A glass curtain wall damage identification method and system

Abstract

The application provides a glass curtain wall damage identification method and system, the method comprising: S1, a collection stage: applying a hammering excitation to a glass curtain wall, and synchronously collecting vibration response signals under the hammering excitation at multiple signal collection measuring points around the periphery of the glass curtain wall; S2, an analysis stage: processing the vibration response signals to calculate time-frequency energy and information entropy thereof; weighting and fusing the time-frequency energy and information entropy to obtain a damage evaluation index vector; S3, an identification stage: mapping the damage evaluation index vector to each signal collection measuring point to generate a damage distribution cloud map representing damage spatial distribution. The method of the application can evaluate the damage condition of the glass curtain wall by analyzing the vibration characteristics of the vibration response signals without disassembling or damaging the glass curtain wall components, can accurately and quickly locate the damage position, and provides a new way for monitoring the service safety of the glass curtain wall.

Description

Technical Field

[0001] This invention relates to the field of structural vibration testing, and in particular to a method and system for identifying damage to glass curtain walls. Background Technology

[0002] Glass curtain walls are a widely used exterior cladding structure in modern buildings. During use, they directly bear normal loads such as wind loads, their own weight, and temperature changes, as well as extreme loads such as earthquakes and typhoons. Their boundary connection conditions (such as silicone structural adhesives) are prone to performance degradation due to rainwater erosion, temperature changes, direct ultraviolet radiation, and external loads, leading to damage or even detachment, which poses safety hazards.

[0003] Although existing technologies have made some progress in damage identification and health monitoring based on vibration response signals of glass curtain wall structures, the following shortcomings still exist, making it difficult to accurately locate the damage to glass curtain walls:

[0004] First, traditional structural dynamic characteristic parameters offer limited information, often establishing only a univariate linear relationship between numerical changes and damage severity, making it difficult to accurately pinpoint the spatial distribution of damage. Furthermore, using other modal characteristics (such as curvature modes) for damage identification requires careful evaluation of the correctness and confidence of the method's assumptions, such as sensitivity to noise and unstable index values. Second, the effectiveness of identifying the spatial distribution of damage in glass curtain walls is positively correlated with the number of measurement points. However, limited by testing costs or the principles of hardware acquisition, most existing technologies employ a single-input, single-output method involving moving excitation or measurement points. This method is based on the reciprocity assumption of input excitation and output response in linear time-invariant systems. However, when damage occurs at the glass curtain wall's boundaries, this directly interferes with the accuracy of the reciprocity assumption. Finally, for signal processing algorithms, the decision to first select a certain decomposition mode, decompose the signal, and then select a specific signal component for subsequent processing lacks physical interpretability.

[0005] To address the aforementioned challenges, there is an urgent need to invent a glass curtain wall damage identification method that is convenient to test, has a clear principle, high analysis efficiency, and requires a small number of testing points, so as to accurately locate the damage position of the glass curtain wall and monitor and warn of its health status. Summary of the Invention

[0006] The main objective of this invention is to propose a method and system for identifying damage to glass curtain walls, which can quickly, reliably and accurately locate the damage to glass curtain walls.

[0007] To achieve the above objectives, the present invention provides the following technical solution:

[0008] In a first aspect, the present invention provides a method for identifying damage to glass curtain walls, comprising the following steps:

[0009] S1. Acquisition Stage: Apply hammering excitation to the glass curtain wall and simultaneously acquire vibration response signals under hammering excitation at multiple signal acquisition points around the perimeter of the glass curtain wall.

[0010] S2. Analysis phase: The vibration response signal is processed to calculate its time-frequency energy and information entropy; the time-frequency energy and information entropy are weighted and fused to obtain the damage evaluation index vector;

[0011] S3. Identification stage: The damage evaluation index vector is mapped to each signal acquisition point to generate a damage distribution cloud map that characterizes the spatial distribution of damage.

[0012] In a second aspect, the present invention provides a glass curtain wall damage identification system for implementing the glass curtain wall damage identification method described in the first aspect, comprising:

[0013] The excitation module is used to apply hammering excitation to the glass curtain wall;

[0014] The acquisition module is used to synchronously acquire the vibration response signal under hammer excitation from multiple signal acquisition points around the perimeter of the glass curtain wall;

[0015] The analysis module, communicatively connected to the acquisition module, is used to receive the vibration response signal and is configured to: process the vibration response signal to calculate its time-frequency energy and information entropy; and perform weighted fusion of the time-frequency energy and information entropy to obtain a damage evaluation index vector.

