Curtain wall structure adhesive damage detection method and device based on modal difference and digital twinning

Abstract

The application discloses a curtain wall structure adhesive damage detection method and device based on modal difference and digital twinning, and belongs to the technical field of digital twinning structure adhesive damage detection and evaluation. The application solves the problem that the traditional glass curtain wall structure adhesive damage detection and evaluation method based on the first-order natural frequency cannot locate the specific position of damage due to the insensitivity to local boundary delamination. The application collects vibration data of the test group of hidden frame glass curtain walls and the non-damage state group of hidden frame glass curtain walls to obtain normal acceleration time history data. The second-order and third-order natural frequencies are preliminarily screened, and whether the structure adhesive is damaged is judged through the difference constraint condition. The boundary relative curvature modal difference is used to further locate the damage, and whether the current test point is a damage point is judged. A digital twinning model and damage early warning are constructed, and a parameter report and a visual damage evaluation result are output. The application effectively improves the boundary damage positioning accuracy and can be applied to glass curtain wall structure adhesive damage detection.

Description

TECHNICAL FIELD

[0001] The present application relates to a curtain wall structure adhesive damage detection method and device, in particular to a curtain wall structure adhesive damage detection method and device based on modal difference and digital twinning, belonging to the technical field of structure adhesive damage detection and evaluation based on digital twinning. BACKGROUND

[0002] The hidden frame glass curtain wall is widely used in modern architecture due to its transparency and beauty, and its structural safety highly depends on the bonding performance of the structural adhesive between the glass panel and the metal frame. With the increase of service life, the structural adhesive is prone to damage (such as delamination and cracking) due to aging, deterioration or construction defects, which may lead to the risk of curtain wall panel falling off, and therefore efficient and accurate non-destructive testing technology is urgently needed. The existing hidden frame curtain wall structural adhesive damage detection technology mainly includes: physical contact-based methods (such as suction cup method and air bag pressurization method), which indirectly judge the adhesive performance by applying external force to evaluate the displacement of the panel, but the efficiency is low, the disturbance is large and the damage cannot be accurately located; fixed sensor-based methods (such as accelerometers and fiber Bragg grating sensors), which require pre-embedding and installation of sensors on the surface of the curtain wall or in the adhesive, and judge the existence of damage by monitoring the decrease of natural frequency or strain anomaly, which can identify some damage, but has problems such as complex installation, limited coverage, difficult high-altitude operation, inability to accurately locate small damage boundaries (especially boundary delamination), and the change of natural frequency is not sensitive to early small damage, making it difficult to locate; vibration modal analysis-based methods (such as curvature modal) are sensitive to local stiffness changes, which have the potential to locate damage in theory, but existing technologies usually rely on densely arranged contact sensors to obtain full-field vibration modes, which are complex to calculate and difficult to apply to high-altitude and large-area curtain wall site detection; in addition, the existing methods generally lack effective means to associate the detection results with dynamic digital twinning technology and conduct damage evaluation.

[0003] In prior art one, patent document with publication (announcement) number CN109959709A discloses a kind of full hidden frame glass curtain wall boundary structure sealant damage identification method, the damage condition of structural sealant is judged by comparing the difference of frequency response function of undamaged glass curtain wall and the frequency response function of to-be-detected curtain wall, specific steps include:1) make undamaged reference curtain wall;2) install acceleration sensor at specific position of glass panel;3) collect force signal and acceleration signal by exciting vibration by force hammer;4) calculate the frequency response function of undamaged and to-be-detected curtain wall respectively;5) quantify damage degree by relative cumulative error formula (RAE), when RAE>0.1, it is judged that there is damage and alarm is triggered, it realizes rapid detection using vibration response characteristic difference, with the advantages of simple equipment, convenient operation, high efficiency, etc., can effectively identify the aging or cracking problem of structural sealant, prevent glass curtain wall from falling off, suitable for large-scale detection in engineering site;In prior art two, patent document with publication (announcement) number CN113866276A discloses a kind of glass curtain wall loosening detection device and method based on piezoelectric transducer, mainly solve the problems of signal coupling, high-altitude operation risk and insufficient early fault identification precision in traditional detection, the core of device includes: excitation signal module, at least two pieces of piezoelectric transducer (as driver and sensor respectively) fixed to curtain wall glass and data acquisition system, impedance characteristic curve is obtained by decoupling processing output signal, and loosening state is judged by analyzing inherent frequency change.The innovation points are:1) double piezoelectric sheet discrete design is adopted, mechanical signal and capacitance signal are decoupled through lock-in amplifier circuit, and the applicability of flexible materials such as PVDF film is improved;2) signal conditioning module containing phase shifter and balanced modulator is proposed, and response amplitude is accurately extracted through Vo=√(Vo1²+Vo2²) algorithm;3) electromechanical coupling dynamics model is established, inherent frequency is located through impedance curve peak value, and sensitivity is improved by more than 10 times compared with traditional knocking method;But the existing technology has the following shortcomings:1) it depends on manual pre-embedding, pasting acceleration sensor and force hammer excitation, high-altitude installation operation is complex, time-consuming, has safety risk and low efficiency, and it is difficult to realize rapid detection of large-area curtain wall;2) only the existence of damage is qualitatively judged by first-order inherent frequency offset, and it is not sensitive to local boundary delamination, and lacks spatial positioning capability, so it cannot provide damage coordinate and local stiffness degradation degree, so that detection data cannot be dynamically mapped to digital twin model, and it cannot support predictive maintenance decision.

