Strain gauge array-based high-rise building structure deformation monitoring method and system

Abstract

The application discloses a high-rise building structure deformation monitoring method and system based on a strain gauge array, relates to the technical field of deformation monitoring, and comprises the following steps: arranging a strain gauge array, and acquiring real-time original strain signals; establishing a linear elastic finite element model, extracting an elastic modal basis vector set, and taking a linear space formed by the elastic modal basis vector set as an elastic principal component subspace; calculating a projection component of the original strain signals in the elastic principal component subspace, subtracting the projection component from the original strain signals, and obtaining an orthogonalized residual strain signal; extracting a damage sensitive feature from the orthogonalized residual strain signal, analyzing a deformation state based on the damage sensitive feature, and obtaining a structure deformation monitoring result. The application solves the technical problem that local damage deformation in a high-rise building structure is difficult to accurately identify in the prior art, and achieves the technical effect of improving the accuracy of high-rise building structure damage deformation monitoring.

Description

Technical Field

[0001] This invention relates to the field of deformation monitoring technology, specifically to a method and system for monitoring the deformation of high-rise building structures based on strain gauge arrays. Background Technology

[0002] In the health monitoring of high-rise building structures, strain sensors are typically deployed to acquire strain response data under wind loads, earthquakes, and other forces. However, the acquired strain signals are often a superposition of overall elastic deformation and abnormal responses caused by local damage, with the overall structural response dominating and easily masking subtle local changes caused by cracks, plastic development, etc. The lack of effective signal separation and feature extraction methods results in insufficient ability to identify local damage deformation, making it difficult to reflect the true damage state of the structure in a timely and accurate manner. Summary of the Invention

[0003] This application provides a method and system for monitoring the deformation of high-rise building structures based on strain gauge arrays, which is used to address the technical problem that it is difficult to accurately identify local damage and deformation in high-rise building structures in the prior art.

[0004] In view of the above problems, this application provides a method and system for monitoring the deformation of high-rise building structures based on strain gauge arrays.

[0005] The first aspect of this application provides a method for monitoring the deformation of high-rise building structures based on strain gauge arrays, the method comprising: Strain gauge arrays are deployed at key stress-bearing components of the high-rise building structure to collect raw strain signals in real time. A linear elastic finite element model corresponding to the high-rise building structure is established. Elastic modal basis vector sets are extracted from the linear elastic finite element model, and the linear space spanned by the elastic modal basis vector sets is used as the elastic principal component subspace. The projection components of the raw strain signals in the elastic principal component subspace are calculated, and the projection components are subtracted from the raw strain signals to obtain orthogonalized residual strain signals. Damage-sensitive features are extracted from the orthogonalized residual strain signals, and deformation state analysis is performed based on the damage-sensitive features to obtain structural deformation monitoring results.

[0006] A second aspect of this application provides a deformation monitoring system for high-rise building structures based on strain gauge arrays, the system comprising: The system includes a signal acquisition module for deploying strain gauge arrays at key stress-bearing components of the high-rise building structure to acquire raw strain signals in real time; a reference model establishment module for establishing a linear elastic finite element model corresponding to the high-rise building structure, extracting a set of elastic modal basis vectors from the linear elastic finite element model, and using the linear space spanned by the set of elastic modal basis vectors as the elastic principal component subspace; a calculation module for calculating the projection components of the raw strain signal in the elastic principal component subspace, and subtracting the projection components from the raw strain signal to obtain the orthogonalized residual strain signal; and a deformation state analysis module for extracting damage-sensitive features from the orthogonalized residual strain signal, performing deformation state analysis based on the damage-sensitive features, and obtaining structural deformation monitoring results.

[0007] One or more technical solutions provided in this application have at least the following technical effects or advantages: This application deploys strain gauge arrays at key stress-bearing components of high-rise building structures to acquire raw strain signals in real time; establishes a linear elastic finite element model corresponding to the high-rise building structure; extracts a set of elastic modal basis vectors from the linear elastic finite element model; and uses the linear space spanned by the set of elastic modal basis vectors as the elastic principal component subspace; calculates the projection component of the raw strain signal in the elastic principal component subspace, and subtracts the projection component from the raw strain signal to obtain the orthogonalized residual strain signal; extracts damage-sensitive features from the orthogonalized residual strain signal, and performs deformation state analysis based on the damage-sensitive features to obtain structural deformation monitoring results. This invention solves the technical problem of accurately identifying local damage and deformation in high-rise building structures in existing technologies. By performing elastic principal component separation on the strain signal and extracting damage-sensitive features from the orthogonalized residual strain signal, it achieves the technical effect of improving the accuracy of damage and deformation monitoring in high-rise building structures. Attached Figure Description

[0008] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0009] Figure 1 A schematic diagram of the process for a high-rise building structure deformation monitoring method based on strain gauge array provided in this application embodiment; Figure 2 A schematic diagram of a high-rise building structure deformation monitoring system based on strain gauge array provided in this application embodiment.

