be inconsistent with the whole data is the mathematical essence of structural damage identification. According to the curve characteristics of curvature mode difference before and after structural damage position damage, the curvature mode damage identification method based on the data deletion model, which is capable of automatically identifying and has high robustness, is established by deleting the influence of each observation point data on the overall curvature mode difference curve form. The damage experiment of the high-pile wharf model proves that the method has better applicability.
It will be appreciated by those skilled in the art that the invention may be embodied in other specific forms without departing from the spirit or essential characteristics thereof. The embodiments disclosed above are therefore to be considered in all respects as illustrative and not restrictive. All changes which come within the scope of or equivalence to the invention are intended to be embraced therein.
Claims (2)
1. A structural damage identification method based on a deletion model is characterized by comprising the following steps:
obtaining the curvature modal difference between the current state and the intact state of the structure;
constructing a regression model according to the curvature modal difference, and performing parameter estimation after matrixing the regression model;
deleting curvature modal differences of a plurality of observation points, rebuilding a regression model, matrixing the regression model and then performing parameter estimation;
calculating WK values of the observation points before and after deletion according to the parameter estimation result, and judging whether the point is damaged in the structure according to the WK values;
the method for constructing the regression model according to the curvature modal difference and matrixing the regression model comprises the following steps of:
let observation data be { x i ,y i },i=1,2,…,n;
Wherein x i Numbering the measuring points, y i Is the curvature mode difference of the corresponding unit;
establishing a regression model:
The model is written in matrix form:
Y=Xβ+ε;
wherein Y is (Y) 1 ,y 2 ,…,y n ) T ;ε=(ε 1 ,ε 2 ,…ε n ) T The random error vector satisfies the white noise condition; x is an n X4 order matrix with the ith action
The method for parameter estimation of the matrixed regression model comprises the following steps:
finding an estimate of the parameter beta using least squares estimationThe sum of the squares of the errors of the following equations:
to obtainThe term P ═ X T X) -1 X T For hat matrix, noteAs the fitting value of Y, the residual value is recorded asNote that the sum of the squares of the residuals is:
The method for deleting curvature modal differences of a plurality of observation points, reconstructing a regression model, matrixing the regression model and then performing parameter estimation comprises the following steps:
and marking the model after the ith data point is deleted as a data deletion model, wherein the expression is as follows:
y j =X j T β+ε j ,j=1,2,…,n,j≠i;
wherein j ≠ i means that the model does not contain the ith data point;
the model matrix form is:
Y(i)=X(i)β+ε(i);
wherein Y (i), X (i), epsilon (i) represents a vector or a matrix of Y, X, epsilon and epsilon after deleting the corresponding ith component;
the least squares regression coefficient of the data deletion model at this time is:
regression residual sum of squares:
and random error variance estimation:
the method for calculating the WK value before and after deletion of the observation point according to the parameter estimation result and judging whether the point is a damage in the structure according to the WK value comprises the following steps:
analyzing observation points by adopting WK statistics, wherein the observation points are defined as:
Whereinp ii Is a hat matrix P diagonal elements, andobeying a t-distribution with a parameter of n-5;
according to the WK distribution, giving a significance level alpha to obtain the WK i The confidence interval with confidence 1- α is:
wherein t is α (n-5) is a t distribution upper side alpha quantile;
thus, the ith data point { x } i ,y i The judgment standard for the corresponding unit as a damaged unit is as follows:
2. The structural damage identification method based on the deletion model according to claim 1, characterized in that: the method for acquiring the curvature modal difference comprises the following steps:
obtaining an initial curvature mode of the structure through a vibration test;
regularly observing the curvature mode of the current state;
and comparing the current curvature mode with the curvature mode in the intact state to obtain the curvature mode difference.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911373177.5A CN111125824B (en) | 2019-12-27 | 2019-12-27 | Structural damage identification method based on deletion model |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911373177.5A CN111125824B (en) | 2019-12-27 | 2019-12-27 | Structural damage identification method based on deletion model |