[0016] The identification module, which is communicatively connected to the analysis module, is used to receive the damage evaluation index vector and is configured to: map the damage evaluation index vector to each signal acquisition measurement point to generate a damage distribution cloud map characterizing the spatial distribution of damage.

[0017] The beneficial effects of this invention compared to existing technologies include: the damage identification method of this invention can not only determine whether a glass curtain wall is damaged, but also accurately locate the specific location of the damage. Furthermore, it has the advantages of reasonable computational assumptions, strong universality, no reliance on sample training, and low economic cost. Overall, this invention can quickly, reliably, and accurately locate the damage location of a glass curtain wall, as detailed below:

[0018] (1) Most previous methods were based on the single-input single-output reciprocity assumption. However, when glass curtain wall components are damaged or the damage state evolves, the structure may exhibit nonlinear response characteristics and change the dynamic transmission path, which in turn causes changes in the transfer function, reducing the applicability of the reciprocity assumption and potentially leading to biased identification results. This invention adopts a single-input multi-output testing and analysis framework, in which the vibration response of each measuring point can be collected synchronously and used for joint analysis, thereby reducing the uncertainty caused by asynchronous measurements, inconsistent operating conditions, or human preprocessing, and improving the rationality of the analysis assumptions and test procedures.

[0019] (2) The core of this invention lies in directly processing the acquired vibration response signal, extracting damage features through two dimensions: time-frequency energy and information entropy, and then weighted and fused to obtain a damage evaluation index vector to further generate a damage distribution cloud map characterizing the spatial distribution of damage. This invention does not rely on a specific signal decomposition paradigm or strong assumptions about a fixed transfer function form. When the dynamic characteristics of the glass curtain wall change due to variations in component size, material properties, or boundary constraints, it will not affect the damage evaluation and location effects of this invention. This invention effectively reduces the risk of algorithm failure caused by changes in structural transfer characteristics, thereby improving the engineering applicability and transferability of the method.

[0020] (3) This invention does not rely on a sample library or model training, but can directly calculate and judge the collected vibration response signals. It can quickly output damage evaluation results and location information after obtaining test data, thus having good real-time performance and engineering application efficiency, and reducing the identification uncertainty caused by insufficient samples, sample bias or working conditions outside the training domain.

[0021] (4) The present invention has low requirements for testing equipment and the number of measurement points. For example, conventional sensors can be used to complete data acquisition, and the number of measurement points can be configured and expanded according to engineering needs. Under the premise of meeting the positioning requirements, the number of measurement points can be reduced to at least 8, thereby reducing costs such as sensor deployment, acquisition channels and test organization, and has high engineering feasibility and economy.

[0022] Other beneficial effects of the embodiments of the present invention will be further described below. Attached Figure Description

[0023] Figure 1 This is a flowchart of the glass curtain wall damage identification method in an embodiment of the present invention;

[0024] Figure 2 This is a schematic diagram of engineering practice data collection and testing in an embodiment of the present invention;

[0025] Figure 3 This is a schematic diagram of short-time Fourier transform time-frequency decomposition of vibration response signal in an embodiment of the present invention;

[0026] Figure 4 This is a schematic diagram illustrating the disordered change and entropy increase of the vibration response signal caused by damage in an embodiment of the present invention;

[0027] Figure 5 This is a flowchart of the fusion process based on the Kalman filter recursive algorithm in an embodiment of the present invention;

[0028] Figure 6 This is a schematic diagram illustrating the actual test data collection and analysis results in an embodiment of the present invention. Detailed Implementation

[0029] To fully demonstrate the technical features and innovative advantages of the present invention, further description and explanation will be provided below in conjunction with the accompanying drawings and specific embodiments.

[0030] In the description of this invention, it should be noted that the following description is merely exemplary and not intended to limit the scope and application of this invention.

[0031] To address the technical challenges of determining the location of glass curtain wall damage in practical engineering projects, as well as the low efficiency and unstable identification results of on-site testing and analysis, such as... Figure 1 As shown in the figure, an embodiment of the present invention proposes a method for identifying damage to glass curtain walls, including:

[0032] S1. Acquisition Stage: Apply hammering excitation to the glass curtain wall and simultaneously acquire vibration response signals under hammering excitation at multiple signal acquisition points around the perimeter of the glass curtain wall.

[0033] S2. Analysis phase: The vibration response signal is processed to calculate its time-frequency energy and information entropy; the time-frequency energy and information entropy are weighted and fused to obtain the damage evaluation index vector;

[0034] S3. Identification stage: The damage evaluation index vector is mapped to each signal acquisition point to generate a damage distribution cloud map that characterizes the spatial distribution of damage.