[0010] Explanation of reference numerals in the attached figures: Signal acquisition module 11, reference model establishment module 12, calculation module 13, deformation state analysis module 14. Detailed Implementation

[0011] This application provides a method and system for monitoring the deformation of high-rise building structures based on strain gauge arrays. It addresses the technical problem of accurately identifying local damage and deformation in high-rise building structures in existing technologies. By performing elastic principal component separation on the strain signal and extracting damage-sensitive features from the orthogonalized residual strain signal, the method improves the accuracy of damage and deformation monitoring in high-rise building structures.

[0012] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0013] It should be noted that any variation of the terms "comprising" and "having" is intended to cover non-exclusive inclusion, for example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or modules that are not explicitly listed or that are inherent to such processes, methods, products, or devices.

[0014] Example 1, as Figure 1 As shown, this application provides a method for monitoring the deformation of high-rise building structures based on strain gauge arrays, the method comprising: Step S100: Install strain gauge arrays at key stress-bearing parts of the high-rise building structure to collect raw strain signals in real time.

[0015] In this embodiment, when monitoring the deformation of a high-rise building structure, the key stress-bearing parts of the high-rise building structure are first extracted from the pre-determined structural design drawings. These key stress-bearing parts are the locations of components in the structural system that are subjected to large forces or are sensitive to deformation. Subsequently, strain gauge arrays are deployed at the aforementioned key stress-bearing parts. The strain gauge array is a collection of multiple strain gauge sensors arranged in a predetermined spatial distribution to simultaneously measure multiple measuring points. Then, each strain gauge continuously senses the strain change at its measuring point during structural operation and outputs a corresponding electrical signal. The output signals of each strain gauge are uniformly collected and processed by a data acquisition device, converting the analog signals into digital signals and summarizing them to obtain the original strain signal composed of the strain values ​​of each measuring point.

[0016] Step S200: Establish a linear elastic finite element model corresponding to the high-rise building structure, extract the set of elastic modal basis vectors from the linear elastic finite element model, and use the linear space spanned by the set of elastic modal basis vectors as the elastic principal component subspace.

[0017] In this embodiment, when establishing the linear elastic finite element model corresponding to the high-rise building structure, the linear elastic finite element model is established based on the design drawings of the high-rise building structure. Next, modal analysis is performed on the linear elastic finite element model to extract the elastic modal basis vector set. Then, an elastic principal component subset space is constructed based on the elastic modal basis vector set.

[0018] Furthermore, in the method provided in the application embodiment, establishing a linear elastic finite element model corresponding to the high-rise building structure, extracting a set of elastic modal basis vectors from the linear elastic finite element model, and using the linear space spanned by the set of elastic modal basis vectors as the elastic principal component subset space, further includes: Based on the design drawings of the high-rise building structure, a linear elastic finite element model is established, wherein the material parameters in the linear elastic finite element model include the elastic modulus and Poisson's ratio; modal analysis is performed on the linear elastic finite element model to extract the first N overall deformation modes, where N is a preset positive integer, and the theoretical strain values ​​of each strain gauge measurement point position under each mode are recorded to form an elastic modal basis vector set; the elastic modal basis vector set is solidified and stored to construct an elastic principal component subset space.

