CN118013808A - Method for identifying looseness health state of bolt of offshore wind power generation structure - Google Patents
Method for identifying looseness health state of bolt of offshore wind power generation structure Download PDFInfo
- Publication number
- CN118013808A CN118013808A CN202410417226.5A CN202410417226A CN118013808A CN 118013808 A CN118013808 A CN 118013808A CN 202410417226 A CN202410417226 A CN 202410417226A CN 118013808 A CN118013808 A CN 118013808A
- Authority
- CN
- China
- Prior art keywords
- data
- power generation
- offshore wind
- wind power
- fusion
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000010248 power generation Methods 0.000 title claims abstract description 79
- 238000000034 method Methods 0.000 title claims abstract description 77
- 230000036541 health Effects 0.000 title claims abstract description 55
- 230000004927 fusion Effects 0.000 claims abstract description 88
- 238000012512 characterization method Methods 0.000 claims abstract description 84
- 238000000513 principal component analysis Methods 0.000 claims abstract description 18
- 238000005259 measurement Methods 0.000 claims abstract description 8
- 230000006870 function Effects 0.000 claims description 57
- 239000011159 matrix material Substances 0.000 claims description 50
- 239000013598 vector Substances 0.000 claims description 31
- 230000008569 process Effects 0.000 claims description 26
- 238000000354 decomposition reaction Methods 0.000 claims description 13
- 238000000605 extraction Methods 0.000 claims description 9
- 238000009499 grossing Methods 0.000 claims description 9
- 238000003384 imaging method Methods 0.000 claims description 9
- 230000001133 acceleration Effects 0.000 claims description 7
- 238000004088 simulation Methods 0.000 claims description 7
- 230000003862 health status Effects 0.000 claims description 6
- 238000005070 sampling Methods 0.000 claims description 6
- 230000008859 change Effects 0.000 claims description 4
- 238000010276 construction Methods 0.000 claims description 2
- 230000010355 oscillation Effects 0.000 claims description 2
- 238000012544 monitoring process Methods 0.000 abstract description 8
- 238000012360 testing method Methods 0.000 description 13
- 238000001514 detection method Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 5
- 230000008901 benefit Effects 0.000 description 4
- 230000006378 damage Effects 0.000 description 4
- 230000007613 environmental effect Effects 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 3
- 230000035945 sensitivity Effects 0.000 description 3
- 238000012935 Averaging Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 2
- 238000013459 approach Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000012423 maintenance Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000005856 abnormality Effects 0.000 description 1
- 230000001808 coupling effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000005284 excitation Effects 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 238000012806 monitoring device Methods 0.000 description 1
- 230000008092 positive effect Effects 0.000 description 1
- 238000011158 quantitative evaluation Methods 0.000 description 1
- 230000002787 reinforcement Effects 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
Landscapes
- Wind Motors (AREA)
Abstract
The invention discloses a method for identifying the looseness health state of a bolt of an offshore wind power generation structure, and belongs to the field of health monitoring of marine engineering structures; firstly, constructing multisource fusion characterization parameters of the offshore wind power generation structure in normal and different bolt loosening states, carrying out feature layer fusion and principal component analysis, and establishing health state characterization data of the offshore wind power generation structure; then acquiring service vibration data of the offshore wind power generation tower barrel structure, constructing multisource fusion characterization parameters in a service state, carrying out feature layer fusion and principal component analysis, and calculating projection data of the multisource fusion characterization parameters in a principal component space; finally, obtaining the distance indexes of the projection data and the characterization data of different health states, and fusing decision layers to further realize the accurate identification of the loosening state of the bolt; the scheme does not need to identify modal parameters, so that the calculated amount is greatly reduced; and vibration signals are decomposed and processed based on empirical modes, so that the vibration signal is not limited by the structural type to be tested, the measurement noise interference in a complex environment is reduced, and the practicability and popularization value are high.
Description
Technical Field
The invention belongs to the field of ships and ocean engineering, focuses on an offshore wind power generation tower structure, and particularly relates to a method for identifying the looseness health state of bolts of the offshore wind power generation tower structure.
Background
The offshore wind power generation structure is an infrastructure for developing and utilizing offshore wind energy, with the technical innovation, the connection among towers is more and more, and the bolt connection is the most main connection mode among the current offshore wind power generation tower structures. Under the coupling action of ocean loads such as wind wave current and sea ice, the connecting structure is required to bear more changeable loads, loosening faults are more prone to occur, the integral rigidity of the offshore wind power generation structure is changed, when loosening reaches a certain degree, the integral vibration characteristics of the offshore wind power generation structure are even abnormal, and the offshore wind power generation structure is induced to resonate and collapse even. It is counted that structural failure accidents caused directly or indirectly by bolt loosening are up to about 20% among wind power accidents occurring worldwide each year from 2007-2023.
The bolt of the connecting structure has strong looseness latency, huge harm and difficult detection, and is valued in academia and engineering industry. Aiming at the problem of connection structure/bolt looseness monitoring, technologies of bolt looseness angle real-time monitoring system and method [ CN117288086A ], [ a bolt looseness position detection method based on SVM and AR model transmissibility ] [ CN117113755A ] and [ a wind turbine generator bolt looseness fracture monitoring device and fatigue evaluation method ] [ CN117231438A ] are disclosed, wherein the technologies are respectively used for monitoring the bolt looseness angle, the bolt looseness position and the fracture state. However, the existing method is mainly concentrated in the fields of machinery and land wind power, and aiming at the offshore wind power generation structure, the problems that fault characteristic information in bolt loosening signals of the offshore wind power generation structure is weak, similar working conditions are difficult to distinguish and the like are considered, and the bolt loosening degree cannot be quantitatively identified, so that the related technologies for monitoring the bolt loosening state of the offshore wind power generation structure are less, and in combination with the existing bolt loosening monitoring scheme, quantitative determination of the bolt loosening degree is difficult to realize.