[0004] In summary, for the safety of building, a non-contact, high-efficiency, precise boundary damage detection method and device based on modal difference and digital twin are needed. SUMMARY

[0005] The following presents a simplified summary of the application in order to provide a basic understanding of some aspects of the application. This summary is not an extensive overview of the application. It is intended to neither identify key or critical elements of the application nor delineate the scope of the application. Its purpose is merely to present some concepts of the application in a simplified form as a prelude to the more detailed description that is presented later.

[0006] In view of this, in order to solve the problem that the conventional glass curtain wall structure adhesive damage detection and evaluation method based on the first-order natural frequency in the prior art cannot locate the specific position of the damage due to the insensitivity to local boundary adhesive peeling, the application provides a curtain wall structure adhesive damage detection method and device based on modal difference and digital twinning.

[0007] The technical solution one is as follows: a curtain wall structure adhesive damage detection method based on modal difference and digital twinning, comprising the following steps:

[0008] S1. Collecting vibration data of a to-be-tested group of hidden-frame glass curtain walls and a no-damage group of hidden-frame glass curtain walls to obtain normal acceleration time history data;

[0009] S2. Preliminary screening of second-order and third-order natural frequencies according to the normal acceleration time history data, and judging whether the structure adhesive is damaged or not through a difference constraint condition;

[0010] S3. Further locating the damage by using boundary relative curvature modal difference according to the judgment result of step S2, determining whether the current test point is a damage point, and completing the damage position locating;

[0011] S4. Constructing a digital twinning model and damage early warning according to the damage position locating result of step S3, outputting a parameter report and a visual damage evaluation result, and realizing damage evaluation.

[0012] Further, in S1, for the collection of vibration data of the to-be-tested group of hidden-frame glass curtain walls, a laser vibration meter and a projectile exciter are controlled to hover within a preset range from the glass panel, the projectile exciter emits nylon projectiles to hit the center of the glass panel, and at the same time, the laser vibration meter synchronously collects the glass vibration response signal according to the arranged collection test point position to obtain the normal acceleration time history data of the glass panel to be tested.

[0013] For the collection of vibration data of the no-damage group of hidden-frame glass curtain walls, based on the geometric size, material properties and boundary constraint condition of the to-be-tested hidden-frame glass curtain wall, a glass curtain wall structure reduced 1:1 is built in the laboratory, and the vibration data collection step of the to-be-tested group of hidden-frame glass curtain walls is repeated to obtain the normal acceleration time history data of the glass panel under the no-damage condition.

[0014] Further, in the S2, according to the solving process of the second order natural frequency and the third order natural frequency, the second order natural frequency and the third order natural frequency of the to-be-tested group of hidden frame glass curtain walls are obtained and compared with the second order natural frequency and the third order natural frequency of the intact group of hidden frame glass curtain walls, and through the comparison, the second order natural frequency difference and the third order natural frequency difference are obtained.

[0015] The solving process of the second order natural frequency and the third order natural frequency is represented as: Fourier transform is performed on the normal acceleration time history collected in the step S1 to obtain an acceleration amplitude spectrum , wherein, the Fourier transform process is represented by , the x-axis frequency value of the second significant peak of the acceleration amplitude spectrum corresponds to the second order natural frequency, and the x-axis frequency value of the third significant peak corresponds to the third order natural frequency.

[0016] When the second order natural frequency and the third order natural frequency satisfy any one of the difference constraint conditions, it is determined that the structural glue is damaged.

[0017] The difference constraint condition is represented as:

[0018]

[0019]

[0020] wherein, is the second order natural frequency of the structural glue in the intact state, is the third order natural frequency of the structural glue in the intact state.

[0021] Further, in the S3, based on the acceleration signals of the to-be-tested group and the intact group, i.e., the normal acceleration time history data, the acceleration amplitude spectrum of the panel boundary measuring point is calculated , the third order natural frequency more sensitive to the boundary constraint is extracted, and the corresponding acceleration peak value is obtained.

[0022] The calculation formula of the third order relative modal displacement is represented as

[0023]

[0024] wherein, ​​The maximum acceleration amplitude of all the measuring points of the to-be-tested group at the third order natural frequency is obtained.

[0025] According to the third order relative modal displacement of the undamaged group and the to-be-tested group, the third order relative curvature modal difference is calculated, and the specific process is as follows:

[0026] Third order relative modal displacement difference is expressed as:

[0027]

[0028] wherein, is the displacement amplitude of the third order modal at the grid point (i, j), is the third order relative modal displacement of the undamaged group, is the third order relative modal displacement of the to-be-tested group;

[0029] According to the third order relative modal displacement difference , combined with the central difference formula, the third order relative curvature modal difference is obtained.

[0030] Third order relative curvature modal difference in x direction is expressed as:

[0031]

[0032] Third order relative curvature modal difference in y direction is expressed as:

[0033]

[0034] Third order relative modal difference of torsional curvature is expressed as:

[0035]

[0036] When the third order relative curvature modal difference peak value of the measuring point exceeds the preset value, the current measuring point is determined as a damage point, and the damage position positioning is completed.