[0035] In some implementations, in step S1, the hammering excitation is applied to the center of the glass curtain wall multiple times. After the previous hammering excitation ends, the amplitude of the vibration response signal decays to below a preset threshold before the next hammering excitation is applied. By applying the hammering excitation to the center of the glass curtain wall and controlling the application time interval, signal overlap is avoided, providing a stable and reliable data foundation for subsequent analysis.

[0036] In some implementations, in step S1, the number of signal acquisition points is no less than eight, and they are evenly distributed along the four boundaries of the glass curtain wall. By setting no less than eight evenly distributed signal acquisition points, testing costs are saved while ensuring positioning accuracy.

[0037] In some implementations, step S2, calculating the time-frequency energy characteristics, includes: performing a short-time Fourier transform on the vibration response signal to obtain its time-frequency power spectrum, and summing the local energy distribution in the time-frequency domain. Through the short-time Fourier transform and the local energy summation operation, the energy distribution characteristics of the vibration response signal in the time-frequency domain are extracted, providing an energy dimension indicator for damage identification.

[0038] In some implementations, step S2, calculating the information entropy, includes: calculating the information entropy value based on the histogram distribution of the vibration response signal amplitude, and normalizing the information entropy value and performing state transition processing based on the response correction transition coefficient of the signal acquisition measurement points to obtain an information entropy at the same scale as the time-frequency energy. By normalizing the information entropy and performing state transition processing based on the response correction transition coefficient, the scale difference between the time-frequency energy and the information entropy is eliminated, enabling the effective fusion of two feature quantities with different physical meanings within the same framework.

[0039] In some implementations, the weighted fusion in step S2 includes: based on the Kalman filter recursive principle, using the processed time-frequency energy as the predicted value and the processed information entropy as the observed value, calculating through a two-step iterative recursive formula, and outputting the damage evaluation index vector. Through the two-step iterative Kalman filter recursive formula, the weighted fusion of the two feature quantities, time-frequency energy and information entropy, is achieved, fully utilizing the advantages of Kalman filtering in fusing uncertain information, thereby obtaining a more robust and accurate damage evaluation index.

[0040] In some implementations, step S2 includes:

[0041] S2-1: Import the collected hammer excitation signal and vibration response signal into the data analysis software, perform baseline drift removal and mean zeroing preprocessing on the vibration response signal, and extract the amplitude and stable waveform signal segment ip(t) of the vibration response signal during the hammer excitation process for analysis. The hammer excitation signal ip is a function of time t.

[0042] S2-2: The vibration response signal of the glass curtain wall under intact boundary conditions is as follows: The vibration response signal under boundary condition damage is Indicates the number of response data points. This represents the total number of signal acquisition points. The vibration response signal of each signal acquisition point is as follows, where is The number representing the acquisition channel:

[0043]

[0044] S2-3: The window function of the short-time Fourier transform is Vibration response signal In To respond to the index of the data points, the window function has a window length of 100 points. The number of points in the short-time Fourier transform is The window width in the time domain is The window width in the frequency domain is The time-frequency power spectrum obtained by performing a short-time Fourier transform on the vibration response signal is as follows:

[0045]

[0046] in, The time-domain variable representing the window function; Indicates a time-domain index; Indicates frequency domain index; It is the imaginary unit, and its square value is -1; It is a natural constant; It is pi (π).

[0047] The window function divides the vibration response signal into multiple segments of size in the time-frequency domain. A grid composed of squares, where the energy of different squares represents the vibration response signal at a certain point. The local energy distribution within the time-frequency range, obtained by summing the local energies distributed in each grid, yields the following time-frequency energy of the vibration response signal:

[0048]

[0049] in Represents the index of the grid in the time and frequency domains;

[0050] The time-frequency energy damage coefficient vector is defined as follows:

[0051] ;

[0052] S2-4: Define the number of bins for information entropy based on the histogram of vibration response signal amplitude. ,but To fall into the first The probability of each box interval and the formula for calculating the information entropy of the vibration response signal are as follows:

[0053]

[0054] The information entropy of a glass curtain wall under complete boundary conditions is expressed as: The information entropy under boundary condition damage is expressed as: The information entropy is processed by state transition to obtain information entropy on the same scale as the time-frequency energy: the information entropy vector is normalized to... and The formulas for the response correction transfer coefficients at each signal acquisition point are defined as follows:

[0055]

[0056] The corrected transfer matrix before and after damage is composed of the response correction transfer coefficients of each signal acquisition point. The formula for calculating the information entropy after state transition processing is as follows:

[0057]

[0058] in," "" indicates Hadamard multiplication of matrices of the same dimension.