The quantitative evaluation of the looseness degree of the bolts of the offshore wind power generation tower cylinder structure plays a vital role in ensuring the safety of a fan, improving the maintenance efficiency, preventing faults, optimizing the design and improving the management level of the whole wind power industry, so that the quantitative recognition method for the looseness degree of the bolts of the offshore wind power generation tower cylinder structure has the advantages of high design sensitivity, high accuracy and strong anti-interference capability, is suitable for the quantitative recognition method for the looseness of the bolts of the offshore wind power generation tower cylinder structure, realizes the real-time and automatic monitoring and diagnosis of the health state of the connection of the bolts of the tower cylinder structure, provides powerful guarantee for the safe operation of the offshore wind power generation structure, and has positive pushing effect on the development of the offshore wind power industry.
Disclosure of Invention
Aiming at the technical problem that the looseness degree of the bolts of the offshore wind power generation tower barrel structure is difficult to quantitatively identify in the prior art, the invention provides the method for identifying the looseness health state of the bolts of the offshore wind power generation structure, and the early damage judgment and identification of the offshore wind power generation structure are obviously improved by establishing a multi-stage fusion strategy of a data layer, a feature layer and a decision layer, the health state of the structure can still be accurately identified under a higher noise level, and the noise robustness of the health state identification method is effectively improved.
The invention is realized by adopting the following technical scheme: a method for identifying the looseness health state of a bolt of an offshore wind power generation structure comprises the following steps:
step A, constructing multisource fusion characterization parameters of the offshore wind power generation structure in a normal state and in different bolt loosening states, carrying out feature layer fusion and principal component analysis, and establishing different health state characterization data of the offshore wind power generation structure;
the construction of the multisource fusion characterization parameters specifically adopts the following modes:
A11, establishing a finite element model corresponding to the offshore wind power generation structure, and obtaining multisource vibration data of the offshore wind power generation structure in normal and different bolt loosening states according to numerical simulation of the finite element model;
step A12, carrying out sample division on multisource vibration data in normal and different bolt loosening states to obtain a plurality of sample data, carrying out data layer fusion on each sample data, and sequentially carrying out empirical mode decomposition to obtain intrinsic mode function components of each sample;
a13, calculating probability density functions of each eigenmode function component, and carrying out imaging and smoothing treatment on the probability density functions;
step A14, carrying out gridding feature extraction on the image processed in the step A13, and assembling and splicing to construct multisource fusion characterization parameters in a normal state and in different bolt loosening states;
In addition, feature layer fusion and principal component analysis are carried out, and the health state characterization data of the offshore wind tower structure is established, specifically adopting the following modes:
respectively carrying out feature layer fusion on data fusion characterization parameters in normal and slight loosening, moderate loosening and severe loosening states to obtain 、、And；
Integrating the fused data in normal and slight looseness, moderate looseness and severe looseness states to obtain an overall data fusion characterization parameter；
Fusion of characterization parameters for overall dataPerforming centering treatment; Calculating covariance matrixCovariance matrix eigenvectorsIs the direction of the main component; for the data matrix to be normalized, Representing the average value of each row of the matrix Z; a diagonal matrix; A feature vector matrix;
Determining principal component order s, and setting energy threshold So that the front partThe characteristic values satisfy；
Determining the front according to the principal component orderThe feature vector is the principal component vectorForming a principal component space projection matrix
Carrying out principal component projection on the integral feature matrix to obtain characterization data of the health state of the offshore wind tower structureComprising health status characterization data at upper and lower locationsAnd；
Step B, vibration data in the running process of the offshore wind power generation structure is obtained, multisource fusion characterization parameters in the running state are constructed, feature layer fusion and principal component analysis are carried out, and projection data of the multisource fusion characterization parameters in a principal component space are calculated;
vibration data in the running process of the offshore wind power generation tower barrel structure is acquired, and multisource fusion characterization parameters in the running state are constructed, specifically adopting the following modes:
Acquiring corresponding multi-source vibration data in the running process of the offshore wind power generation structure to be detected in real time according to acceleration sensors arranged at the upper position and the lower position in the running process AndAndAnd；
Sample division is carried out on multi-source vibration data in the running process, and a plurality of sample data are obtained; carrying out data layer fusion on each sample data, and sequentially carrying out empirical mode decomposition to obtain intrinsic mode function components of each sample;
Calculating probability density functions of each eigenmode function component, and carrying out imaging and smoothing treatment on the probability density functions;
extracting gridding characteristics of the image, and forming data fusion characterization parameters in the operation process of the upper position and the lower position by assembling and splicing the image AndRepresenting measured data symbols.
Further, in the step B, when calculating projection data, the following method is specifically adopted:
Feature layer fusion is carried out on the operation state data fusion characterization parameters ; Then, fusion characterization parametersPerforming centering treatment; Calculating covariance matrixThe covariance matrix eigenvectors are the principal component directions;
Determining the front according to the principal component order The feature vector is the principal component vectorForming a principal component space projection matrix
Fusion of characterization parameters for overall dataPerforming main component projection to obtain operation state characterization data of the offshore wind tower structureComprising corresponding operating state characterization data at upper and lower positionsAnd；
And C, acquiring the distance indexes of the projection data obtained in the step B and the characterization data of different health states obtained in the step A, and then fusing decision layers, wherein the accurate identification of the looseness health state of the marine power generation structure bolt is realized through the approach classification criterion:
Step C1, calculating the distance index between the projection data of the upper part of the running state and the characterization data of different health states to obtain the first step Individual operating state sample projection data and the firstDistance indicators of characterization data of different health state samples are expressed as; Similarly, the distance index between the position projection data and the characterization data of different health states in the running state is calculated and expressed as；
Step C2, fusing decision layers, and calculating the average value of the distance indexes of the upper position and the lower positionAnd carrying out running state bolt loosening degree identification based on the approaching classification criterion, and classifying and identifying the running state sample bolt loosening degree by taking the minimum distance as a standard measurement.