[0037] Further, the S4 comprises the following steps:

[0038] S41. Establishing an initial structure model, combining with damage area coordinates, completing damage position mapping:

[0039] In the S41, the initial structure model, i.e. the high-precision finite element model, is established based on the geometric size, material properties and boundary constraint conditions of the hidden frame glass curtain wall, wherein the glass adopts a shell element, the structural adhesive adopts a spring element to simulate the adhesive bonding behavior, a spring element is established at each boundary, and the initial stiffness is calibrated according to the second and third order natural frequencies of the undamaged group At this time, the second and third order natural frequencies of the initial structure model are the same as the second and third order natural frequencies of the undamaged group 、 consistent;

[0040] The damage area coordinates (i, j) located by step S3 are mapped to the corresponding nodes of the finite element model, and the spring elements of the damage area are marked;

[0041] S42. A target function for model correction is established, an inversion iterative algorithm is used to solve the target function to iteratively correct the initial model of the structure, a corrected model is obtained, and the digital twin model construction is realized;

[0042] In the S42, based on the damage area coordinates (i, j) located by step S3 and the measured natural frequency offset, i.e., the second-order natural frequency , the third-order natural frequency of the to-be-measured group, a target function for model correction is established and solved, i.e., the damage coordinates (i, j) are mapped to the corresponding nodes of the finite element model, the spring elements of the current damage area are marked as the correction objects, the initial value of the stiffness reduction coefficient α is set as the engineering experience value 0.8, then modal analysis is performed, the second-order natural frequency and the third-order natural frequency of the current model are calculated, and the relative error , between the current frequencies and the measured damage frequencies is calculated. If δ ≤ 1%, it is considered that the target function has met the correction condition, the current α value is output and the iteration is terminated, otherwise, the stiffness reduction coefficient α is updated according to α = α - 0.05, and the modal analysis is performed again for the next iteration calculation until the target function has met the correction condition;

[0043] Finally, the digital twin model consistent with the measured damage state is obtained, and the output data includes the damage position coordinates (X, Y) converted from the damage area coordinates (i, j), the stiffness reduction coefficient α, and the matching results of the second-order and third-order natural frequencies of the corrected model and the measured values;

[0044] S43. According to the designed wind load working condition, the glass panel deflection is calculated, a safety warning grading mechanism is set, and the damage evaluation result is output by the digital twin model;

[0045] In the S43, according to the building structure load specification, the standard value of the wind load at the position of the curtain wall is calculated , wherein, is the gust factor, is the wind load shape factor, is the wind pressure height variation coefficient, is the basic wind pressure;

[0046] In the digital twin model, design wind loads with positive and negative pressures are uniformly applied to the glass panel, respectively. Design value of wind load Standard value of wind load 1.4 times that of the wind load, which is applied as a static uniformly distributed pressure;

[0047] The maximum normal displacement of the glass in the digital twin model under the design wind load is calculated using geometric nonlinearity as the maximum deflection df, with the unit being mm. The maximum deflection df is then normalized to obtain the normalized deflection.

[0048] The normalized deflection is expressed as:

[0049]

[0050] Among them, the benchmark value is the maximum deflection value of the glass in the undamaged group, and the damage limit value is the maximum deflection value of the glass when the structural adhesive is severely damaged.

[0051] Based on the normalized deflection value, the damage is divided into four levels: Level 1 damage: normalized deflection < 0.1; Level 2 damage: 0.1 ≤ normalized deflection < 0.5; Level 3 damage: 0.5 ≤ normalized deflection < 1; Level 4 damage: normalized deflection > 1.

[0052] Damage assessment is achieved by outputting the following parameter reports and visualized damage assessment results through a digital twin model: Damage coordinates: (X, Y) = (__, __); Corrected stiffness reduction factor: α; Maximum deflection of glass panel at mid-span under standard wind load: df = __ mm and normalized deflection; Warning level: [mild|moderate|severe|damaged], __ indicates the current value;

[0053] The visual damage assessment results include the damage-marked areas in the digital twin model and the glass deflection visualization cloud map.

[0054] Technical Solution 2: A curtain wall structural adhesive damage detection device based on modal difference and digital twin, used to execute the curtain wall structural adhesive damage detection method based on modal difference and digital twin described in Technical Solution 1, including a UAV carrier unit, a laser vibrometer, and a projectile vibrator;

[0055] The UAV carrier unit is connected to the laser vibration meter and the projectile vibrator, respectively.

[0056] The laser vibration meter is used to collect vibration signals;

[0057] The projectile vibrator is used to excite the glass to vibrate.

[0058] The beneficial effects of the present application are as follows: the present application breaks through the limitation of the detection result of the prior art away from the digital model, maps the detection result to the digital twin model dynamically by positioning the damage coordinates (i, j) and correcting the stiffness reduction coefficient alpha, realizes damage evolution prediction and grading warning, improves the detection efficiency through non-contact detection of the unmanned aerial vehicle vehicle-mounted laser vibration instrument (LDV) combined with projectile excitation, completely eliminates the risk of high-altitude manual operation, overcomes the low efficiency and safety risk problem caused by the dependence of the prior art on manual installation of sensors or hammer excitation, and realizes full-process non-contact detection.