[0059] The information entropy damage coefficient vector is defined as follows:

[0060]

[0061] S2-5: Based on the Kalman filter recursive formula, As a predicted value, As observed values, First, iterate through the Kalman filter recursive formula, using... Observational corrections are made for weighted fusion:

[0062] The first step of the weighted fusion iterative formula is as follows:

[0063]

[0064]

[0065]

[0066]

[0067]

[0068]

[0069] in, These are the time-frequency energy and information entropy logarithmic domain representations of the vibration response signal, respectively. It is the initial state estimate; It is the prior estimate of the first step. This is the posterior estimate after the first update. It is the initial error covariance, and the two are equal; It is the process noise covariance; It is the prior error covariance of the first step; It predicts the noise covariance; It is the Kalman gain matrix of the first step; It is the covariance after the first update is completed; It is the identity matrix;

[0070] The second iterative recursive formula for weighted fusion is as follows:

[0071]

[0072]

[0073]

[0074]

[0075]

[0076] in, It is the prior estimate from the second step. It is the posterior estimate after the second step update; It is the prior error covariance of the second step; It is the Kalman gain matrix of the second step; It is the observation noise covariance; This is a vector of damage assessment indicators.

[0077] In some implementations, step S3 includes: constructing a two-dimensional boundary mapping matrix, wherein the positions of its non-zero elements correspond one-to-one with the actual layout coordinates of the signal acquisition measurement points; filling the values ​​in the damage evaluation index vector into the corresponding non-zero element positions of the two-dimensional boundary mapping matrix to form a discrete index matrix; and performing two-dimensional spatial interpolation processing on the discrete index matrix to generate a continuous damage distribution cloud map.

[0078] In some implementations, the two-dimensional spatial interpolation process employs a cubic interpolation method.

[0079] By constructing a two-dimensional boundary mapping matrix and performing two-dimensional spatial interpolation, discrete boundary measurement point indicators are transformed into continuous damage distribution cloud maps, realizing an intuitive and visual expression of damage location and degree, which greatly facilitates the interpretation by engineers.

[0080] In some implementations, the signal acquisition points acquire vibration response signals through contact sensors arranged on the glass curtain wall, or through non-contact sensors.

[0081] This invention also provides a glass curtain wall damage identification system for implementing the aforementioned glass curtain wall damage identification method, comprising:

[0082] The excitation module is used to apply hammering excitation to the glass curtain wall;

[0083] The acquisition module is used to synchronously acquire the vibration response signal under hammer excitation from multiple signal acquisition points around the perimeter of the glass curtain wall;

[0084] The analysis module, communicatively connected to the acquisition module, is used to receive the vibration response signal and is configured to: process the vibration response signal to calculate its time-frequency energy and information entropy; and perform weighted fusion of the time-frequency energy and information entropy to obtain a damage evaluation index vector.

[0085] The identification module, which is communicatively connected to the analysis module, is used to receive the damage evaluation index vector and is configured to: map the damage evaluation index vector to each signal acquisition measurement point to generate a damage distribution cloud map characterizing the spatial distribution of damage.

[0086] In some embodiments, the excitation module is a force hammer; the acquisition module includes multiple sensors and a data acquisition instrument; and the analysis module and the identification module are integrated into a computer with data analysis software installed.

[0087] The embodiments of the present invention are further described below.

[0088] This embodiment aims to provide a convenient and accurate method for identifying glass curtain wall damage. Its core features include: acquiring vibration response through fixed-point hammer impact excitation and boundary synchronous acquisition; comprehensively utilizing the time-frequency energy and information entropy characteristics of the vibration response signal to characterize the damage; employing an algorithm based on the Kalman filter recursive principle to weightedly fuse the two features to obtain a robust damage evaluation index; and finally visualizing the damage distribution through spatial mapping and interpolation. In this example, the hammer impact excitation is applied by a hammer, and the glass curtain wall damage identification method includes:

[0089] S1. Acquisition Phase: Apply hammering excitation to the glass curtain wall and use multiple sensors arranged around the perimeter of the glass curtain wall to synchronously acquire the vibration response signal of the glass curtain wall.

[0090] S2. Analysis stage: Import the vibration response signal into the data analysis software (MATLAB in this example). First, perform a short-time Fourier transform on the vibration response signal to calculate its time-frequency energy. Then, calculate the information entropy of the vibration response signal. Finally, substitute the two into the Kalman filter recursive algorithm for weighted fusion to obtain the damage evaluation index vector.

[0091] S3. Identification Stage: The damage evaluation index vector is mapped to the signal acquisition points of the sensors, and a damage distribution cloud map (numerical thermal cloud map) representing the spatial distribution of damage is drawn in the data analysis software. In the cloud map, the darker the color, the more severe the damage.