Compared with the prior art, the invention has the advantages and positive effects that:
The scheme is based on the analysis of the near-end data of the offshore wind power generation tower tube connecting structure, and the near-end data is sensitive to early damage/tiny defects of the structure, so that the judgment can be finished by only arranging a few (such as two) sensors, the detection difficulty and the cost are greatly reduced, and the method has higher guiding significance and practical value for early warning, maintenance and reinforcement decision of the offshore equipment connecting structure; moreover, based on probability density function images, through a probability density function feature extraction method for building gridding and by combining special features and feature vector assembly strategies, the state characterization system of the offshore wind power generation structure is enriched, the sensitivity of feature parameters is improved, and the method has higher sensitivity to slight looseness and early defect identification of the offshore wind power generation structure;
In addition, the interference of environmental factors and measurement noise on the loosening judgment method is eliminated through principal component analysis, only the vibration response of the structure is required to be measured, the current environmental factors are not required to be measured, and the accurate judgment of the health state of the offshore wind power generation connection structure under environmental and noise pollution is effectively realized; compared with the traditional modal parameter detection method, modal parameter identification is not needed in the scheme, so that the calculated amount is greatly reduced; in addition, vibration signals are processed based on an empirical mode decomposition method, interference of actual measurement environmental noise is reduced, signal decomposition can be performed according to time scale characteristics of data, therefore, the method is not limited by a structural type to be detected, characterization parameters are constructed based on multi-source data, structural health monitoring precision is greatly improved, and practicality and popularization value are high.
Drawings
FIG. 1 is a diagram of multi-source acceleration data for upper measurement points in a healthy state according to an embodiment of the present invention, wherein (a) is upper x-axis acceleration data and (b) is upper y-axis acceleration data;
FIG. 2 is a schematic diagram showing bolt distribution according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of health status characterization data of an offshore wind tower structure according to an embodiment of the present invention, wherein (a) is an upper position, (b) is a lower position, and D1, D2, and D3 respectively represent slight, moderate, and severe looseness, and H is a health status;
FIG. 4 is a schematic diagram showing the result of determining the health status of the operation process of the offshore wind turbine tower according to the embodiment of the invention, wherein (a) is the upper position and (b) is the lower position; d1, D2, D3 represent mild, moderate, severe loosening, H is healthy; test represents a Test state, test A is a slightly loosened state of bolt loosening, and Test B is a severely loosened state of four bolt loosening;
FIG. 5 is a schematic diagram of a probability density image according to an embodiment of the present invention.
Detailed Description
In order that the above objects, features and advantages of the invention will be more readily understood, a further description of the invention will be rendered by reference to specific embodiments thereof which are illustrated in the appended drawings. In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention, but the present invention may be practiced otherwise than as described herein, and therefore the present invention is not limited to the specific embodiments disclosed below.
The embodiment of the method for identifying the looseness health state of the bolt of the offshore wind power generation structure comprises the following steps:
step A, constructing multisource fusion characterization parameters of the offshore wind power generation structure in a normal state and in different bolt loosening states, carrying out feature layer fusion and principal component analysis, and establishing different health state characterization data of the offshore wind power generation structure;
step B, vibration data in the running process of the offshore wind power generation tower structure is obtained, multisource fusion characterization parameters in the running state are constructed, feature layer fusion and principal component analysis are carried out, and projection data of the multisource fusion characterization parameters in a principal component space are calculated;
and C, acquiring the distance indexes of the projection data obtained in the step B and the different health state representation data obtained in the step A, fusing decision layers, and accurately identifying the loosening health state of the bolt of the power generation structure through the approaching classification criterion.
The step A is specifically realized by the following modes:
A1, constructing multisource fusion characterization parameters under normal states and different bolt loosening states of an offshore wind tower structure:
(1) Establishing a finite element model corresponding to the offshore wind power generation structure, and obtaining multi-source vibration data of the offshore wind power generation tower barrel structure in a normal state and in different bolt loosening states according to numerical simulation of the finite element model;
Numerical simulation is carried out on the finite element model, continuous vibration time interval data of the upper part and the lower part of the connection structure of the normal state of the offshore wind power generation tower barrel structure are obtained, and the sampling time length is set as S, sampling frequency isHz, continuous acquisitionUpper positionsDirection and sumTo vibration dataAndContinuous collection ofLower positionsDirection and sumTo vibration dataAndUnder the normal state of the offshore wind power generation structure to be detected, the multisource vibration data corresponding to the upper position a and the lower position b are recorded asAndAndAnd；
In addition, through carrying out numerical simulation on the finite element model, continuous vibration time interval data of two typical positions of the upper part and the lower part of the connecting structure under different bolt loosening states of the offshore wind power generation structure are respectively obtained, the three bolt loosening states of slight loosening, moderate loosening and heavy loosening can be divided according to the bolt loosening quantity, the bolt loosening quantity respectively corresponds to (0, 25% ] (25%, 50% ] and (50, 75% ], and the sampling time length is set as the total bolt quantityS, sampling frequency isHz, continuously acquiring and obtaining multisource vibration data of corresponding upper position and lower position of three bolts of offshore wind power generation structure to be detected in loosening stateAndAndAnd、AndAndAndAndAndAndAnd。
(2) Sample division is carried out on multisource vibration data in normal and different bolt loosening states, and a plurality of sample data are obtained;
sample division is carried out on multisource vibration data under normal and different bolt loosening states so as to Dividing the original data for time intervals to obtainData of individual samples, each sample consisting ofThe vibration data is composed; dividing the samples needs to comprehensively consider the time spent by each sample and the data point number factors.