[0019] In this embodiment, based on the design drawings of the high-rise building structure, all input data required for modeling is first compiled. This input data includes at least the total number of floors, floor height, axial dimensions, beam and column locations, shear wall locations, floor slab range, cross-sectional dimensions of each component, node connection relationships, and foundation fixation status. After data compilation, the planar coordinates and elevation coordinates of each node are input into the finite element analysis software according to the design drawings to establish a spatial geometric model consistent with the actual structure. After the geometric model is established, beams and columns are discretized into beam elements, and shear walls and floor slabs are discretized into shell elements. Node connections between elements are established according to the component connection relationships given in the design drawings. Subsequently, material parameters and cross-sectional parameters are input to each element, where the material parameters include elastic modulus. The material and Poisson's ratio, elastic modulus, and cross-sectional parameters are entered according to the material category corresponding to the design drawings. The cross-sectional parameters are entered according to the actual dimensions of the component. After the material and cross-sectional parameters are entered, displacement constraints are applied to the bottom nodes according to the foundation fixing conditions in the design drawings, and connection constraints consistent with the design connection type are applied to the node connection positions. After all constraints are entered, the software reads the length, thickness, cross-sectional dimensions, elastic modulus, and Poisson's ratio of each element. First, the stiffness and mass data of each element are calculated. Then, according to the connection relationship between the element and the node, the stiffness data of all elements are summarized into the stiffness data of the entire structure, and the mass data of all elements are summarized into the mass data of the entire structure. After the above processing, a linear elastic finite element model corresponding to the high-rise building structure, as well as the overall stiffness and overall mass data corresponding to this linear elastic finite element model, are obtained.

[0020] After the online elastic finite element model is established, modal analysis is performed on it. Specifically, the modal analysis module is called in the finite element analysis software, and the Lanczos solution method is used to read the aforementioned overall stiffness and overall mass data. During the calculation process, the software first extracts eigenvalues ​​step by step based on the overall stiffness and overall mass data. For each eigenvalue extracted, a set of nodal displacement results corresponding to that eigenvalue is obtained simultaneously. This set of nodal displacement results represents the relative displacement distribution of all nodes in the structure under that mode. After all eigenvalues ​​and corresponding nodal displacement results are calculated, the results are sorted from low to high according to the natural frequencies, and the first N overall deformation modes are extracted from them. N is a preset positive integer, and the value of N is preset before modeling based on the number of strain gauges and the required modal order for monitoring. When extracting the first N overall deformation modes, the modal order, natural frequency, and all nodal displacement data are saved for each result, so that the corresponding nodal displacement results can be directly called according to the modal order in subsequent calculations. After the above processing, the first N global deformation modes, as well as the natural frequencies and all nodal displacement data corresponding to the first N global deformation modes, are obtained.

[0021] After obtaining the first N overall deformation modes, the theoretical strain value of each strain gauge measuring point under each overall deformation mode is calculated. Specifically, based on the strain gauge layout record, the number, installation floor, installation component, measuring point coordinates, and installation direction of each strain gauge are entered into the finite element analysis software item by item, and a correspondence is established between each strain gauge measuring point and the specific element position in the linear elastic finite element model. After establishing the correspondence, the first overall deformation mode is processed first, and the displacement data of all nodes connected to the element where a certain strain gauge measuring point is located under the first overall deformation mode are read. Since the strain gauge measuring point is usually located at a specific position on the element surface rather than at the element node, the displacement data of each node in the element is interpolated according to the actual coordinate position of the measuring point in the element to obtain the displacement data of the measuring point position. After obtaining the displacement data of the measuring point position, the strain value of the measuring point position on the component surface is calculated according to the relationship between the displacement inside the element and the spatial position. The data is modified; when the strain gauges are pasted along a fixed direction, the strain component consistent with the installation direction is extracted from the strain data at the measuring point according to the installation direction of the strain gauge, and used as the theoretical strain value of the strain gauge in the first overall deformation mode. After completing the calculation of one strain gauge measuring point, the same calculation process is continued for the next strain gauge measuring point until the theoretical strain values ​​of all strain gauge measuring points in the first overall deformation mode are calculated. Then, the same calculation process is followed to calculate the theoretical strain values ​​of all strain gauge measuring points in the second overall deformation mode, the third overall deformation mode, and so on up to the Nth overall deformation mode. After the calculation of each overall deformation mode is completed, the theoretical strain values ​​of all strain gauge measuring points in that mode are arranged in a fixed order according to the strain gauge number from smallest to largest, forming a column of theoretical strain data corresponding to the current mode order. After the above processing, the theoretical strain values ​​of each strain gauge measuring point in the first N overall deformation modes are obtained, as well as multiple columns of theoretical strain data arranged in order of mode order and strain gauge number.

