CN116046980B - Structural fatigue damage diagnosis method based on strain monitoring - Google Patents
Structural fatigue damage diagnosis method based on strain monitoring Download PDFInfo
- Publication number
- CN116046980B CN116046980B CN202211623049.3A CN202211623049A CN116046980B CN 116046980 B CN116046980 B CN 116046980B CN 202211623049 A CN202211623049 A CN 202211623049A CN 116046980 B CN116046980 B CN 116046980B
- Authority
- CN
- China
- Prior art keywords
- strain
- crack
- data
- monitoring
- fatigue damage
- 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
- 238000012544 monitoring process Methods 0.000 title claims abstract description 100
- 238000000034 method Methods 0.000 title claims abstract description 74
- 238000003745 diagnosis Methods 0.000 title claims abstract description 21
- 230000008569 process Effects 0.000 claims abstract description 48
- 238000012549 training Methods 0.000 claims abstract description 18
- 238000000605 extraction Methods 0.000 claims abstract description 6
- 238000012360 testing method Methods 0.000 claims description 28
- 238000004088 simulation Methods 0.000 claims description 16
- 230000008859 change Effects 0.000 claims description 15
- 230000035515 penetration Effects 0.000 claims description 8
- 238000010206 sensitivity analysis Methods 0.000 claims description 8
- 238000009661 fatigue test Methods 0.000 claims description 7
- 230000000977 initiatory effect Effects 0.000 claims description 5
- 238000005259 measurement Methods 0.000 claims description 5
- 230000009471 action Effects 0.000 claims description 3
- 230000036541 health Effects 0.000 abstract description 21
- 230000006870 function Effects 0.000 description 18
- 238000004422 calculation algorithm Methods 0.000 description 13
- 239000013598 vector Substances 0.000 description 8
- 238000010801 machine learning Methods 0.000 description 7
- 239000011159 matrix material Substances 0.000 description 7
- 230000003247 decreasing effect Effects 0.000 description 6
- 238000013528 artificial neural network Methods 0.000 description 5
- 230000003287 optical effect Effects 0.000 description 5
- 238000012417 linear regression Methods 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 3
- 230000006872 improvement Effects 0.000 description 3
- 238000013507 mapping Methods 0.000 description 3
- 230000004044 response Effects 0.000 description 3
- 101001095088 Homo sapiens Melanoma antigen preferentially expressed in tumors Proteins 0.000 description 2
- 102100037020 Melanoma antigen preferentially expressed in tumors Human genes 0.000 description 2
- 230000001174 ascending effect Effects 0.000 description 2
- 238000005336 cracking Methods 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 239000004973 liquid crystal related substance Substances 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 230000000750 progressive effect Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000001960 triggered effect Effects 0.000 description 2
- 238000012795 verification Methods 0.000 description 2
- 229910001250 2024 aluminium alloy Inorganic materials 0.000 description 1
- 238000007476 Maximum Likelihood Methods 0.000 description 1
- 238000009825 accumulation Methods 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000002939 conjugate gradient method Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 239000006185 dispersion Substances 0.000 description 1
- 238000013210 evaluation model Methods 0.000 description 1
- 230000005284 excitation Effects 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 230000000644 propagated effect Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 210000001550 testis Anatomy 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N33/00—Investigating or analysing materials by specific methods not covered by groups G01N1/00 - G01N31/00
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01B—MEASURING LENGTH, THICKNESS OR SIMILAR LINEAR DIMENSIONS; MEASURING ANGLES; MEASURING AREAS; MEASURING IRREGULARITIES OF SURFACES OR CONTOURS
- G01B7/00—Measuring arrangements characterised by the use of electric or magnetic techniques
- G01B7/16—Measuring arrangements characterised by the use of electric or magnetic techniques for measuring the deformation in a solid, e.g. by resistance strain gauge
- G01B7/18—Measuring arrangements characterised by the use of electric or magnetic techniques for measuring the deformation in a solid, e.g. by resistance strain gauge using change in resistance
-
- G01N33/0083—
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/23—Design optimisation, verification or simulation using finite element methods [FEM] or finite difference methods [FDM]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/04—Ageing analysis or optimisation against ageing
-
- 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
Abstract
The invention relates to a structural fatigue damage diagnosis method based on strain monitoring, which comprises the following steps: determining a strain monitoring position, obtaining strain peak value data of the strain monitoring position, and performing feature extraction on the strain peak value data to obtain dimensionless strain peak value relative error percentage data; setting a plurality of levels of threshold values based on the strain peak value relative error percentage data of the monitoring position, and judging whether fatigue damage is generated at the strain monitoring position; and establishing a Gaussian process regression model, training the Gaussian process regression model through the strain peak value relative error percentage data and crack information, and identifying crack length based on the trained Gaussian process regression model to finish fatigue damage diagnosis. According to the structural fatigue damage diagnosis method, the structural health state can be monitored, and the structural fatigue damage degree can be accurately identified.
Description
Technical Field
The invention relates to the technical field of fatigue damage monitoring, in particular to a structural fatigue damage diagnosis method based on strain monitoring.