(3) Carrying out data layer fusion on each sample data, and sequentially carrying out empirical mode decomposition to obtain intrinsic mode function components of each sample, wherein the method is specifically as follows:
data layer fusion is carried out on various local oscillation data by the method AndProjecting on an angular bisector formed by two coordinate axis directions, taking the sum of the modes of the two projections as reference data after the fusion of the data layersThe method specifically comprises the following steps:， ; z a and z b are reference data obtained by fusing data layers of an upper position a point and a lower position b point respectively, and are obtained by calculating a point data x a and y a and b point data x b and y b;
Respectively to the upper positions of the connecting structures And a lower positionA kind of electronic deviceThe empirical mode decomposition is sequentially carried out on the sample data, and each sample can be obtained after decompositionEach of the eigenmode function components, the upper position and the lower position samples are obtainedAnd each eigenmode function.
(4) Calculating probability density functions of each eigenmode function component, and carrying out imaging and smoothing treatment on the probability density functions; respectively calculating probability density functions of eigenvalue function components of each layer of each sample to obtain each sampleProbability density function of layer IMF; K is the number of eigenvalue function components of each sample after empirical mode decomposition, and n samples can be obtained altogetherA plurality of eigenmode functions; IMF is an abbreviation for eigenmode function, and may be specifically extended to IMF1, IMF2, … IMFk,
The probability density functions of the 1 st and 2 … k layers IMF are respectively corresponding to each other, the probability density functions are subjected to imaging and smoothing, the probability density functions calculated by the IMF of each layer are subjected to imaging and smoothing by adopting a neighbor point averaging method, burrs caused by noise or data abnormality in the images are removed, noise interference in the signal measurement process is eliminated, and the accuracy of damage identification is improved, namely, the neighbor point averaging method is adoptedThe mean value in the neighborhood replaces the value of each point in the neighborhood: In which, in the process, The average value of the adjacent points is the data points of the points in the neighborhood SIs denoted as g (z), and N is the number of data points in the neighborhood S.
(5) Carrying out gridding feature extraction on the image, assembling and splicing to construct multisource fusion characterization parameters in a normal state and in different bolt loosening states:
The method comprises the steps of performing gridding feature extraction on each processed probability density function image, dividing the images by grids with specific density, replacing data points in the grids by mathematical matrixes with the same dimension, changing the probability density images of signals when the data are changed, and accurately extracting the change of the probability density images through the change of the data points of the grids corresponding to the mathematical matrixes;
Set the first Sample number 1The probability density function of the layer IMF isThe maximum and minimum amplitude values of the amplitude value data in the statistical probability density function are respectivelyAndThe maximum value of probability density isOn the abscissa ofAnd the ordinateEquidistant demarcation within an areaThe grids, R and L respectively correspond to the number of the transverse square and the longitudinal square of the square background when the probability density function is imaged, and a zero matrix is correspondingly definedA zeroing matrix corresponding to the probability density function of the k-th IMF layer, wherein the dimension is R multiplied by L, each element in the matrix corresponds to a grid at a corresponding position in the graph, the number of data points contained in the grid is the number of the element at the corresponding position of the matrix, if the point falls on a grid line, the point is specified to be distributed to an adjacent grid at the right side or the upper side (for example, the grid at the position A in FIG. 5 contains 3 data points, the grid at the position corresponding to the matrix is 3;B and does not contain data points, and the grid at the position corresponding to the matrix is 0);
Extracting features of the imaged data to obtain the first First oneFeature matrix of probability density function of layer IMFAll columns of the grid are combined in sequence and head to form column vector：
，Represents the firstSample number 1The 1 st row and 1 st column data of the feature matrix of the probability density function of the layer IMF are the same as the other similar data; assembling and splicing to form the firstIndividual sample data feature vectors, allIndividual column vectorsAre spliced and combined into columns according to the sequence from head to tail,Splice column vectors for the kth layer, by correspondingEach row is formed by connecting end to end; constructing reference data feature vectors，Splicing column vectors from all k layersComposition, wherein feature vectorsIs comprised ofA high-dimensional vector of the individual features;
Similarly, the health state sample data is subjected to feature extraction, and the upper part and the lower part are assembled and spliced to form the normal state data fusion feature parameters And，AndThe data fusion characteristic parameters of the normal state of the two measuring points corresponding to the a and the b consist of reference data characteristic vectors,Reference data feature vectors corresponding to IMFs of layers 1 and 2 … k respectively corresponding to the point a,Similarly.
Similarly, characteristic extraction is carried out on sample data of different bolt loosening states, and fusion characteristic parameters of the bolt loosening state data with different upper and lower positions are formed by assembling and splicingAnd、AndAndAnd，AndData fusion characteristic parameters corresponding to the slight looseness state of the point a and the point b,AndAndAndRespectively corresponding to moderate loosening and severe loosening states.
Step A2, feature layer fusion and principal component analysis are carried out, and the health state representation data of the offshore wind turbine tower structure are established:
respectively carrying out feature layer fusion on the data fusion characterization parameters of the normal bolt loosening state and the different bolt loosening state 、、And；The matrix respectively corresponds to the probability density characteristics of all sample signals in normal and slight looseness, moderate looseness and severe looseness states;
integrating the data after the fusion of the normal bolt loosening state and the different bolt loosening state to obtain the fusion characterization parameter of the overall data ；
The principal component analysis is carried out on the overall data fusion characterization parameters, and a principal component space projection matrix is calculated, wherein the specific method is as follows:
Fusion of characterization parameters for overall data Performing centering treatment；
Calculating covariance matrixThe covariance matrix eigenvectors are the principal component directions;
Wherein, For the data matrix to be normalized,Representing the average value of each row of the matrix Z; Is composed of Diagonal matrix of individual eigenvalues, an；Is a eigenvector matrix corresponding to the eigenvalue.