[0092] In this embodiment, the acquisition stage in step S1 specifically includes:

[0093] The object to be collected is the glass curtain wall 1 connected to the main building structure. After preparing a computer 2 (equipped with data analysis and data acquisition software), a data acquisition instrument 3, sensors 4, and a hammer 5 at the data collection site, sensors 4 are deployed along the four boundaries of the glass curtain wall 1. For example... Figure 2 As shown, the center of the glass curtain wall 1 is repeatedly struck by a hammer 5. The hammering force used in this example is in the range of 8 N - 30 N. The sensor 4 collects and captures the vibration response signal after each hammering excitation and transmits it to the data acquisition instrument 3. The data acquisition instrument 3 then transmits it to the data acquisition software in the computer 2 to display the signal waveform.

[0094] In this embodiment, the sensor collection points are only deployed on the four sides of the glass curtain wall. The number of sensors on each side of the glass curtain wall is the same, and the distance between adjacent measurement points is a fixed value (such as a fixed value in the range of 50 mm to 500 mm). The total number of sensors is not less than 8.

[0095] In this embodiment, the hammer impact point is fixed at the center of the curtain wall. When applying the hammer, the hammer head axis is kept perpendicular to the plane of the glass curtain wall, and the duration of each impact is the same. After the previous impact ends, the next impact is applied only after the amplitude of the vibration response signal decays to below a preset threshold (e.g., 5% of the maximum absolute value) to avoid signal overlap.

[0096] In this embodiment, the analysis stage in step S2 specifically includes the following steps:

[0097] S2-1: Import the collected hammer excitation signal and vibration response signal into the data analysis software, perform baseline drift removal and mean zeroing preprocessing on the vibration response signal, and extract the amplitude and stable waveform signal segment ip(t) of the vibration response signal during the hammer excitation process for analysis. The hammer excitation signal (force signal) ip is a function of time t.

[0098] S2-2: The vibration response signal of the glass curtain wall under intact boundary conditions is as follows: The vibration response signal under boundary condition damage is (Signal acquisition can be carried out during the regular inspection and testing of glass curtain wall projects.) Indicates the number of response data points. This represents the total number of signal acquisition points (sensors). The vibration response signals of each sensor are as follows, where is... The number representing the acquisition channel:

[0099]

[0100] S2-3: The window function of the short-time Fourier transform is (In this example, the Hamming window is used for analysis), vibration response signal In To respond to the index of the data points, the window function has a window length of 100 points. (In this example, the value is 512), the number of points in the short-time Fourier transform is (In this example, the value is 1024), and the window width in the time domain is... The window width in the frequency domain is The window width is determined based on the actual sampling frequency; in this example, the sampling frequency is 500 Hz. The time-frequency power spectrum obtained by performing a short-time Fourier transform on the vibration response signal is as follows:

[0101]

[0102] in, The time-domain variable representing the window function; Indicates a time-domain index; Indicates frequency domain index; It is the imaginary unit, and its square value is -1; It is a natural constant; It is pi (π).

[0103] like Figure 3 As shown in the left figure, the window function divides the vibration response signal into multiple segments of size in the time-frequency domain. The grid is composed of small squares, and the energy of different squares represents the vibration response signal at a certain point. Local energy distribution within the time-frequency range, such as Figure 3 As shown in the right figure, after meshing, the local time-frequency domain energy in each mesh can be represented as a histogram using the short-time Fourier transform. The time-frequency energy of the vibration response signal obtained by summing the local energies distributed in each mesh is as follows:

[0104]

[0105] in Represents the index of the grid in the time and frequency domains;

[0106] The time-frequency energy damage coefficient vector is defined as follows:

[0107] ;

[0108] S2-4: Define the number of bins for information entropy based on the histogram of vibration response signal amplitude. The number of bins can be determined by the sample size, or it can be analyzed by taking values ​​in the range [100, 200] based on engineering experience. To fall into the first The probability of each box interval and the formula for calculating the information entropy of the vibration response signal are as follows:

[0109]

[0110] The information entropy of a glass curtain wall under complete boundary conditions is expressed as: The information entropy under boundary condition damage is expressed as: To perform weighted fusion of time-frequency energy and information entropy at the same scale, a state transition process needs to be applied to one of them. In this embodiment, the information entropy is processed as follows to obtain an information entropy at the same scale as the time-frequency energy: the information entropy vector is normalized to... and The formulas for the response correction transfer coefficients of each sensor are defined as follows:

[0111]

[0112] The modified transfer matrix before and after damage is composed of the modified transfer coefficients of the responses of each sensor. The formula for calculating the information entropy after state transition processing is as follows:

[0113]

[0114] in," "" indicates Hadamard multiplication of matrices of the same dimension.