Background
Structural fatigue damage diagnosis is an important component of the aircraft structure SPHM. There are a variety of structural health monitoring means including active and passive monitoring means. Research on structural fatigue damage diagnosis is carried out on the basis of a plurality of monitoring means, and appropriate structural fatigue damage diagnosis strategies need to be formulated for specific sensors because the data types and the characteristics of the different types of sensors are different. Most of active monitoring means can only acquire wave signals by applying excitation, and continuous monitoring is difficult, so that the application of the active monitoring means is limited. In the passive monitoring means, the strain fatigue damage accumulation effect of the structural key points is sensitive and easy to monitor, and the strain sensor performance parameters are relatively stable, has high response speed and is convenient for continuously monitoring the structural fatigue damage. Thus, structural strain based health monitoring remains one of the most promising methods at present.
The strain monitoring of the aircraft structure is mainly applied in two aspects by collecting strain histories of key parts in real time through strain sensors (traditional strain gauges or optical fiber sensors and the like) to track strain distribution and change information. On the one hand, the load equation of the key part of the aircraft structure can be constructed, verified and corrected through actually measured strain data, and the prediction accuracy of the local load or stress of the key part is improved. On the other hand, based on the strain history of the monitoring part, a mapping relation between the strain and the structural damage is constructed, and the structural fatigue damage is directly monitored and diagnosed. The method has the advantages of quick response, higher precision and the like in the aspects of structural fatigue damage monitoring and diagnosis, is suitable for structural on-line health monitoring, and is less in application. Therefore, in order to expand the application of strain monitoring to the on-line health monitoring of a structure, the invention provides a structural damage diagnosis method based on strain monitoring.
Along with the improvement of the precision of the finite element software simulation technology, the precision of crack propagation simulation is higher and higher, the research of the relationship between local strain response and crack propagation is deeper and deeper, the determination of the strain-crack sensitive position is greatly promoted, the possibility is provided for optimizing the sensor arrangement, and the accuracy of strain monitoring is improved. According to the linear elasticity assumption in fracture mechanics, when the structure is not damaged, the strain value of a certain measuring position is almost unchanged under the same load state, and after a crack is expanded to a certain extent, according to different crack expansion forms, local strain near the crack can be increased, decreased, increased and decreased or increased and increased. Accordingly, the structural health state can be monitored, and fatigue cracks can be found in time.
The traditional structural fatigue damage diagnosis method mainly comprises manual detection, and has high manual/time cost and low precision and reliability. Crack propagation is a nonlinear behavior, and different strain combinations exist in different crack propagation stages, so that the strain-crack mapping relation is difficult to express in the crack propagation process. The strain-based structure monitoring method solves the problems of difficult and low efficiency of manually collecting the fatigue damage information of the structure, and the machine learning algorithm has the advantages of high efficiency/precision and the like when acquiring the complex object relationship. Therefore, strain-crack information can be deeply excavated based on strain monitoring and combining a machine learning algorithm, so that the structural health state can be monitored timely, accurately, automatically and intelligently, and structural fatigue damage can be identified.
Disclosure of Invention
The invention aims to provide a structural fatigue damage diagnosis method based on strain monitoring, which not only can monitor the structural health state, but also can accurately identify the structural fatigue damage degree.
In order to achieve the above object, the present invention provides the following solutions:
a structural fatigue damage diagnosis method based on strain monitoring, comprising:
determining a strain monitoring position, obtaining strain peak value data of the strain monitoring position, and performing feature extraction on the strain peak value data to obtain dimensionless strain peak value relative error percentage data;
setting a plurality of levels of threshold values based on the strain peak value relative error percentage data of the monitoring position, and judging whether fatigue damage is generated at the strain monitoring position;
and establishing a Gaussian process regression model, training the Gaussian process regression model through the strain peak value relative error percentage data and crack information, and identifying crack length based on the trained Gaussian process regression model to finish fatigue damage diagnosis.
Preferably, acquiring strain peak data of the strain monitoring location includes:
and performing crack expansion simulation through finite element software, obtaining information of a strain peak value, a crack and a load cycle number, performing strain-crack sensitivity analysis, selecting a position with larger and uniform change for monitoring, obtaining a strain monitoring position in the crack expansion simulation process, performing a strain monitoring fatigue test based on the strain monitoring position, and obtaining strain peak value data and crack information.
Preferably, the performing crack propagation simulation by the finite element software includes:
under the action of a pulling-pulling fatigue load, initiating a fatigue crack by a sample, and dividing the crack into a crack on one side of a center hole and a crack on two sides of the center hole based on the evolution of the fatigue crack;
and carrying out local encryption and simulation test loading on the crack propagation area grid on the single-side penetration cracks of the central hole and the double-side penetration cracks of the central hole through the finite element software.
Preferably, performing the simulated test loading comprises:
one end of the specimen is completely fixed, and the other end of the specimen only retains freedom in the length direction and applies a constant-amplitude fatigue load.
Preferably, the strain-crack sensitivity analysis based on the strain, crack and load cycle number information comprises:
based on the increase of the load circulation number, the crack is continuously expanded, the strain field of stress near the crack is changed, an x-y rectangular coordinate system is established on the sample by taking the circle center as the origin of coordinates, and strain peaks at adjacent monitoring positions along the y axis and the x axis are respectively selected for sensitivity analysis.
Preferably, acquiring the strain peak data and the crack information includes:
and (3) sticking a strain gauge at the strain monitoring position, acquiring strain peak value data by adopting a dynamic strain gauge, applying a tensile-tensile fatigue load to the sample, loading by adopting a sine wave, observing the surface of the pattern, and acquiring the crack information.
Preferably, the method for obtaining dimensionless data is as follows:
wherein delta i Is the non-dimensional data of the data,for the strain reference value, Δε i Is the difference between the strain measurement and the reference value.