[0059] The two-stage diagnosis mechanism (two-stage inherent frequency preliminary screening + boundary relative curvature modal positioning) innovatively proposed by the present application does not need to obtain excitation force information, accurately positions the centimeter-level damage interval, visualizes the damage area, dynamically maps the positioned damage coordinates (i, j) to the digital twin model, and quantifies the damage degree through the stiffness reduction coefficient alpha; the present application simulates the design wind load, calculates the maximum deflection of the glass panel under the action of the wind load, establishes a four-level warning mechanism (mild | moderate | severe | damage) in combination with deflection normalization, finally outputs a key parameter report and a visualization result, realizes quantitative analysis and warning of the damage state, provides a quantitative basis for maintenance decision, improves the boundary damage positioning accuracy, and the proposed boundary relative curvature modal difference positioning method can realize accurate spatial positioning of the centimeter-level damage interval. BRIEF DESCRIPTION OF DRAWINGS

[0060] The drawings described herein are used to provide further understanding of the present application, and form a part of the present application. The schematic embodiments of the present application and the description thereof are used to explain the present application, and do not constitute an improper limitation on the present application. In the drawings:

[0061] Figure 1 It is a flowchart of the curtain wall structure adhesive damage detection method based on modal difference and digital twin;

[0062] Figure 2 It is a schematic diagram of the position of the measurement point of the laser vibration instrument;

[0063] Figure 3 It is a schematic diagram of the position of the measurement point of the boundary relative curvature modal method;

[0064] Figure 4 It is a flowchart of the mapping and warning of the digital twin model;

[0065] Figure 5 It is an embodiment flowchart of the curtain wall structure adhesive damage detection method based on modal difference and digital twin;

[0066] Figure 6 It is a structural schematic diagram of the curtain wall structure adhesive damage detection device based on modal difference and digital twin.

[0067] Figure descriptions: 1. Unmanned aerial vehicle (UAV) carrier unit; 2. Laser vibration meter; 3. Projectile vibrator. Detailed Implementation

[0068] To make the technical solutions and advantages of the embodiments of the present invention clearer, the exemplary embodiments of the present invention will be further described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not an exhaustive list of all embodiments. It should be noted that, unless otherwise specified, the embodiments and features in the embodiments of the present invention can be combined with each other.

[0069] Example 1: Reference Figures 1-6 This embodiment describes a method for detecting structural adhesive damage in curtain walls based on modal difference and digital twins, specifically including the following steps:

[0070] S1. Collect vibration data of the hidden frame glass curtain wall under test and the undamaged hidden frame glass curtain wall to obtain normal acceleration time history data;

[0071] S2. Based on the normal acceleration time history data, the second and third natural frequencies are initially screened, and the structural adhesive is judged to be damaged by the difference constraint condition.

[0072] S3. Based on the judgment result of step S2, further locate the damage using the boundary relative curvature mode difference, determine whether the current measuring point is a damage point, and complete the damage location;

[0073] S4. Based on the damage location results in step S3, construct a digital twin model and damage early warning system, output parameter reports and visualized damage assessment results, and realize damage assessment.

[0074] Specifically, the two-stage damage diagnosis method is as follows: 1) Initial screening stage: the presence of damage is determined by the shift of the second and third natural frequencies, which is more sensitive to changes in boundary stiffness; 2) Localization stage: the modal displacement of the boundary measuring points is extracted, the relative curvature modal difference is calculated by the central difference method, and the peak value of the relative curvature modal difference is used as the damage criterion to accurately locate the centimeter-level debonding interval.

[0075] Damage-digital twin dynamic mapping technology: The damage coordinates (i, j) are mapped to the digital twin model, and the stiffness reduction factor α is calculated by combining the natural frequency offset to quantify the degree of damage. Wind load is applied to the digital twin model to obtain the maximum deflection of the glass and normalize it. A four-level early warning mechanism (mild|moderate|severe|damage) is established to realize the automatic comparison between the detection results and the early warning mechanism.

[0076] Further, in the S1, for the collection of vibration data of the hidden frame glass curtain wall to be detected, i.e., the test group, the laser vibration tester and the projectile exciter are controlled to hover in a preset range (less than 5 m) from the glass panel, the projectile exciter emits nylon projectiles to hit the center of the glass panel, and the laser vibration tester synchronously collects the vibration response signals of the glass according to the arranged collection point positions. The vibration response of each glass panel is collected three times with an interval of 5 seconds, and the normal acceleration time history data of the glass panel to be detected is obtained by taking the average value of the vibration responses.

[0077] For the collection of vibration data of the hidden frame glass curtain wall in a non-damaged state, i.e., the non-damaged group, based on the geometric size, material properties (glass elastic modulus E, Poisson's ratio v, and structural adhesive shear modulus G) and boundary constraint conditions of the hidden frame glass curtain wall to be detected, a glass curtain wall structure with a 1:1 reduction thereof is built in the laboratory, and the structural adhesive bonding is complete and firm. The vibration data collection steps of the hidden frame glass curtain wall to be detected are repeated to obtain the normal acceleration time history data of the glass panel in a non-damaged state.