[0115] like Figure 4 As shown, the disorder of the vibration response signal differs before and after the damage. To assess the difference in information entropy before and after the damage, the information entropy damage coefficient vector is defined as follows:

[0116]

[0117] S2-5: As Figure 5 As shown, based on the Kalman filter recursive formula, As a predicted value, As observed values, First, iterate through the Kalman filter recursive formula, using... Observational corrections are made for weighted fusion:

[0118] The first step of the weighted fusion iterative formula is as follows:

[0119]

[0120]

[0121]

[0122]

[0123]

[0124]

[0125] in, These are the time-frequency energy and information entropy logarithmic domain representations of the vibration response signal, respectively. It is the initial state estimate; It is the prior estimate of the first step. This is the posterior estimate after the first update. It is the initial error covariance, and the two are equal; It is the process noise covariance; the initialization of the matrix can be determined according to the actual test conditions. It is the prior error covariance of the first step; It is the prediction noise covariance, which reflects the degree of disturbance in the predicted value; It is the Kalman gain matrix of the first step; It is the covariance after the first update is completed; It is the identity matrix;

[0126] The second iterative recursive formula for weighted fusion is as follows:

[0127]

[0128]

[0129]

[0130]

[0131]

[0132] in, It is the prior estimate from the second step. It is the posterior estimate after the second step update; It is the prior error covariance of the second step; It is the Kalman gain matrix of the second step; It is the observation noise covariance, which reflects the degree of disturbance in the observed values; This is a vector of damage assessment indicators; the higher the value, the more severe the damage.

[0133] In this embodiment, the analysis stage in step S3 specifically includes:

[0134] In actual testing and data acquisition, sensors are deployed only along the curtain wall boundary. To achieve quantitative representation of the deployment locations, a two-dimensional matrix corresponding to the discrete grid of the glass curtain wall is constructed as a spatial mapping carrier. All internal grid cells except the outermost ring are assigned a value of zero, forming a boundary matrix that only represents the positions of boundary measurement points. In this embodiment, a one-to-one correspondence is established between the coordinates of the outermost non-zero cells of this boundary matrix and the actual sensor deployment coordinates, while grid cells without deployed sensors remain at zero values. The non-zero elements in this matrix represent the effective locations of the sensor deployments, and the number of non-zero elements is equal to the number of sensors.

[0135] The vibration response signals collected by each sensor are calculated and analyzed to obtain damage evaluation indicators, which are then combined into an indicator vector. Based on the coordinate correspondence mentioned above, the indicator vector is mapped to the corresponding non-zero unit of the boundary matrix, thereby obtaining a discrete indicator matrix that characterizes the spatial distribution of damage.

[0136] In the visualization output stage, the color mapping boundary of the heat map is pre-defined, and spatial smoothing and interpolation processing of the discrete index matrix is ​​determined based on the number and density of measurement points. In this example, cubic interpolation using two-dimensional interpolation is employed to generate a continuous damage distribution cloud map. In the damage distribution cloud map, darker colors indicate a higher degree of damage in that area. Figure 6 As shown, the left figure is a schematic diagram of silicone structural adhesive delamination damage at the glass curtain wall boundary. The silicone structural adhesive is represented by black bars in the figure, and the length range of the delamination damage (305 mm) is marked in the figure. The middle figure is the experimentally measured glass curtain wall boundary delamination condition, and the area marked by the box is the actual delamination damage range of the glass curtain wall. The right figure shows the damage location identification of the glass curtain wall that has actually suffered delamination damage using the method and system of the present invention. It can be clearly seen that the position of the dark block identified by the cloud map is consistent with the actual location of the delamination damage on the glass curtain wall, verifying the feasibility of the method of the present invention, and the result can be stably reproduced. Figure 6The parameters selected for the example effect shown are as follows: the number of sensor acquisition points is 16, and the type is a contact IEPE accelerometer; the spacing between the sensors is 120 mm; the sampling frequency set in the data acquisition is 500 Hz; the window length of the short-time Fourier transform is 512 points, and the number of short-time Fourier transform points is 1024; the information entropy bin number is 100; the cloud map interpolation uses the "cubic" cubic interpolation command in the "interp2" interpolation command of MATLAB software.

[0137] In summary, specific embodiments of the present invention provide a method for identifying damage to glass curtain walls. Under more reasonable mechanical testing settings, the damage identification method of the present invention can accurately locate the specific location of damage through numerical cloud maps of damage evaluation indicators. It also possesses adaptability to damage-induced nonlinearity and changes in transmission paths, weak dependence on models and decomposition paradigms, rapid discrimination capability without sample training, and low-cost engineering feasibility, significantly improving the stability, transferability, and application timeliness of damage identification results.