Preferably, determining whether the strain monitoring location is experiencing fatigue damage includes:
wherein n is 1 For the number exceeding the set threshold in the measured data, n 2 To measure the total number of data, P is confidence, delta i Is the non-dimensional data of the data,is a multi-level threshold;
setting a probability value, if the dimensionless data is larger than the confidence coefficient of the threshold value, judging that the strain monitoring position generates fatigue damage if the confidence coefficient of the threshold value exceeds the probability value, and if the confidence coefficient does not exceed the probability value, continuing to the next level threshold value, and judging whether the strain monitoring position generates fatigue damage or not.
Preferably, the expression of the gaussian process regression model is:
wherein n is the number of input sample points, f (·) is a function corresponding to X, ε is white noise σ n Is the variance.
The beneficial effects of the invention are as follows:
according to the structural fatigue damage diagnosis method based on strain monitoring, according to different crack expansion forms, local strain near the crack can be increased, decreased, increased before decreased or increased after decreased before increased, and the like, according to the phenomenon, the structural health state can be monitored, fatigue cracks can be found in time, a strain-crack mapping relation is established by combining a machine learning algorithm, and structural fatigue damage can be accurately identified. In a word, the method can monitor the structural health state in time, accurately, automatically and intelligently, identify the structural fatigue damage, and provide support for the application of strain monitoring in structural health monitoring and fatigue damage diagnosis.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions of the prior art, the drawings that are needed in the embodiments will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the present invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of a structural fatigue damage diagnosis method based on strain monitoring according to an embodiment of the present invention;
FIG. 2 is a sample size and notch location for an embodiment of the present invention;
wherein, (a) is the size of the sample, and (b) is the notch position;
FIG. 3 is a finite element modeling crack propagation result for an embodiment of the present invention;
FIG. 4 is a graph showing the change of the normal stress cloud graph along the load direction along with the crack length according to the embodiment of the invention;
wherein, (a) is a normal stress cloud image change when the crack length is 2.32mm, (b) is a normal stress cloud image change when the crack length is 3.82mm, (c) is a normal stress cloud image change when the crack length is 5.32mm, and (d) is a normal stress cloud image change when the crack length is 8.32 mm;
FIG. 5 shows strain peak histories at different monitoring positions according to the embodiments of the present invention in cases 1 and 2. Establishing an x-y rectangular coordinate system on the sample by taking the circle center as a coordinate origin, and comparing adjacent position strains;
wherein, (a) is the strain peak variation in case 1 (-5.5,2), (-5.5,4) and (-5.5,6), (b) is the strain peak crack growth variation in case 1 (-5.5,4), (-8.5,4) and 0 (-11.5,4), and (c) takes (-11.5,0) as the monitoring location;
FIG. 6 is a strain monitoring position according to an embodiment of the present invention;
FIG. 7 is a graph showing the results of fatigue test cracking as a function of life for an embodiment of the present invention;
FIG. 8 shows the strain peak histories for sample 1# and finite element modeling case 1 according to an embodiment of the present invention;
wherein (a) is the strain peak history of sample 1#, and (b) is the strain peak history of the finite element simulation 1
FIG. 9 is a graph showing the variation of the strain peak relative error percentage with fatigue life for six strain monitoring positions for samples 1# through 6# according to an embodiment of the present invention;
FIG. 10 is a plot of the relative error percent of the TH 1-monitoring position strain peak and crack length as a function of fatigue life for samples 2#, 4# and 5# of the examples of the present invention;
FIG. 11 is a graph showing crack lengths of samples # 2, # 4, # 5, and # 6 identified by the Gaussian process regression model of the example of the present invention;
FIG. 12 is a graph showing crack lengths of sample No. 1 identified by a Gaussian process regression model of an embodiment of the invention;
FIG. 13 is a graph showing the correlation between the crack length of sample No. 1 identified by different algorithms and the crack length identified by different machine learning algorithms and test results in accordance with an embodiment of the present invention;
wherein, (a) is the crack length of the sample No. 1 identified by different algorithms, and (b) is a correlation diagram of the crack length identified by different algorithms and the test result;
FIG. 14 is a graph showing crack lengths of sample 3# identified by 6 monitoring position strain peak histories and Gaussian process regression for sample 3# in accordance with an embodiment of the present invention;
wherein, (a) is the strain peak history of 6 monitoring positions of sample 3#, and (b) is the crack length of sample 3# identified by gaussian process regression.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
In order that the above-recited objects, features and advantages of the present invention will become more readily apparent, a more particular description of the invention will be rendered by reference to the appended drawings and appended detailed description.
As shown in fig. 1, a structural fatigue damage diagnosis method based on strain monitoring is characterized by comprising the following steps: determining a strain monitoring position, obtaining strain peak value data of the strain monitoring position, and performing feature extraction on the strain peak value data to obtain dimensionless strain peak value relative error percentage data; setting a plurality of levels of threshold values, respectively comparing the dimensionless data with the threshold values, and judging whether fatigue damage occurs at the strain monitoring position; and establishing a Gaussian process regression model, training the Gaussian process regression model through the dimensionless data and the crack information, and identifying the crack length based on the trained Gaussian process regression model.
And (3) data acquisition:
as shown in FIG. 2, 6 pieces of the test pieces were used, and the material was 2024 aluminum alloy sheet 3mm thick, 100mm×30mm×3mm in geometry, and 6mm in diameter in the L-T direction, as shown in FIG. 2 (a). Of these, 3 were prepared with a notch of 0.8mm length on the side of the center Kong Bianshan, see FIG. 2 (b). The purpose is to initiate fatigue cracks at this location, and the other 3 pieces are not subjected to the pre-notch treatment.