Publications (2)
Publication Number | Publication Date |
---|---|
CN111125824A CN111125824A (en) | 2020-05-08 |
CN111125824B true CN111125824B (en) | 2022-08-09 |
Family
ID=70503752
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201911373177.5A Active CN111125824B (en) | 2019-12-27 | 2019-12-27 | Structural damage identification method based on deletion model |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111125824B (en) |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105064420A (en) * | 2015-08-06 | 2015-11-18 | 交通运输部天津水运工程科学研究所 | High-pile wharf foundation pile damage diagnosis method based on structural residual modal force |
CN106897543A (en) * | 2017-04-25 | 2017-06-27 | 湘潭大学 | The girder construction damnification recognition method of On Modal Flexibility Curvature matrix norm |
CN109543303A (en) * | 2018-11-22 | 2019-03-29 | 华北水利水电大学 | A method of the Damage Assessment Method to be perturbed based on class curvature of the flexibility difference matrix and frequency |
CN110455476A (en) * | 2019-07-29 | 2019-11-15 | 河海大学 | A kind of multidimensional dynamical dactylogram damnification recognition method based on MCD abnormal point checking method |
-
2019
- 2019-12-27 CN CN201911373177.5A patent/CN111125824B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105064420A (en) * | 2015-08-06 | 2015-11-18 | 交通运输部天津水运工程科学研究所 | High-pile wharf foundation pile damage diagnosis method based on structural residual modal force |
CN106897543A (en) * | 2017-04-25 | 2017-06-27 | 湘潭大学 | The girder construction damnification recognition method of On Modal Flexibility Curvature matrix norm |
CN109543303A (en) * | 2018-11-22 | 2019-03-29 | 华北水利水电大学 | A method of the Damage Assessment Method to be perturbed based on class curvature of the flexibility difference matrix and frequency |
CN110455476A (en) * | 2019-07-29 | 2019-11-15 | 河海大学 | A kind of multidimensional dynamical dactylogram damnification recognition method based on MCD abnormal point checking method |
Also Published As
Publication number | Publication date |
---|---|
CN111125824A (en) | 2020-05-08 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN113838054B (en) | Mechanical part surface damage detection method based on artificial intelligence | |
US6367040B1 (en) | System and method for determining yield impact for semiconductor devices | |
CN110706213B (en) | Bridge cluster structure damage judgment method based on strain response cumulative distribution function difference | |
CN115294139A (en) | Image-based slope crack monitoring method | |
US20200393347A1 (en) | Imaging Method of Internal Defects in Longitudinal Sections of Trees | |
CN112411371A (en) | Highway bridge comprehensive detection method and system based on mobile sensing and vision | |
CN110852001A (en) | Bridge structure safety assessment method based on image processing | |
CN110728089B (en) | Large-span bridge stay cable structure damage diagnosis method based on BOTDA technology | |
CN110455476B (en) | Multi-dimensional dynamic fingerprint damage identification method based on MCD abnormal point detection algorithm | |
Lang et al. | Pavement cracking detection and classification based on 3d image using multiscale clustering model | |
CN111125824B (en) | Structural damage identification method based on deletion model | |
CN111062083B (en) | Structural damage identification method based on gradual deletion model | |
CN111141472B (en) | Anti-seismic support and hanger detection method and system | |
KR101255022B1 (en) | Detecting method of building crack using point group | |
CN112732773B (en) | Method and system for checking uniqueness of relay protection defect data | |
CN114563771A (en) | Double-threshold laser radar cloud layer detection algorithm based on cluster analysis | |
CN114882009A (en) | Fatigue crack tip automatic detection method and system capable of adapting to various surface states | |
CN114186312A (en) | Tunnel-related landslide type identification method based on tunnel deformation characteristics | |
CN112161785A (en) | Ocean engineering structure micro-damage judgment method | |
CN117421965B (en) | Post-earthquake bridge detection method and system based on unmanned aerial vehicle technology | |
KR102061539B1 (en) | Safety Inspection Method of Structure Using Drone with Thermal Infrared Sensor | |
CN115980890B (en) | Rainfall station abnormal data detection method based on space-time elements | |
CN114841939B (en) | Grey cloth defect degree analysis method based on image processing | |
CN116482622A (en) | Lightning refined early warning effect inspection method | |
CN109472083B (en) | Civil engineering structure damage detection method based on statistical process control |
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 |