[0078] Reference Figure 2 In the embodiment, the glass curtain wall is a square with a size of 600 mm x 600 mm, 7 rows and 7 columns can be set, and 49 laser vibration testers can be arranged to collect the measurement points. The density of the measurement points can be adjusted based on the desired accuracy. The higher the density, the higher the accuracy.

[0079] Further, in the S2, according to the solving process of the second-order natural frequency and the third-order natural frequency, the second-order natural frequency and the third-order natural frequency of the test group hidden frame glass curtain wall are obtained and compared with those of the non-damaged group hidden frame glass curtain wall. The difference between the second-order natural frequencies is and the difference between the third-order natural frequencies is .

[0080] The solving process of the second-order natural frequency and the third-order natural frequency is represented as: performing Fourier transform on the normal acceleration time history collected in step S1 to obtain the acceleration amplitude spectrum , , wherein, represents the Fourier transform process, the x-axis frequency value (unit: HZ) of the second significant peak of the acceleration amplitude spectrum corresponds to the second-order natural frequency, and the x-axis frequency value (unit: HZ) of the third significant peak corresponds to the third-order natural frequency.

[0081] According to the principle that structural adhesive damage will lead to a decrease in boundary constraint and ultimately cause a decrease in natural frequency, the existence of structural adhesive damage can be judged according to the second-order natural frequency and the third-order natural frequency.

[0082] When the second-order natural frequency and the third-order natural frequency If any of the difference constraints is satisfied, it is determined that the structural glue has damage;

[0083] The difference constraint is expressed as:

[0084]

[0085]

[0086] wherein, is the second-order natural frequency of the structural glue in the undamaged state, is the third-order natural frequency of the structural glue in the undamaged state;

[0087] Further, in the S3, based on the acceleration signals, i.e., normal acceleration time history data, of the to-be-tested group and the undamaged group, the acceleration amplitude spectrum of the panel boundary measuring point is calculated , the third-order natural frequency corresponding to the acceleration peak value (unit: m / s2) is extracted, and the acceleration peak values of the to-be-tested group and the undamaged group are normalized to calculate the third-order relative modal displacement, respectively.

[0088] The calculation formula of the third-order relative modal displacement is expressed as

[0089]

[0090] wherein, is the maximum acceleration amplitude of all measuring points of the to-be-tested group at the third-order natural frequency;

[0091] According to the third-order relative modal displacements of the undamaged group and the to-be-tested group, the third-order relative curvature modal difference is calculated, and the specific process is as follows:

[0092] The third-order relative modal displacement difference is expressed as:

[0093]

[0094] wherein, represents the displacement amplitude of the third-order modal at the grid point (i, j), is the third-order relative modal displacement of the undamaged group, is the third-order relative modal displacement of the to-be-tested group;

[0095] According to the third-order relative modal displacement difference , the third-order relative curvature modal difference is obtained in combination with the central difference formula.

[0096] The third-order relative curvature modal difference in the x direction is expressed as:

[0097]

[0098] third order relative curvature modal difference in y direction is expressed as:

[0099]

[0100] third order relative modal difference of torsional curvature is expressed as (may not be calculated in detection, and the structural adhesive damage mainly affects the normal bending of the panel):

[0101]

[0102] When the third order relative curvature modal difference peak value of the measuring point exceeds the preset value, that is, it is obviously too large, it is determined that the current measuring point is a damage point, that is, the damage position positioning is completed.

[0103] Specifically, referring to the boundary measuring point of Figure 3 , the transverse measuring point calculates the relative curvature modal difference in y direction, and the longitudinal measuring point calculates the relative curvature modal difference in x direction;

[0104] The process of traditional curvature modal difference positioning damage is expressed as:

[0105] First, the acceleration frequency response function is calculated, so as to obtain the displacement frequency response function , ( wherein, is the angular frequency, is the frequency), the imaginary part of the displacement frequency response function is taken at the natural frequency, and discrete displacement data is obtained, wherein i and j are the indices of the grid points in x and y directions, denotes the displacement amplitude of the (m, n) order mode at the grid point (i, j).

[0106] The central difference method is used to numerically differentiate the discrete displacement data , so as to approximately calculate the curvature components in x and y directions and torsion;

[0107] The curvature component in x direction is expressed as:

[0108]

[0109] The curvature component in y direction is expressed as:

[0110]

[0111] twisted curvature component is expressed as:

[0112]

[0113] wherein, , are the grid spacing in x, y direction respectively, represents the displacement amplitude of the (m, n)th mode at the grid point (i, j+1), represents the displacement amplitude of the (m, n)th mode at the grid point (i, j-1), represents the displacement amplitude of the (m, n)th mode at the grid point (i+1, j), represents the displacement amplitude of the (m, n)th mode at the grid point (i-1, j);

[0114] The modal displacement difference is calculated as ;

[0115] The modal displacement difference is calculated as is expressed as:

[0116]

[0117] wherein, is the modal displacement under the undamaged state, is the modal displacement under the damaged state;

[0118] The modal displacement difference is substituted into the central difference formula to obtain the x and y direction curvature modal difference;

[0119] The x direction curvature modal difference is expressed as:

[0120]

[0121] The y direction curvature modal difference is expressed as:

[0122]

[0123] Finally, the conventional method locates the structural adhesive damage by using the peak point of the curvature modal difference;