In this example, conventional resistive strain gages were used for structural health monitoring, and in order to determine the strain gage placement location, FRANC3D and ABAQUS finite element software was used for crack growth simulation. And extracting strain information of the key part, analyzing strain changes of different positions in the crack propagation process, performing strain-crack sensitivity analysis, determining a strain monitoring position, and taking the strain monitoring position as a basis for pasting a strain sensor. Carrying out strain monitoring fatigue test and collecting strain and crack information, wherein the strain monitoring fatigue test comprises the following steps:
under the action of a tensile-tensile fatigue load, the sample finally penetrates the thickness of the sample to be changed into a penetration crack no matter where the fatigue crack is initiated. Therefore, considering the extreme damage case, crack growth simulation was performed in two cases, case 1: 0.8mm penetration crack exists on one side of the central hole, and the condition 2 is that: the center hole had 0.8mm penetration cracks on both sides. Fatigue crack growth simulation was performed using ABAQUS and FRANC3D software. The crack propagation area is locally encrypted, and is automatically divided by the FRANC3D, and the grid encryption is used for improving the calculation accuracy. Strain gage adhering position areaThe grid is evenly divided by hexahedron with the size of 0.5mm by 0.3mm, and the grid attribute is C3D8R. The other regional grids are hexahedrons with the size of 1mm by 0.5mm by 0.3mm, and the grid attribute is C3D8R. And (3) simulating test loading, wherein one end of the sample is completely fixed, and the other end only retains the freedom degree along the length direction. A constant amplitude fatigue load was applied at the end where the degree of freedom was maintained, the load peak was 9.45kN and the stress ratio was 0.1. The crack length is determined by Paris formula, the elastic modulus E is 70GPa, the Poisson ratio C is 0.33, the material parameters C and n are 3.7435E-12 and 3.0167 respectively, and the fatigue crack growth threshold value delta K is obtained th ＝1.58MPa·m 1/2 Fracture toughness K C ＝34.79MPa·m 1/2 . In both cases, inputting the parameters into finite element software, and calculating the variation of the crack distribution with the cycle number, as shown in fig. 3; wherein the load parameters include only: peak and stress ratio.
Fig. 4 is a stress cloud graph during crack propagation, and it can be seen that as the number of load cycles increases, the crack is continuously propagated, and the stress-strain field near the crack is significantly changed, which makes it possible to use a strain monitoring mode.
The strain monitoring location is determined taking into account several factors: (1) sensitivity of strain monitoring location; (2) The limitation of the pasting technology cannot be directly pasted on the crack propagation path; (3) Symmetry of geometry, constraints and crack propagation during use of the actual structure. The center of the circle is used as the origin of coordinates, and an x-y rectangular coordinate system is established as shown in fig. 6. And (3) according to factors (1) and (2), respectively selecting strain peaks at adjacent monitoring positions along the y axis and the x axis for sensitivity analysis. Along the y-axis, as shown in FIG. 5 (a), the strain in case 1 (-5.5,4) is large relative to (-5.5,2) and uniform compared to (-5.5,6). Similarly, the same is true for the strain peak contrast along the y-axis elsewhere on the x-axis. Thus, strain at y=4 is primarily considered to be relatively sensitive to cracking. Comparing along y=4, as shown in fig. 5 (b), the strain at (-5.5,4) in case 1 is greatly and uniformly changed with crack growth compared to the adjacent position. Thus (-5.5,4) was taken as a monitoring location. In comparison between case 1 and case 2, the difference in strain peak is more pronounced at y=0, whereas the strain peak at (-11.5,0) position is larger and positive as shown in fig. 5 (c), so (-11.5,0) is taken as a monitoring position. The actual crack growth is affected by a number of factors, which cause the crack growth path to shift. Thus, the y-axis symmetric position (5.5,4) of (-5.5,4) is taken as a one-place monitoring position, according to the factor (3). Since the test piece has a thickness of 3mm, three positions (-5.5,4), (5.5,4) and (-11.5,0) are symmetrical to the other side of the test piece in consideration of more complicated use cases such as crack initiation positions, shapes and the like, and the final determined positions are shown in fig. 6. Where TH1, TH2 and TH3 are on one side of the sample and TH4, TH5 and TH6 are on the other side of the sample.
Fatigue test: the strain gauge paste was performed at the position shown in fig. 6, the strain gauge size was BX120-1AA, the resistance was 120±0.1Ω, and the sensitive gate size was 1mm×1mm. Strain data acquisition was performed using a DH5921 dynamic strain gauge with a sampling frequency of 500Hz. Clamping the sample on an MTS810-100kN electrohydraulic servo fatigue testing machine to apply a pull-pull fatigue load, wherein the stress peak value is sigma max The stress ratio r=0.1, using sine wave loading, room temperature atmosphere, frequency 10Hz, =105 MPa. During the test, the surface of the test piece was observed by using an optical microscope with a magnification of 5 times, and the crack length was measured.
And (3) data processing:
data preprocessing: and extracting strain peak value data, and corresponding the strain, crack and load cycle number information.
Feature extraction: the strain value epsilon of a certain measuring position under the same load state is assumed that the structure is not damaged i Almost unchanged, and calculating a strain reference value under the same load state after the load is stable:
wherein, the liquid crystal display device comprises a liquid crystal display device,for the strain reference value, i=1, 2,3, …, n is the number of strain measurements under the same load condition.