Determining the order of main components and setting an energy thresholdSo that the front partThe characteristic values satisfy；
Determining the front according to the principal component orderThe feature vector is the principal component vectorForming a principal component space projection matrix
Carrying out principal component projection on the integral feature matrix to obtain characterization data of the health state of the offshore wind tower structureComprising health status characterization data at upper and lower locationsAnd，Is the transposed matrix of P.
In addition, the step B is specifically realized by the following modes:
Step B1, vibration data in the running process of the offshore wind power generation tower barrel structure is obtained, and multisource fusion characterization parameters in the running state are constructed:
arranging acceleration sensors at the upper part and the lower part in the operation process of the offshore wind power generation tower cylinder connecting structure to be detected, and acquiring multisource vibration data in the service process of the offshore wind power generation tower cylinder connecting structure to be detected in real time AndAndAnd；
Sample division is carried out on multi-source vibration data in the running process, and a plurality of sample data are obtained; then carrying out data layer fusion on each sample data, and sequentially carrying out empirical mode decomposition to obtain intrinsic mode function components of each sample; calculating probability density functions of each eigenmode function component, and carrying out imaging and smoothing treatment on the probability density functions;
extracting gridding characteristics of the image, and forming data fusion characterization parameters in the operation process of the upper position and the lower position by assembling and splicing the image AndWherein, the method comprises the steps of, wherein,AndAnd respectively fusing characterization parameters corresponding to the upper position a and the lower position b, wherein the parameters represent measured data (the same applies below), and n represents n layers of IMFs.
And step B2, carrying out feature layer fusion and principal component analysis, and calculating projection data of the feature layer fusion and principal component analysis in a principal component space:
Feature layer fusion is carried out on the operation state data fusion characterization parameters ; And performing principal component analysis on the running state data fusion characterization parameters, and calculating projection data of the running state data fusion characterization parameters in a principal component space, wherein the method comprises the following steps of:
Fusion of characterization parameters for overall data Performing centering treatment; Calculating covariance matrixThe covariance matrix eigenvectors are the principal component directions;
Determining the front according to the principal component order The feature vector is the principal component vectorForming a principal component space projection matrix
Fusion of characterization parameters for overall dataPerforming main component projection to obtain operation state characterization data of the offshore wind tower structureComprising operational state characterization data at upper and lower positionsAnd。
In the step C, the distance index of the projection data and the characterization data of different health states is obtained, and the specific method is as follows:
Calculating distance indicators of projection data of upper part of running state and characterization data of different health states, such as the first Individual operating state sample projection data and the firstThe distance index of the characterization data of the samples with different health states is specifically as follows；
Similarly, the distance index between the position projection data and the different health state characterization data in the running state is calculated, in particular。
And then fusing decision layers, and realizing accurate identification of the loosening health state of the tower structural bolts through approaching classification criteria:
performing decision layer fusion, and calculating the average value of the upper position and lower position distance indexes And carrying out running state bolt loosening degree identification by using an adjacent classification criterion, and classifying and identifying the running state sample bolt loosening degree by taking the minimum distance as a standard measurement.
In order to make the objects, technical solutions and advantages of the embodiments of the present invention more clear, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.
Taking a 5MW three-pile foundation offshore wind power generation structure as an example, a finite element model is built, multisource vibration data of the offshore wind power generation tower barrel structure in a normal state under the typical wind load effect and in the upper and lower typical positions under different bolt loosening states are analyzed, and the upper position of the connecting structure under the excitation of the wind load in the normal state is shown in fig. 1And the acceleration response in two directions is taken as normal state vibration data for analysis after 60s.
Similarly, multi-source vibration data at typical positions of the upper and lower parts of the offshore wind power generation connection structure in the slightly loosened, moderately loosened and severely loosened states are measured, 8 bolts are connected in the finite element model (fig. 2), symmetry is considered, the slightly loosened (D1) considers 6 types of sample data of bolt 1, bolt 2, bolt 3, bolt 1+2, bolt 1+3 and bolt 1+4, the moderately loosened (D2) considers 6 types of sample data of bolt 1+2+3, bolt 1+3+4, bolt 1+3+6, bolt 1+2+3+4, bolt 1+3+5+7, bolt 1+5+7+8, and the severely loosened (D3) considers 6 types of sample data of bolt 1+2+3+4+5+6, bolt 1+4+6, bolt 1+2+3+5+6, bolt 1+3+4+6+7 and bolt 1+3+4+6+7, and 18 types of sample data are accumulated.
And constructing multisource fusion characterization parameters under the normal state of the offshore wind tower structure and different bolt loosening states by utilizing multisource data, carrying out feature layer fusion and principal component analysis, and establishing the health state characterization data of the offshore wind tower structure, as shown in fig. 3.
Vibration data of the offshore wind power generation tower cylinder structure in the operation process are acquired, two operation process bolt loosening states are considered, namely a Test A (bolt 8 loosening, slight loosening) and a Test B (bolt 1+4+6+7+8 loosening, severe loosening), multisource fusion characterization parameters are constructed in the operation state, feature layer fusion and principal component analysis are carried out, projection data of the offshore wind power generation tower cylinder structure in the principal component space are calculated, the Test A projection data falls into a D1 (slight loosening) cluster, the Test B projection data falls into a D3 (severe loosening) cluster, and the bolt loosening degree of the offshore wind power generation tower cylinder structure in the operation process is accurately identified.