[0124] But the present application uses the relative curvature modal difference to locate the damage, which is different from the conventional method in that: when the conventional method calculates the frequency response function , the hammer excitation force The time history signal is limited to the laboratory, and the present application uses a UAV to launch a projectile to excite, does not need to obtain the time history signal of the projectile excitation force, and innovatively uses relative curvature modal difference positioning, which is better than the traditional method in positioning effect; The theoretical basis of the present application is the consistency of vibration mode shape, whether it is traditional modal displacement difference (dependent on excitation force) or relative modal displacement difference (only acceleration signal is needed), the core of its damage positioning function is: retain the spatial distribution characteristics of the vibration mode, and amplify the local distortion through curvature calculation; Relative modal displacement is essentially a linear scaling of traditional modal displacement, and curvature is the second derivative of displacement, linear scaling does not affect the difference result, therefore, the peak position (i.e. damage position) of the relative curvature modal difference is relatively consistent with the traditional method, which is more accurate, and under the condition of not needing to measure the excitation force, it perfectly reproduces the damage positioning function of the traditional method.

[0125] Further, in the S4, the following steps are included:

[0126] S41. Establish an initial model of the structure, combine the damage area coordinates, and complete damage position mapping:

[0127] In the S41, based on the geometric size (glass panel size, thickness) of the hidden frame glass curtain wall, the material properties (density, elastic modulus, Poisson's ratio) and boundary constraint conditions, an initial model of the structure, i.e. a high-precision finite element model, is established, wherein the glass adopts a shell element, the structural adhesive adopts a spring element to simulate the adhesive bonding behavior, a spring element is established at each boundary, and the initial stiffness is calibrated according to the second and third order natural frequencies of the undamaged group ; At this time, the second and third order natural frequencies of the initial model of the structure are consistent with the second and third order natural frequencies of the undamaged group

[0128] The damage area coordinates (i, j) located by the step S3 are mapped to the corresponding nodes of the finite element model, and the spring elements of the damage area are marked;

[0129] S42. Establish a target function for model correction, solve the target function by using an inversion iterative algorithm to iteratively correct the initial model of the structure, obtain a corrected model, and realize construction of a digital twin model;

[0130] In the S42, based on the damage area coordinates (i, j) located by the step S3 and the measured natural frequency offset, i.e. the second order natural frequency and the third order natural frequency ​, the damage coordinates (i, j) are mapped to the corresponding nodes of the finite element model, the spring elements in the current damage area are marked as the correction objects, the initial value of the stiffness reduction coefficient a is set as the engineering experience value 0.8, then the modal analysis is performed, and the second-order natural frequency and the third-order natural frequency of the current model are calculated , If the relative error δ between the current frequency and the measured damage frequency is less than or equal to 1%, it is considered that the objective function has met the correction condition, the current a value is output, and the iteration is terminated, otherwise, the stiffness reduction coefficient a is updated according to a=a-0.05, and the modal analysis is performed again for the next iteration until the objective function has met the correction condition.

[0131] Finally, the digital twin model consistent with the measured damage state is obtained, and the output data includes the accurate positioning damage position coordinates (X, Y) converted by the damage area coordinates (i, j), the stiffness reduction coefficient a, and the matching results of the second-order and third-order natural frequencies of the corrected model and the measured values.

[0132] S43. According to the designed wind load working condition, the glass panel deflection is calculated, a safety warning grading mechanism is set, and the damage assessment result is output by the digital twin model;

[0133] In the S43, the wind load standard value at the position of the curtain wall is calculated according to the building structure load specification (GB 50009) , , wherein, is the gust factor, is the wind load shape factor, is the wind pressure height variation factor, is the basic wind pressure.

[0134] In the digital twin model, the designed wind load is uniformly applied to the glass panel, and the designed wind load is uniformly applied to the glass panel, the designed wind load is 1.4 times the wind load standard value, and the wind load application mode is static uniform pressure.

[0135] The maximum normal displacement of the glass of the digital twin model under the designed wind load is calculated as the maximum deflection df by geometric nonlinearity (large deformation theory considering the influence of structural deformation on its own stiffness), and the maximum deflection df is normalized to obtain the normalized deflection.

[0136] The normalized deflection is represented as:

[0137]

[0138] Wherein, the reference value is the maximum deflection value of the undamaged group of glass, and the damage limit value is the maximum deflection value of the glass when the structural adhesive is judged to be severely damaged, and according to the simulation report result, the two are respectively 1.721 mm and 2.317 mm;

[0139] According to the value of the normalized deflection, the damage is divided into four levels, namely, first-level damage (mild): normalized deflection < 0.1, second-level damage (moderate): 0.1 ≤ normalized deflection < 0.5, third-level damage (severe): 0.5 ≤ normalized deflection < 1, and fourth-level damage (destruction): normalized deflection > 1;

[0140] The following parameter report and visual damage assessment result are output by the digital twin model to realize damage assessment: damage coordinates: (X, Y) = (__, __); corrected stiffness reduction coefficient: alpha; maximum deflection of the glass panel at the midspan under the standard value of wind load: df = __ mm and normalized deflection; warning level: [mild | moderate | severe | destruction], __ represents the current value.