The difference between the strain measurement and the reference value is:
the relative error percentages of the strain measurement and the reference value are:
delta to be dimensionless i The method is used for feature extraction, and is used for structural health state monitoring and damage degree assessment, and specifically comprises the following steps:
crack length: as shown in FIG. 7, 2 samples of the 3 samples with unilateral notches, numbered 1# and 2# are seen, and when the samples are broken, only unilateral cracks exist, and the 3 samples with notches and the 3 samples without notches, numbered 4# and 5# are both side cracks.
Strain: the strain data was collected using a dynamic strain gauge, and as shown in fig. 8 (a), the strain peak history extracted at six monitoring positions was shown as sample 1# for example, and it was found that there was little change in the strain peak when no fatigue crack was generated in the sample. And once a crack occurs, strain gage information for each location begins to change. The strain peak history sample 1# of the finite element simulation case 1 as shown in fig. 8 (b) is similar to the case 1 of the finite element simulation. Considering symmetry of the penetration crack, only TH1, TH3 and TH5 position strain peaks are of interest here. The TH1 position strain peak value of the sample No. 1 gradually rises, and the change is relatively large; the strain peak value at the TH3 position is in a descending trend and changes relatively slowly; the TH5 position strain peak value has a decreasing trend and relatively large change, which is consistent with the trend obtained by calculation simulation.
Performing dimensionless treatment according to formulas (1) - (3) to obtain delta of six monitoring positions i Delta of 6 samples at 6 monitoring positions i As shown in fig. 9. Samples 1# and 2# are unilateral crack breaks, the strains at TH1 and TH2 are in an ascending trend, the strains at TH3, TH4, TH5 and TH6 are in a descending trend,but the strain changes relatively slowly at TH3 and TH 4. Samples 3#, 4#, 5# and 6# are double sided crack fractures. The lengths of cracks at two sides of a 5# hole of the sample are close, strains at TH 1-TH 16 are in descending trend and approximate in value, the lengths of the cracks at two sides of a 3# hole, a 4# hole and a 6# hole are different, strains at TH 3-6 are in descending trend, and strains at TH1 and TH2 are in ascending and descending trend.
Structural health state monitoring:
structural health is monitored during the test. According to the test result, the test result is a result of analyzing the internal relation among delta, cracks and load cycle numbers, the three are drawn in a graph, and the numerical value suitable for the used test piece is searched from small delta to large delta, so that the multilevel threshold value is presetEnsuring that fatigue cracks can be detected. Counting +.>Is set, the probability P of (2). If the first-level threshold detects P>95%, observing by using a 5-time optical microscope, if fatigue cracks are observed, identifying the structural fatigue damage degree, and if no fatigue cracks are observed, monitoring a next-stage threshold value until the multi-stage threshold value monitoring is finished, wherein the method specifically comprises the following steps:
analysis of delta i Internal connection between cracks and load cycle number, structure health state detection is carried out by adopting a threshold method, and multi-level threshold values are selectedEnsuring that fatigue cracks can be found. If->Then there are:
wherein n is 1 To measureNumber n of exceeding set threshold in quantity data 2 For measuring the total number of data. P is the confidence, if P is greater than 95%, then the ith cycle is considered to have crack generation, and further observation is performed by using an optical microscope.
If the fatigue crack is observed, judging that the structure is subjected to fatigue damage, and evaluating the damage degree of the structure in the next step; and if the structural health state of the next-stage threshold is not observed, monitoring the structural health state of the next-stage threshold.
The method comprises the following steps: the strain characteristic change process of the TH5 and TH6 positions of 6 samples is monotonous and has large change, and the TH5 and TH6 positions are symmetrical, so that the threshold value is set according to the strain characteristic of the TH5 position. Determination of threshold using samples 2#, 4# and 5#The strain peak relative error percentage delta ∈ 9% for samples 4# and 5# without the pre-notch and the strain peak relative error percentage delta ∈ 3% for sample 2# with the pre-notch were plotted together with the variation of the crack length with load cycle number for 3 samples TH5, as shown in fig. 10, for crack lengths within 1 mm-1.5 mm.
Selecting four-level preset threshold values in consideration of sample processing dispersion For sequentially determining whether fatigue cracks are generated in the structure. N herein 2 Taking 1000, when one of the first-order set threshold appears +.>And the probability is more than 95%, namely +.>When an alarm is triggered, then the crack is observed by means of an optical microscope. If the crack can be observed, ending the link; if no crack is observed, the process is continued, but at this time, it is notFocusing on the alarm of the level threshold value, and extending to the next level threshold value; wherein the thresholds 3% and 9% are determined according to fig. 10, and then supplemented with 1% and 6%. Delta obtained for test data under the condition of constant number of the four i And carrying out statistical retrieval, and finally obtaining 95% by comparing the adjustment probability with the test result.
Taking sample 1 as an example, structural health monitoring was performed during the test using a thresholding method, and the results are shown in table 1 as sample 1 verification results. At a first level thresholdN=3000, an alarm is triggered. At this time, observation should be performed using an optical microscope, but no crack observation was performed when the specimen 1# test had ended and n=3000. According to the test results, the crack length was about 1.2mm at n=2000 and about 1.4mm at n=4000, so that the crack length was within 1.2 to 1.4mm at n=3000. Therefore, the selected threshold value can better complete structural health monitoring and give structural health.