[0138] In the method embodiments provided by this invention, vibration response signals of the glass curtain wall are acquired at signal acquisition points. The medium for acquiring the vibration response signals at the signal acquisition points is not strictly limited in form. It can be implemented using contact acquisition methods (e.g., accelerometers, velocity sensors, displacement sensors, strain gauges, or strain measurement devices connected by cables), or non-contact acquisition methods (e.g., testing equipment based on optical vibration measurement, laser vibration measurement, visual measurement, infrared or electrical remote measurement principles), or any combination of the above methods. Without contradiction, those skilled in the art can replace or modify the type, layout, and signal transmission method of the acquisition device according to actual testing conditions, installation environment, and accuracy requirements; all such modifications should be considered within the protection scope of this invention.

[0139] In the method embodiments provided by this invention, all or part of the steps of the method can be executed by related hardware driven by program instructions. The program instructions can be called and executed by a processor (e.g., a general-purpose processor, a digital signal processor, a field-programmable gate array, an application-specific integrated circuit, or a combination thereof); the related hardware may include, but is not limited to, a data acquisition module, a signal conditioning module, a memory, a communication interface, and a cooperating computing unit. Accordingly, this invention can also be implemented as a computer program product, which is stored in a computer-readable storage medium and, when executed by a processor, implements all or part of the steps of the method of this invention; the computer-readable storage medium includes, but is not limited to, a read-only memory, a random access memory, a magnetic disk, an optical disk, a portable storage device, and other media capable of storing program instructions.

[0140] Furthermore, although the present invention has been described in conjunction with the above embodiments, those skilled in the art should understand that, without departing from the concept of the present invention, the execution carrier of the steps can be replaced by software implementation, hardware implementation, or a combination of software and hardware implementation; the combination, decomposition, order adjustment, or equivalent substitution of the steps can also be varied according to specific application requirements. All variations of equivalent substitutions made in accordance with the present invention specification should be considered to fall within the protection scope of the present invention.

[0141] The above description, in conjunction with specific preferred embodiments, provides a further detailed explanation of the present invention. It should not be construed that the specific implementation of the present invention is limited to these descriptions. For those skilled in the art, various equivalent substitutions or obvious modifications can be made without departing from the concept of the present invention, and all such modifications, achieving the same performance or application, should be considered within the scope of protection of the present invention.

Claims

1. A method for identifying damage to glass curtain walls, characterized in that, Includes the following steps: S1. Acquisition Stage: Apply hammering excitation to the glass curtain wall and simultaneously acquire vibration response signals under hammering excitation at multiple signal acquisition points around the perimeter of the glass curtain wall. S2. Analysis phase: The vibration response signal is processed to calculate its time-frequency energy and information entropy; the time-frequency energy and information entropy are weighted and fused to obtain the damage evaluation index vector; S3. Identification stage: Map the damage evaluation index vector to each signal acquisition point to generate a damage distribution cloud map that characterizes the spatial distribution of damage. The weighted fusion in step S2 includes: based on the Kalman filter recursive principle, using the processed time-frequency energy as the predicted value and the processed information entropy as the observed value, calculating through a two-step iterative recursive formula, and outputting the damage evaluation index vector; including the following steps: S2-1: Import the acquired hammer excitation signal and vibration response signal into the data analysis software. Perform baseline drift removal and mean zeroing preprocessing on the vibration response signal, and extract the signal segments with stable amplitude and waveform during the hammer excitation process. IP (t) is analyzed to determine the hammer excitation signal. IP It is a function of time t; S2-2: The vibration response signal of the glass curtain wall under intact boundary conditions is as follows: The vibration response signal under boundary condition damage is Indicates the number of response data points. This represents the total number of signal acquisition points. The vibration response signal of each signal acquisition point is as follows, where is The number representing the acquisition channel: ； S2-3: The window function of the short-time Fourier transform is Vibration response signal In To respond to the index of the data points, the window function has a window length of 100 points. The number of points in the short-time Fourier transform is The window width in the time domain is The window width in the frequency domain is The time-frequency power spectrum obtained by performing a short-time Fourier transform on the vibration response signal is as follows: ； in, The time-domain variable representing the window function; Indicates a time-domain index; Indicates frequency domain index; It is the imaginary unit, and its square value is -1; It is a natural constant; It is pi; The window function divides the vibration response signal into multiple segments of size in the time-frequency domain. A grid composed of squares, where the energy of different squares represents the vibration response signal at a certain point. The local energy distribution within the time-frequency range, obtained by summing the local energies distributed in each grid, yields the following time-frequency energy of the vibration response signal: ； in Represents the index of the grid in the time and frequency domains; The time-frequency energy damage coefficient vector is defined as follows: ； S2-4: Define the number of bins for information entropy based on the histogram of vibration response signal amplitude. ,but To fall into the first The probability of each box interval and the formula for calculating the information entropy of the vibration response signal are as follows: ； The information entropy of a glass curtain wall under complete boundary conditions is expressed as: The information entropy under boundary condition damage is expressed as: The information entropy is processed by state transition to obtain an information entropy on the same scale as the time-frequency energy: the information entropy vector is normalized to... and The formulas for the response correction transfer coefficients at each signal acquisition point are defined as follows: ； The corrected transfer matrix before and after damage is composed of the response correction transfer coefficients of each signal acquisition point. The formula for calculating the information entropy after state transition processing is as follows: ； in," " represents Hadamard multiplication of matrices of the same dimension; The information entropy damage coefficient vector is defined as follows: ； S2-5: Based on the Kalman filter recursive formula, As a predicted value, As observed values, First, iterate through the Kalman filter recursive formula, using... Observational corrections are made for weighted fusion: The first step of the weighted fusion iterative formula is as follows: ； ； ； ； ； ； in, These are the time-frequency energy and information entropy logarithmic domain representations of the vibration response signal, respectively. It is the initial state estimate; It is the prior estimate of the first step. This is the posterior estimate after the first update. It is the initial error covariance, and the two are equal; It is the process noise covariance; It is the prior error covariance of the first step; It predicts the noise covariance; It is the Kalman gain matrix of the first step; It is the covariance after the first update is completed; It is the identity matrix; The second iterative recursive formula for weighted fusion is as follows: ； ； ； ； ； in, It is the prior estimate from the second step. It is the posterior estimate after the second step update; It is the prior error covariance of the second step; It is the Kalman gain matrix of the second step; It is the observation noise covariance; This is a vector of damage assessment indicators.