[0141] The visual damage assessment result includes the damage marking area in the digital twin model and the visual cloud map of the glass deflection.

[0142] Specifically, the correction aims to find the stiffness reduction coefficient alpha of the spring unit in the damage area, 0 < alpha ≤ 1, so that the second-order natural frequency and the third-order natural frequency of the corrected digital twin model respectively satisfy the following conditions: The present application can also use acoustic excitation, electromagnetic pulse instead of projectile excitation.

[0143] Embodiment 2: Reference Figure 6 Detailed description of this embodiment, based on the modal difference and the digital twin of the curtain wall structural adhesive damage detection device, for executing the modal difference and the digital twin of the curtain wall structural adhesive damage detection method described in embodiment 1, including unmanned aerial vehicle unit 1, laser vibration meter 2, projectile exciter 3;

[0144] The unmanned aerial vehicle unit 1 is connected with the laser vibration meter 2 and the projectile exciter 3 through the rigid support respectively;

[0145] The laser vibration meter 2 is used for collecting vibration signals;

[0146] The projectile exciter 3 is used for exciting glass vibration.

[0147] Specifically, in the present embodiment, the unmanned aerial vehicle carrier unit 1 is composed of an unmanned aerial vehicle bottom integrated RTK positioning module and a ducted fan negative pressure adsorption device (negative pressure value ≥ 0.5 atm), which ensures stable hovering in strong wind environment;

[0148] The laser vibration tester 2 (LDV) is equipped with a Polytec RSV-150 remote laser Doppler vibration tester, with a wavelength of 532 nm, a spot diameter of 1 mm, and a scanning frequency of 0-1.5 kHz.

[0149] The projectile exciter 3 emits nylon projectiles to the center of the glass to induce glass vibration, and each projectile has consistent material, mass, size and emission speed.

[0150] Although the present application has been described in terms of limited embodiments, those skilled in the art, in light of the above teachings, will appreciate that other embodiments are possible within the scope of the present application as described herein. Furthermore, it should be noted that the language used in the specification has been principally selected for readability and instructional purposes and can not have been selected to expressly convey the scope of the present application. Accordingly, many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the scope of the appended claims. The disclosures of the present application are illustrative only and not restrictive of the scope of the present application, which is defined by the appended claims.

Claims

1. A method for detecting damage to structural adhesives in curtain walls based on modal difference and digital twin, characterized in that, Includes the following steps: S1. Collect vibration data of the hidden frame glass curtain wall under test and the undamaged hidden frame glass curtain wall to obtain normal acceleration time history data; S2. Based on the normal acceleration time history data, the second and third natural frequencies are initially screened, and the structural adhesive is judged to be damaged by the difference constraint condition. S3. Based on the judgment result of step S2, further locate the damage using the boundary relative curvature mode difference, determine whether the current measuring point is a damage point, and complete the damage location; S4. Based on the damage location results in step S3, construct a digital twin model and damage early warning system, output parameter reports and visualized damage assessment results, and realize damage assessment. In step S2, based on the solution process for the second and third natural frequencies, the second and third natural frequencies of the frameless glass curtain wall under test are obtained and compared with those of the frameless glass curtain wall under test. The difference in the second natural frequency is obtained through comparison. and the difference between the third-order natural frequency ; The solution process for the second and third natural frequencies is expressed as follows: The normal acceleration time history acquired in step S1... Perform a Fourier transform to obtain the acceleration amplitude spectrum. , ,in, Represents the Fourier transform process, acceleration amplitude spectrum The x-axis frequency value of the second significant peak corresponds to the second natural frequency, and the x-axis frequency value of the third significant peak corresponds to the third natural frequency. When the second natural frequency and third-order natural frequency If any one of the difference constraints is met, the structural adhesive is determined to be damaged. The difference constraint is expressed as: ； ； in, The second natural frequency of the structural adhesive in a non-destructive state. The third natural frequency of the structural adhesive in a non-destructive state; In step S3, the acceleration amplitude spectrum of the panel boundary measuring points is calculated based on the acceleration signals (i.e., normal acceleration time history data) of the test group and the non-destructive group. Extracting the third-order natural frequencies that are more sensitive to boundary constraints. Corresponding peak acceleration The peak acceleration values ​​of both the test group and the non-destructive group were normalized and calculated to obtain the third-order relative modal displacements. Third-order relative modal displacement The calculation formula is expressed as follows: ； in, The maximum acceleration amplitude at the third natural frequency for all measurement points in the test group; The third-order relative curvature mode difference is calculated based on the third-order relative modal displacements of the non-destructive group and the test group. The specific process is as follows: Third-order relative modal displacement difference Represented as: ； in, This represents the displacement amplitude of the third mode at grid point (i, j). For the third-order relative modal displacement of the lossless group, The third relative modal displacements of the test group; Based on the third-order relative modal displacement difference By combining the central difference formula, the third-order relative curvature mode difference is obtained; Third relative curvature mode difference in the x-direction Represented as: ； Third relative curvature mode difference in the y-direction Represented as: ； Third relative modal difference of torsional curvature Represented as: ； When the peak value of the third relative curvature mode difference of the measuring point exceeds the preset value, the current measuring point is determined to be a damage point, that is, the damage location is completed. S4 includes the following steps: S41. Establish an initial structural model and, based on the coordinates of the damaged area, complete the damage location mapping: In step S41, a high-precision finite element model is established based on the geometric dimensions, material properties, and boundary constraints of the frameless glass curtain wall. The glass is represented by shell elements, and the structural adhesive is represented by spring elements to simulate the adhesive bonding behavior. Spring elements are established at each boundary, and the initial stiffness is determined based on the undamaged state. Calibration, at this point, the second and third natural frequencies of the initial structural model and the second and third natural frequencies of the lossless group are... , Consistent; The coordinates (i, j) of the damaged area located in step S3 are mapped to the corresponding nodes of the finite element model, and the spring elements of the damaged area are marked. S42. Establish the objective function for model correction, use the inversion iterative algorithm to solve the objective function to iteratively correct the initial structural model, obtain the corrected model, and realize the construction of the digital twin model; In step S42, the second-order natural frequency of the test group is determined based on the coordinates (i, j) of the damaged area located in step S3 and the natural frequency offset measured in step S2. Third-order natural frequency The objective function for model correction is established and solved, which maps the damage coordinates (i, j) to the corresponding nodes of the finite element model. The spring elements in the current damaged region are marked as correction objects. The initial value of the stiffness reduction factor α is set to an empirical value of 0.