TABLE 1
Structural damage degree identification:
training a Gaussian process regression model, and identifying the structural damage degree after the fatigue crack can be observed, namely identifying the length of the fatigue crack. The gaussian process regression algorithm is used, and the gaussian process is a combination of a set of random variables, and any finite number of variables obey a joint gaussian distribution. The Gaussian process regression model calculates super parameters by selecting reasonable kernel functions according to sample information, and the final purpose is to realize regression, and the basic principle is as follows:
the gaussian process regression model properties are determined by mean and covariance functions and can be expressed simply as:
f(x)～GP(m'(x),k(x,x * ))(5)
m'(x)＝E[f(x)] (6)
k(x,x * )＝E[(f(x)-m'(x))(f(x * )-m'(x * ))] (7)
where m' (. Cndot.) is the mean function, k (. Cndot.) is the covariance function, x and x * Representing two different input samples, E is the expectation and GP is a gaussian process.
Given training sample set d= { (x) i ,y i ) I=1, 2,..n = { X, y }, where X i ∈R d Represents d-dimensional input variables, X represents n X d-dimensional input matrix, y i For scalar representation of the corresponding output, y represents an n x 1-dimensional output vector. For the gaussian process regression problem, consider the following model:
wherein n is the number of input sample points, f (·) is a function corresponding to X, ε is the mean value of 0 and variance σ n White noise of (a) is provided.
The posterior distribution of the function is determined by a joint a priori distribution. Given a new prediction input sample x * Then training output y and prediction output f * Is:
wherein k (X, X * )＝k(x * ,X) T K (X, X) is a positive and symmetrical covariance matrix of order n multiplied by n of the training sample, k (X) * ,x * ) To predict the covariance matrix of the input samples themselves in the samples, k (X, X * ) Covariance matrix for training sample input and predicting sample input, I n For an n-dimensional identity matrix,as super parameters, m' (. Cndot.) is a mean function, k (. Cndot.) is a covariance function, and y is training output data;
the training output is the prior, the prior distribution is known, the prediction output is the posterior, and the posterior distribution is obtained through Bayesian updating. The input is dimensionless data for six strain monitoring locations and the output is the corresponding crack length.
Predicted value f * The posterior distribution of (2) is:
in the method, in the process of the invention,and cov (f) * ) Respectively, are the predicted input points x * Corresponding mean and variance, X is training input, y is training output, X * Is a predictive input.
The Materrn covariance kernel function is used herein and is defined as:
wherein v 1 For the degree of freedom, Γ (·) represents the gamma function, K ν Representing the modified Bessel function, the hyper-parametersAnd theta l Scaling factors in the magnitude of the function change amplitude and the horizontal axis direction are respectively, and τ is the distance τ= |x between input variables i -x j |，ν＝5/2，k f As a kernel function, x i x j Is an input variable.
Using maximum likelihood estimation combined with conjugate gradient method to realize super parameter theta= [ sigma ] n ,σ f ,θ l ]The marginal likelihood function in logarithmic form is:
wherein θ= [ σ ] n ,σ f ,θ l ]Covariance matrix for vectors containing all super parametersp (y|X, θ) is a probability symbol, θ is Q -1 Is the inverse of Q, y T Is the transposed matrix of y, n being the super-parameters of the different regions.
Identifying fatigue crack length: taking other test pieces of the same kind to perform strain monitoring test, and using six strain sensors delta i As input, crack length identification was performed using a trained gaussian process regression model.
The 6 monitoring position strain characteristic data of the four samples of the sample No. 2, the sample No. 4, the sample No. 5 and the sample No. 6 are taken as input, the crack lengths at the two sides of the hole are respectively taken as output, and 2 Gaussian process regression models for respectively identifying the crack lengths at the two sides of the hole are trained. Training was performed using Matlab regression toolbox, and the super-parameter estimation results are shown in table 2. As shown in FIG. 11, the training results and the test results show that the recognition results and the test results basically coincide with each other for the sample data, and the recognition effect of the Gaussian process regression model is good.
TABLE 2
As shown in FIG. 1, the Gaussian process regression model identifies crack lengths for samples # 2, # 4, # 5, and # 6.
And (3) carrying out Gaussian process regression model verification:
taking sample 1# as an example, the structural damage degree was evaluated, and as a result, as shown in FIG. 12, the crack length of sample 1# identified by the Gaussian process regression model of FIG. 12
As can be seen from fig. 13, the crack length identified by the gaussian process regression model substantially matches the test data of sample 1# during the crack steady propagation phase.
Comparison of identification algorithm:
identification of crack length using an artificial neural network, support vector regression and an evaluation model trained by a stepwise linear regression algorithm, comparing the identification result with the identification result of a gaussian process regression algorithm, and using the mean absolute error (Mean Absolute Error, MAE), the mean absolute percent error (Mean Absolute Percentage Error, MAPE), the root mean square error (Root Mean Square Error, RMSE), the correlation coefficient (Correlation coefficient, R) and the decision coefficient (Coefficient Of Determination, R) 2 ) Five indices evaluate the accuracy of the model.
The five indices are calculated as follows:
wherein y is the crack length measured by the test expressed in the form of a column vector;the crack length identified for the machine learning model of the column vector representation; y is i As the first testi times the detected crack length; />Is the crack length identified by the ith time of the machine learning model.