Accurately and quantitatively identifying the bolt looseness degree of sample data in the operation process, calculating the distance indexes of projection data of a sample to be detected and characterization data of different health states, and fusing decision layers, wherein the projection data of Test A (bolt 8 looseness) and the characterization data of No.2 (bolt 1 looseness) are closest in distance by utilizing upper position data as shown in a table 1; the lower position data is utilized, the distance between the lower position data and No.3 representing data (bolt 2 loosening) is closest, the decision layer is used for fusing and comprehensively judging that the distance between Test A projection data and No.3 representing data (bolt 2 loosening) is closest, and the comprehensive judging that the bolt loosening state is slightly loosened (D1) when 1 bolt is loosened is completely consistent with the preset loosening degree, so that the effectiveness of the invention is verified. Notably, test A is that bolt 8 loosens, and is symmetrical with No.3 characterization data (bolt 2 loosens), so that the invention is verified to accurately position a single bolt looseness position or a symmetrical position thereof, and the detection and investigation range is greatly reduced.
Similarly, using the upper position data, the Test B projection data (bolt 1+4+6+7+8 loose) and No.15 representation data (bolt 1+3+4+5+6 loose) are closest to each other, using the lower position data, it can be determined that it is closest to No.16 representation data (bolt 1+4+5+6+8 loose), and by decision layer fusion, the Test B projection data (bolt 1+4+6+7+8 loose) and No.16 (bolt 1+4+5+6+8 loose) representation data are closest to each other, and the total determination that the bolt loose state is the severe looseness (D3) when the bolt is 5 looseness is completely consistent with the preset looseness level, the validity of the present invention is verified.
Table 1 in the operation process, the loosening degree of different bolts approaches to classifying and fusing decision results:
；
The present invention is not limited to the above-mentioned embodiments, and any equivalent embodiments which can be changed or modified by the technical content disclosed above can be applied to other fields, but any simple modification, equivalent changes and modification made to the above-mentioned embodiments according to the technical substance of the present invention without departing from the technical content of the present invention still belong to the protection scope of the technical solution of the present invention.
Claims (10)
1. The method for identifying the looseness health state of the bolt of the offshore wind power generation structure is characterized by comprising the following steps of:
step A, constructing multisource fusion characterization parameters of the offshore wind power generation structure in normal states and in different bolt loosening states, carrying out feature layer fusion and principal component analysis, and establishing different health state characterization data of the offshore wind power generation structure, wherein the bolt loosening states are divided according to bolt loosening quantity, and include slight loosening, moderate loosening and severe loosening;
Step B, vibration data in the running process of the offshore wind power generation structure is obtained, multisource fusion characterization parameters in the running state are constructed, feature layer fusion and principal component analysis are carried out, and projection data of the multisource fusion characterization parameters in a principal component space are calculated;
and C, acquiring the distance indexes of the projection data obtained in the step B and the different health state representation data obtained in the step A, and carrying out decision layer fusion to realize identification of the loosening health state of the bolt of the power generation structure based on the adjacent classification criterion.
2. The method for identifying the looseness health state of the bolts of the offshore wind power generation structure according to claim 1, wherein the method comprises the following steps of: in the step A, the construction of the multisource fusion characterization parameters specifically adopts the following modes:
A11, establishing a finite element model corresponding to the offshore wind power generation structure, and obtaining multisource vibration data of the offshore wind power generation structure in normal and different bolt loosening states according to numerical simulation of the finite element model;
step A12, carrying out sample division on multisource vibration data in normal and different bolt loosening states to obtain a plurality of sample data, carrying out data layer fusion on each sample data, and sequentially carrying out empirical mode decomposition to obtain intrinsic mode function components of each sample;
a13, calculating probability density functions of each eigenmode function component, and carrying out imaging and smoothing treatment on the probability density functions;
and step A14, carrying out gridding feature extraction on the image processed in the step A13, and assembling and splicing to construct multisource fusion characterization parameters in a normal state and in different bolt loosening states.
3. The method for identifying the looseness health state of the bolts of the offshore wind power generation structure according to claim 2, wherein the method comprises the following steps of: in the step a11, the following specific method is adopted:
Numerical simulation is carried out on the finite element model, continuous vibration time interval data of the upper position and the lower position of the connection structure in the normal state of the offshore wind power generation structure are obtained, and the sampling time length is set as Sampling frequency is/>Respectively and continuously collect/>/>, Upper and lower positions a and bDirection and/>Obtaining the/>, of the upper position a of the offshore wind power generation structure to be detected in the normal state, from the vibration dataDirection and/>To multisource vibration data, and/>, lower position bDirection and/>To multisource vibration data, the corresponding is denoted/>, respectivelyAnd/>/>And/>；
Similarly, performing numerical simulation on the finite element model to respectively acquire continuous vibration time course data of the upper position and the lower position of the connecting structure under different bolt loosening states of the offshore wind power generation structure, and continuously acquiring to obtain corresponding upper position and lower position of the offshore wind power generation structure to be detected under the conditions of slight loosening, moderate loosening and heavy looseningDirection and/>To multisource vibration data, respectively corresponding to the representation as/>And/>/>And/>、/>And/>/>And/>/>And/>/>And/>。
4. The method for identifying the looseness and health state of the bolts of the offshore wind power generation structure according to claim 3, wherein the method comprises the following steps of: in the step A12, the multi-source vibration data under the normal and different bolt loosening states are subjected to sample division to obtainDividing the original data for the time interval to obtain/>Sample data, per sample by/>The vibration data is composed;
Comprehensively considering the time and data point data factors of each sample when dividing the samples; then, data layer fusion is carried out on various local oscillation data by the method And/>Projecting the two coordinate axis directions on an angular bisector formed by the two coordinate axis directions to obtain the datum data/>, after the data layers are fusedExpressed as: /(I)，/>；
Respectively to the upper positions of the power generation structuresAnd lower part/>/>The empirical mode decomposition is sequentially carried out on the sample data, and each sample can be obtained after decompositionEach of the eigenmode function components, the upper position and the lower position samples are obtained/>And each eigenmode function.