8. Then, modal analysis is performed to calculate the second natural frequency of the current model. and third-order natural frequency The relative error between the current frequency and the measured damage frequency is calculated. , If δ ≤ 1%, the objective function is considered to have met the correction condition, the current α value is output and the iteration is terminated; otherwise, the stiffness reduction coefficient α is updated according to α = α - 0.05, and the modal analysis is performed for the next iteration calculation until the objective function has met the correction condition. Finally, a digital twin model consistent with the measured damage state is obtained. Its output data includes the damage location coordinates (X, Y) transformed from the damage region coordinates (i, j), the stiffness reduction factor α, and the matching results of the corrected second and third natural frequencies of the model with the measured values. S43. Apply the wind load according to the design conditions, calculate the deflection of the glass panel, set up a safety early warning classification mechanism, and output the damage assessment results through a digital twin model; In step S43, the standard value of wind load at the location of the curtain wall is calculated according to the building structure load code. , ,in, This is the gust coefficient. This is the wind load shape coefficient. This is the coefficient of wind pressure height variation. This is the basic wind pressure; In the digital twin model, design wind loads with positive and negative pressures are uniformly applied to the glass panel, respectively. Design value of wind load Standard value of wind load 1.4 times that of the wind load, which is applied as a static uniformly distributed pressure; The maximum normal displacement of the glass in the digital twin model under the design wind load is calculated using geometric nonlinearity as the maximum deflection df, with the unit being mm. The maximum deflection df is then normalized to obtain the normalized deflection. The normalized deflection is expressed as: ； Among them, the benchmark value is the maximum deflection value of the glass in the undamaged group, and the damage limit value is the maximum deflection value of the glass when the structural adhesive is severely damaged. Based on the normalized deflection value, the damage is divided into four levels: Level 1 damage: normalized deflection < 0.1; Level 2 damage: 0.1 ≤ normalized deflection < 0.5; Level 3 damage: 0.5 ≤ normalized deflection < 1; Level 4 damage: normalized deflection > 1. Output parameter reports and visualized damage assessment results through digital twin model: Damage coordinates: (X, Y) = (__, __); Corrected stiffness reduction factor: α; Maximum deflection of glass panel at mid-span under standard wind load: df = __ mm and normalized deflection; Warning level: [mild|moderate|severe|damaged]; The visual damage assessment results include the damage-marked areas in the digital twin model and the glass deflection visualization cloud map.

2. The method for detecting structural adhesive damage in curtain walls based on modal difference and digital twins according to claim 1, characterized in that, In S1, for the acquisition of vibration data of the hidden frame glass curtain wall to be tested, i.e. the test group, the laser vibrometer and the shot exciter are controlled to hover within a preset range from the glass panel. The shot exciter launches nylon shot to hit the center of the glass panel. At the same time as the impact, the laser vibrometer synchronously acquires the glass vibration response signal according to the set acquisition point position to obtain the normal acceleration time history data of the glass panel to be tested. For the acquisition of vibration data of frameless glass curtain walls (i.e., the undestructed group) under non-destructive conditions, based on the geometric dimensions, material properties and boundary constraints of the frameless glass curtain wall to be tested, a glass curtain wall structure that is 1:1 replicated with it is built in the laboratory. The vibration data acquisition steps of the frameless glass curtain wall to be tested (i.e., the group to be tested) are repeated to obtain the normal acceleration time history data of the glass panel under non-destructive conditions.

3. A curtain wall structural adhesive damage detection device based on modal difference and digital twin, characterized in that, The method for detecting damage to curtain wall structural adhesive based on modal difference and digital twin as described in any one of claims 1 and 2 includes an unmanned aerial vehicle (UAV) carrier unit (1), a laser vibrometer (2), and a projectile vibrator (3). The UAV carrier unit (1) is connected to the laser vibration meter (2) and the projectile vibrator (3) respectively; The laser vibration meter (2) is used to collect vibration signals; The projectile vibrator (3) is used to excite glass vibration.

Images