The ANN adopts a single hidden layer BP neural network, and the number of hidden layer neurons is 5.SVR adopts a second order polynomial kernel function training model. Crack length results identified by gaussian process regression, artificial neural network, support vector regression, and stepwise linear regression algorithms are shown in fig. 13 (a), and five evaluation index calculation results are shown in table 3. The left crack length of the hole identified by the trial and machine learning models is zero and is therefore not shown in fig. 13 (a). As can be seen from Table 3, the comparison of the different algorithms shows that MAE is 99%, 175% and 219% greater than the Gaussian process regression model, MAPE is 125%, 241% and 324% greater, and RMSE is 45%, 63% and 79% greater, respectively, for crack lengths identified by the artificial neural network, support vector regression and progressive linear regression models, and R and R for the Gaussian process regression, artificial neural network, support vector regression and progressive linear regression 2 The phase difference is within 1%. Therefore, according to the comparison of the five evaluation indexes, the trained Gaussian process regression model can be considered to be more suitable for identifying the crack length. This result can be intuitively obtained as well in fig. 13 (b).
TABLE 3 Table 3
For sample 3 with unknown crack initiation and propagation form and less crack data, crack length was identified using only gaussian process regression model in combination with measured strain peak data to restore the crack propagation process, and the results are shown in fig. 14 (a). From the figure, it can be seen that the trained gaussian process regression model can basically well identify the crack growth state. However, around 30000 cycles, the right crack length recognition result was degraded. As shown in fig. 14 (b), the reason for this phenomenon may be that when the fatigue life of sample 3 is around 30000 cycles, fatigue cracks start to develop on the left side of the sample, so that the strain peaks measured at TH1 and TH2 positions are reduced, and there are only 4 samples available for training, and the number of training samples is small. Although the length of the right crack predicted by the gaussian process regression model decreases during the time from initiation of the left crack to identification, the right crack identification length gradually returns to normal as the left crack propagates, substantially matching the crack length observed during the test.
The above embodiments are merely illustrative of the preferred embodiments of the present invention, and the scope of the present invention is not limited thereto, but various modifications and improvements made by those skilled in the art to which the present invention pertains are made without departing from the spirit of the present invention, and all modifications and improvements fall within the scope of the present invention as defined in the appended claims.
Claims (2)
1. A structural fatigue damage diagnosis method based on strain monitoring, comprising:
determining a strain monitoring position, obtaining strain peak value data of the strain monitoring position, and performing feature extraction on the strain peak value data to obtain dimensionless strain peak value relative error percentage data;
setting a plurality of levels of threshold values based on the strain peak value relative error percentage data of the monitoring position, and judging whether fatigue damage is generated at the strain monitoring position;
establishing a Gaussian process regression model, training the Gaussian process regression model through the strain peak value relative error percentage data and crack information, and identifying crack length based on the trained Gaussian process regression model to finish fatigue damage diagnosis;
the obtaining of strain peak data for the strain monitoring location includes:
performing crack expansion simulation through finite element software, obtaining information of a strain peak value, a crack and a load cycle number, performing strain-crack sensitivity analysis, selecting a position with larger and uniform change for monitoring, obtaining a strain monitoring position in the crack expansion simulation process, performing a strain monitoring fatigue test based on the strain monitoring position, and obtaining strain peak value data and crack information;
performing crack growth simulation by the finite element software includes:
under the action of a pulling-pulling fatigue load, initiating a fatigue crack by a sample, and dividing the crack into a crack on one side of a center hole and a crack on two sides of the center hole based on the evolution of the fatigue crack;
carrying out local encryption and simulation test loading on the crack propagation area grid on the single-side penetration cracks of the central hole and the double-side penetration cracks of the central hole through the finite element software;
performing the simulated test loading includes:
one end of the sample is completely fixed, the other end of the sample only maintains the freedom degree along the length direction, and constant-amplitude fatigue load is applied;
the strain-crack sensitivity analysis based on the strain peak, crack and load cycle number information includes:
based on the increase of the load circulation number, the crack is continuously expanded, the strain field of stress near the crack is changed, an x-y rectangular coordinate system is established on the sample by taking the circle center as the origin of coordinates, and strain peaks at adjacent monitoring positions along the y axis and the x axis are respectively selected for sensitivity analysis;
the obtaining of the strain peak data and the crack information includes:
pasting strain gauges at the strain monitoring positions, acquiring strain peak value data by adopting a dynamic strain gauge, applying a tensile-tensile fatigue load to the sample, loading by adopting sine waves, observing the surface of the sample, and acquiring crack information;
judging whether fatigue damage is generated at the strain monitoring position comprises the following steps:
wherein n is 1 For the number exceeding the set threshold in the measured data, n 2 For the total number of measured data, P is confidenceDegree, delta i Is the non-dimensional data of the data,is a multi-level threshold;
setting a probability value, if the dimensionless data is larger than the confidence coefficient of the threshold value, judging that the strain monitoring position generates fatigue damage if the confidence coefficient of the threshold value exceeds the probability value, and if the confidence coefficient does not exceed the probability value, continuing to the next level threshold value, and judging whether the strain monitoring position generates fatigue damage or not;
the expression of the Gaussian process regression model is as follows:
wherein n is the number of input sample points, f (·) is a function corresponding to X, ε is white noise, σ n Is the variance.