5. The method for identifying the looseness and health state of the bolts of the offshore wind power generation structure according to claim 4, wherein the method comprises the following steps of: in the step A13, probability density functions of eigenvalue function components of each layer of each sample are calculated respectively to obtain each sampleProbability Density function of layer IMF/>；/>And (3) respectively corresponding to probability density functions of IMFs of the 1 st layer and the 2 … k layer, and carrying out imaging and smoothing treatment on the probability density functions obtained by calculating the IMFs of each layer by adopting a near point average method.
6. The method for identifying the looseness and health state of the bolts of the offshore wind power generation structure according to claim 5, wherein the method comprises the following steps of: in the step a14, gridding feature extraction is performed on each processed probability density function image, the images are divided by grids with specific density, the data points in the grids are replaced by mathematical matrixes with the same dimension, when the data are changed, the probability density images of the signals are changed along with the data points of the grids, and the change of the probability density images can be accurately extracted through the change of the data points of the grids corresponding to the mathematical matrixes;
Set the first Sample No./>The probability density function of the layer IMF is/>The maximum and minimum amplitude values of the amplitude data in the statistical probability density function are respectively/>And/>The probability density maximum is/>On the abscissa/>And ordinate/>Equidistant demarcation/>A grid of zero matrix/>, correspondingly definedEach element in the matrix corresponds to a grid at a corresponding position in the graph, wherein the number of data points contained in the grid is the numerical value of the element at the corresponding position of the matrix, and if the point falls on a grid line, the point is specified to be distributed to an adjacent grid on the right side or the upper side;
Extracting features of the imaged data to obtain the first Second/>Feature matrix/>, of probability density function of layer IMFAll columns of the grid are combined in sequence and head to form column vector/>：
Wherein/>Represents the/>Sample No./>Line 1 and column 1 data of the feature matrix of the probability density function of the layer IMF …/>Represents the/>Sample No./>The R row and L column data of the feature matrix of the probability density function of the IMF layer are assembled and spliced to form the/>Individual sample data feature vectors, will all/>Individual column vectors/>Are spliced and combined into a column according to the sequence from beginning to end to form the reference data feature vectorWherein the feature vector/>For containing/>A high-dimensional vector of the individual features;
and similarly, extracting the characteristics of the health state sample data, and assembling and splicing to form the upper and lower normal state data fusion characteristic parameters And/>Characteristic extraction is carried out on sample data in different bolt loosening states, and the data are assembled and spliced to form corresponding data fusion characteristic parameters/>, under the conditions of slight loosening, moderate loosening and severe loosening, of the upper position and the lower positionAnd/>、/>And/>/>And/>。
7. The method for identifying the looseness and health state of the bolts of the offshore wind power generation structure according to claim 6, wherein the method comprises the following steps of: in the step A, feature layer fusion and principal component analysis are carried out, and the health state representation data of the offshore wind turbine tower structure is established, specifically adopting the following modes:
respectively carrying out feature layer fusion on data fusion characterization parameters in normal and slight loosening, moderate loosening and severe loosening states to obtain 、/>、/>And/>；
Integrating the fused data in normal and slight looseness, moderate looseness and severe looseness states to obtain an overall data fusion characterization parameter；
Fusion of characterization parameters for overall dataPerform the centering treatment/>; Calculating covariance matrixCovariance matrix eigenvectors/>Is the direction of the main component; /(I)For the data matrix to be normalized,Representing the average value of each row of the matrix Z; /(I)A diagonal matrix; /(I)A feature vector matrix;
Determining principal component order s, and setting energy threshold So that front/>The individual eigenvalues satisfy/>；
Determining the front according to the principal component orderThe individual feature vectors are principal component vectors/>Forming a principal component space projection matrixCarrying out principal component projection on the integral feature matrix to obtain characterization data/>, of the health state of the offshore wind tower structureComprising health status characterization data/>, at upper and lower locationsAnd/>。
8. The method for identifying the looseness health state of the bolts of the offshore wind power generation structure according to claim 1, wherein the method comprises the following steps of: in the step B, vibration data in the running process of the offshore wind power generation tower structure is obtained, and multisource fusion characterization parameters in the running state are constructed, specifically:
Acquiring corresponding multi-source vibration data in the running process of the offshore wind power generation structure to be detected in real time according to acceleration sensors arranged at the upper position and the lower position in the running process And/>/>And/>；
Sample division is carried out on multi-source vibration data in the running process, and a plurality of sample data are obtained; carrying out data layer fusion on each sample data, and sequentially carrying out empirical mode decomposition to obtain intrinsic mode function components of each sample;
Calculating probability density functions of each eigenmode function component, and carrying out imaging and smoothing treatment on the probability density functions;
extracting gridding characteristics of the image, and forming data fusion characterization parameters in the operation process of the upper position and the lower position by assembling and splicing the image And/>Representing measured data symbols.