2. The glass curtain wall damage identification method as described in claim 1, characterized in that, In step S1, the hammering excitation is applied to the center of the glass curtain wall and is applied multiple times. After the previous hammering excitation is completed, the amplitude of the vibration response signal is attenuated to below a preset threshold before the next hammering excitation is applied.

3. The glass curtain wall damage identification method as described in claim 1, characterized in that, In step S1, the number of signal acquisition points is no less than 8, and they are evenly distributed along the four boundaries of the glass curtain wall.

4. The glass curtain wall damage identification method as described in claim 1, characterized in that, In step S2, the calculation of time-frequency energy characteristics includes: performing a short-time Fourier transform on the vibration response signal to obtain its time-frequency power spectrum, and summing the local energy distribution in the time-frequency domain.

5. The glass curtain wall damage identification method as described in claim 1, characterized in that, In step S2, calculating the information entropy includes: calculating the information entropy value based on the histogram distribution of the vibration response signal amplitude, and performing normalization processing on the information entropy value and state transition processing based on the response correction transition coefficient of the signal acquisition measurement point to obtain the information entropy on the same scale as the time-frequency energy.

6. The glass curtain wall damage identification method as described in claim 1, characterized in that, Step S3 includes: constructing a two-dimensional boundary mapping matrix, the positions of its non-zero elements corresponding one-to-one with the actual layout coordinates of the signal acquisition measurement points; filling the values ​​in the damage evaluation index vector into the corresponding non-zero element positions of the two-dimensional boundary mapping matrix to form a discrete index matrix; performing two-dimensional spatial interpolation processing on the discrete index matrix to generate a continuous damage distribution cloud map.

7. The glass curtain wall damage identification method as described in claim 6, characterized in that, The two-dimensional spatial interpolation process employs a cubic interpolation method.

8. A glass curtain wall damage identification system, used to implement the glass curtain wall damage identification method according to any one of claims 1-7, characterized in that, include: The excitation module is used to apply hammering excitation to the glass curtain wall; The acquisition module is used to synchronously acquire the vibration response signal under hammer excitation from multiple signal acquisition points around the perimeter of the glass curtain wall; The analysis module, communicatively connected to the acquisition module, is used to receive the vibration response signal and is configured to: process the vibration response signal to calculate its time-frequency energy and information entropy; and perform weighted fusion of the time-frequency energy and information entropy to obtain a damage evaluation index vector. The identification module, which is communicatively connected to the analysis module, is used to receive the damage evaluation index vector and is configured to: map the damage evaluation index vector to each signal acquisition measurement point to generate a damage distribution cloud map characterizing the spatial distribution of damage.

Images