2. The method for diagnosing structural fatigue damage based on strain monitoring as recited in claim 1, wherein the method for obtaining dimensionless data is as follows:
wherein delta i Is the non-dimensional data of the data,for the strain reference value, Δε i Is the difference between the strain measurement and the reference value.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211623049.3A CN116046980B (en) | 2022-12-16 | 2022-12-16 | Structural fatigue damage diagnosis method based on strain monitoring |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211623049.3A CN116046980B (en) | 2022-12-16 | 2022-12-16 | Structural fatigue damage diagnosis method based on strain monitoring |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116046980A CN116046980A (en) | 2023-05-02 |
| CN116046980B true CN116046980B (en) | 2023-09-01 |
Family
ID=86122888
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211623049.3A Active CN116046980B (en) | 2022-12-16 | 2022-12-16 | Structural fatigue damage diagnosis method based on strain monitoring |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116046980B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108263639A (en) * | 2018-01-28 | 2018-07-10 | 北京工业大学 | Aircaft configuration key position fatigue life on-line monitoring method based on indirect measuring strain under spectrum carries |
| CN112520064A (en) * | 2020-12-04 | 2021-03-19 | 中国航空工业集团公司成都飞机设计研究所 | Automatic damage identification method based on strain monitoring |
| CN114818417A (en) * | 2022-04-11 | 2022-07-29 | 上海理工大学 | Finite element simulation method for macroscopic fatigue crack propagation |
| CN114936493A (en) * | 2022-05-19 | 2022-08-23 | 华东理工大学 | Method and system for calculating creep fatigue crack propagation of damage and fracture coupling |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2006006412A1 (en) * | 2004-07-09 | 2006-01-19 | Kyushu Tlo Company, Limited | Fatigue crack growth curve estimation method, estimation program, and estimation device |
| CN109885874B (en) * | 2019-01-11 | 2022-12-23 | 华东理工大学 | ABAQUS-based multi-axial creep fatigue prediction method |
| CN110222439B (en) * | 2019-06-12 | 2020-01-07 | 四川大学 | Fatigue damage and life evaluation method based on Abaqus platform |
-
2022
- 2022-12-16 CN CN202211623049.3A patent/CN116046980B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108263639A (en) * | 2018-01-28 | 2018-07-10 | 北京工业大学 | Aircaft configuration key position fatigue life on-line monitoring method based on indirect measuring strain under spectrum carries |
| CN112520064A (en) * | 2020-12-04 | 2021-03-19 | 中国航空工业集团公司成都飞机设计研究所 | Automatic damage identification method based on strain monitoring |
| CN114818417A (en) * | 2022-04-11 | 2022-07-29 | 上海理工大学 | Finite element simulation method for macroscopic fatigue crack propagation |
| CN114936493A (en) * | 2022-05-19 | 2022-08-23 | 华东理工大学 | Method and system for calculating creep fatigue crack propagation of damage and fracture coupling |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116046980A (en) | 2023-05-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Gorji et al. | Machine learning predicts fretting and fatigue key mechanical properties | |
| CN110555247A (en) | structure damage early warning method based on multipoint sensor data and BilSTM | |
| CN109345032A (en) | Particle filter multiple cracks based on Dynamic Crack number extend prediction technique | |
| CN110555235A (en) | Structure local defect detection method based on vector autoregressive model | |
| CN109779791B (en) | Intelligent diagnosis method for abnormal data in solid rocket engine | |
| Sattarifar et al. | Damage localization and characterization using one-dimensional convolutional neural network and a sparse network of transducers | |
| Vazirizade et al. | Online nonlinear structural damage detection using Hilbert Huang transform and artificial neural networks | |
| Zhong et al. | A process-data-driven BP neural network model for predicting interval-valued fatigue life of metals | |
| CN116046980B (en) | Structural fatigue damage diagnosis method based on strain monitoring | |
| Ferreira et al. | A contribution to the identification of the critical plane using the maximum variance method | |
| CN113723496A (en) | Reactor iron core loosening fault diagnosis method, device and equipment | |
| CN111914474B (en) | Fractional order KVFD multi-parameter machine learning optimization method for soft matter viscoelastic mechanical characterization | |
| Jiao et al. | A machine learning perspective on the inverse indentation problem: uniqueness, surrogate modeling, and learning elasto-plastic properties from pile-up | |
| CN108896625B (en) | Strain field fusion smart network structure crack monitoring method | |
| CN111238927A (en) | Fatigue durability evaluation method and device, electronic equipment and computer readable medium | |
| CN113973403B (en) | Temperature-induced strain field redistribution intelligent sensing method based on structure discrete measurement point topology | |
| CN116050252B (en) | Structure fatigue damage prediction method and device and storage medium | |
| Mahmoudi et al. | A novel method to determine material properties and residual stresses simultaneously using spherical indentation | |
| CN112945775B (en) | Method for identifying tensile property parameters of metal material based on indentation profile | |
| CN115153549A (en) | BP neural network-based man-machine interaction interface cognitive load prediction method | |
| CN114387545A (en) | Intelligent corneal biomechanical characteristic detection method based on feedforward network | |
| CN114925716B (en) | Carbon fiber composite material damage positioning method based on ensemble learning algorithm | |
| CN117233347B (en) | Carbon steel spheroidization grade measuring method, system and equipment | |
| Xue et al. | Deep reference autoencoder convolutional neural network for damage identification in parallel steel wire cables | |
| Shen | MULTI-SCALE MODELING OF DEFORMATION AND MICRO-TEXTURE REGION EFFECT ON FATIGUE CRACK NUCLEATION IN TWO-PHASE TITANIUM ALLOY USING PARAMETRICALLY UPSCALED CONSTITUTIVE MODELS |
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 |