9. The method for identifying the looseness health state of the bolts of the offshore wind power generation structure according to claim 1, wherein the method comprises the following steps of: in the step B, when projection data is calculated, the method is specifically implemented by the following steps:
Feature layer fusion is carried out on the operation state data fusion characterization parameters ; Then for fusion characterization parameters/>Perform the centering treatment/>; Calculate covariance matrix/>The covariance matrix eigenvectors are the principal component directions;
Determining the front according to the principal component order The individual feature vectors are principal component vectors/>Forming a principal component space projection matrix; Fusion of characterization parameters to overall dataPerforming principal component projection to obtain running state characterization data/>, of the offshore wind tower structureComprising corresponding operating state characterization data/>, at an upper position and a lower positionAnd/>。
10. The method for identifying the looseness health state of the bolts of the offshore wind power generation structure according to claim 1, wherein the method comprises the following steps of: the step C is specifically realized by adopting the following modes:
Step C1, calculating the distance index between the projection data of the upper part of the running state and the characterization data of different health states to obtain the first step Individual run state sample projection data and/>Distance indicators of characterization data of samples of different health states are expressed as/>; Similarly, the distance index between the position projection data and the characterization data of different health states in the running state is calculated and expressed as/>；
Step C2, fusing decision layers, and calculating the average value of the distance indexes of the upper position and the lower positionAnd carrying out running state bolt loosening degree identification based on the approaching classification criterion, and classifying and identifying the running state sample bolt loosening degree by taking the minimum distance as a standard measurement.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410417226.5A CN118013808B (en) | 2024-04-09 | Method for identifying looseness health state of bolt of offshore wind power generation structure |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410417226.5A CN118013808B (en) | 2024-04-09 | Method for identifying looseness health state of bolt of offshore wind power generation structure |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN118013808A true CN118013808A (en) | 2024-05-10 |
| CN118013808B CN118013808B (en) | 2024-07-09 |
Family
ID=
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105910806A (en) * | 2016-05-30 | 2016-08-31 | 重庆大学 | Filling pump early health status monitoring method |
| CN107977301A (en) * | 2017-11-21 | 2018-05-01 | 东软集团股份有限公司 | Detection method, device, storage medium and the electronic equipment of unit exception |
| US20210123830A1 (en) * | 2019-10-24 | 2021-04-29 | National Chung Cheng University | Machine tool health monitoring method |
| CN116070527A (en) * | 2023-03-07 | 2023-05-05 | 南京航空航天大学 | Milling cutter residual life prediction method based on degradation model |
| CN116226974A (en) * | 2023-01-06 | 2023-06-06 | 深圳防灾减灾技术研究院 | Damage identification method and system based on data fusion and self-adaptive sparse regularization |
| CN116910879A (en) * | 2023-08-03 | 2023-10-20 | 大连理工大学 | Cable force abnormality diagnosis method and device for bridge stay cable under random vehicle-mounted action |
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105910806A (en) * | 2016-05-30 | 2016-08-31 | 重庆大学 | Filling pump early health status monitoring method |
| CN107977301A (en) * | 2017-11-21 | 2018-05-01 | 东软集团股份有限公司 | Detection method, device, storage medium and the electronic equipment of unit exception |
| US20210123830A1 (en) * | 2019-10-24 | 2021-04-29 | National Chung Cheng University | Machine tool health monitoring method |
| CN116226974A (en) * | 2023-01-06 | 2023-06-06 | 深圳防灾减灾技术研究院 | Damage identification method and system based on data fusion and self-adaptive sparse regularization |
| CN116070527A (en) * | 2023-03-07 | 2023-05-05 | 南京航空航天大学 | Milling cutter residual life prediction method based on degradation model |
| CN116910879A (en) * | 2023-08-03 | 2023-10-20 | 大连理工大学 | Cable force abnormality diagnosis method and device for bridge stay cable under random vehicle-mounted action |
Non-Patent Citations (2)
| Title |
|---|
| RENXIANG CHEN ET AL.: "Looseness diagnosis method for connecting bolt of fan foundation based on sensitive mixed-domain features of excitation-response and manifold learning", NEUROCOMPUTING, vol. 219, no. 5, 31 January 2017 (2017-01-31), pages 376 - 388 * |
| 孔祥娜: "基于两路信号融合分析的螺栓松动故障诊断方法研究", 中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑, vol. 2017, no. 03, 15 March 2017 (2017-03-15), pages 029 - 342 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110704801B (en) | Bridge cluster structure operation safety intelligent monitoring and rapid detection complete method | |
| CN101660969B (en) | Intelligent fault diagnosis method for gear box | |
| CN103076394B (en) | Safety evaluation method for ocean platform based on integration of vibration identification frequencies and vibration mode | |
| CN114559819B (en) | Electric automobile battery safety early warning method based on signal processing | |
| CN104198138B (en) | Early warning method and system for abnormal vibration of wind driven generator | |
| CN117076935B (en) | Digital twin-assisted mechanical fault data lightweight generation method and system | |
| CN110319957A (en) | The irregular exceptional value method for diagnosing faults of Ship Structure stress monitoring system sensor | |
| CN115526515A (en) | Safety monitoring system of gate for water conservancy and hydropower | |
| CN110186384B (en) | Ship structure stress monitoring system sensor deviation fault diagnosis method | |
| CN109781863B (en) | Structure corrosion two-stage detection method and system based on rapid vibration test | |
| CN107977679A (en) | Method based on frequency response function and operation response characteristic diagnosis of complex device initial failure | |
| CN111311872A (en) | Long-term monitoring and alarming system for stress of hull structure | |
| CN118013808B (en) | Method for identifying looseness health state of bolt of offshore wind power generation structure | |
| CN118013808A (en) | Method for identifying looseness health state of bolt of offshore wind power generation structure | |
| CN109974837B (en) | Ship structure damage identification method based on rule reasoning | |
| CN114881527B (en) | Online visual intelligent management system that traces back of car intelligence equipment | |
| CN114139383A (en) | Dynamic fatigue reliability risk assessment method for fuel cell vehicle | |
| CN115235542A (en) | Ship reception chamber online monitoring method and monitoring system for ship lift | |
| CN118053046B (en) | Fusion characterization and judgment method for health state of offshore wind power generation connection structure | |
| CN115436037A (en) | Transmission tower health state discrimination method and device based on SSI parameter identification | |
| CN110553807B (en) | Open-circuit fault diagnosis method for sensor of ship structure stress monitoring system | |
| CN114492636A (en) | Transformer winding state signal acquisition system | |
| Li et al. | Research on intelligent structural health monitoring system | |
| CN112161785A (en) | Ocean engineering structure micro-damage judgment method | |
| CN116108726B (en) | Ocean fixed platform inspection method, device and equipment |
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 |