[0022] After all theoretical strain values ​​are calculated, the theoretical strain data corresponding to each overall deformation mode are organized into a fixed storage format. Specifically, the theoretical strain values ​​of all strain gauge measurement points under the first overall deformation mode are first saved as the first column of data in the order of strain gauge numbers. The theoretical strain values ​​of all strain gauge measurement points under the second overall deformation mode are saved as the second column of data in the same order, and so on up to the Nth overall deformation mode. After the column data is organized, the mode order, corresponding natural frequency, strain gauge number, measurement point coordinates, installation direction, and corresponding theoretical strain value are written into the parameter file or database table of the monitoring system, while keeping the strain gauge number order, mode order order, and field format fixed during the writing process. When all data is written, a set of modal theoretical strain data that can be directly called by the monitoring system is formed. Then, all theoretical strain data corresponding to the first N overall deformation modes are stored as a whole, and the linear combination range corresponding to this whole data is determined as the spatial range on which subsequent projection calculations are based. After the above processing, a fixed-stored set of elastic modal basis vectors and an elastic principal component subset space constructed from the set of elastic modal basis vectors are obtained.

[0023] Step S300: Calculate the projection component of the original strain signal in the elastic principal component subspace, and subtract the projection component from the original strain signal to obtain the orthogonalized residual strain signal.

[0024] In this embodiment, when calculating the projection components of the original strain signal in the elastic principal component subspace, the measured strain values ​​of each measuring point acquired by the strain gauge array at the current moment are combined in a predetermined order to form the original strain vector; then, using each elastic modal basis vector in the elastic principal component subspace as a reference, the inner product between the original strain vector and each elastic modal basis vector is calculated to determine the projection coefficients of the original strain vector in the direction of each elastic modal basis vector, thereby obtaining a set of projection coefficients; after obtaining the set of projection coefficients, each projection coefficient is multiplied by its corresponding elastic modal basis vector, and the results are accumulated to obtain the projection components of the original strain vector in the elastic principal component subspace.

[0025] Next, when subtracting the projection component from the original strain signal, the original strain vector is first subtracted from the projection component element by element to obtain the initial residual strain vector; then the residual projection component of the initial residual strain vector in the elastic principal component subspace is calculated, and the initial residual strain vector is further subtracted from the residual projection component to obtain the orthogonalized residual strain signal that is orthogonal to the elastic principal component subspace.

[0026] Furthermore, in the method provided in the application embodiments, calculating the projection component of the original strain signal in the elastic principal component subspace further includes: The original strain signal is represented as the original strain vector composed of the measured values ​​of each strain gauge point in the strain gauge array at the current moment; the inner product of the original strain vector and each elastic modal basis vector in the elastic principal component subspace is calculated in turn to obtain the projection coefficients corresponding to each elastic modal basis vector, thus obtaining the projection coefficient set; each projection coefficient in the projection coefficient set is multiplied by the corresponding elastic modal basis vector and then summed to obtain the projection component.

[0027] In this embodiment, the original strain signal is first processed by reading the measured values ​​of each strain gauge point in the strain gauge array at the current moment according to a preset numbering order and arranging them in the same order to form the original strain vector. Specifically, the numbering order of all strain gauge points in the strain gauge array is first determined, and then the measured strain values ​​of each strain gauge point are read at the same sampling time. Subsequently, the measured strain values ​​are written into the same column of data according to the numbering order, thereby completing the construction of the original strain vector.

[0028] After obtaining the original strain vector, the inner product of the original strain vector with each elastic modal basis vector in the elastic principal component subspace is calculated sequentially. In this process, the first elastic modal basis vector is read first, and the original strain vector is multiplied term by term by the corresponding value in that elastic modal basis vector. All product results are then summed to obtain the projection coefficients corresponding to the first elastic modal basis vector. Then, the second elastic modal basis vector is read, and the second projection coefficient is calculated in the same way. The inner product calculations for the remaining elastic modal basis vectors are then performed sequentially. After all inner product calculations are completed, the projection coefficients are arranged according to the order of the elastic modal basis vectors to obtain the projection coefficient set.

[0029] After obtaining the set of projection coefficients, the projection components are calculated. Specifically, the first projection coefficient in the set is multiplied term by term with the corresponding value of the first elastic mode basis vector to obtain the first set of product results; then the second projection coefficient is multiplied term by term with the corresponding value of the second elastic mode basis vector to obtain the second set of product results; this process is repeated for all projection coefficients and their corresponding elastic mode basis vectors. After all scalar multiplications are completed, the product results are added term by term according to their corresponding positions to obtain the projection components.

[0030] According to the above processing procedure, the original strain signal is first constructed into an original strain vector. Then, the projection coefficient set is obtained by calculating the inner product of the original strain vector and the basis vectors of each elastic mode. Finally, the projection coefficients are multiplied by the corresponding basis vectors of the elastic modes and summed to obtain the projection components of the original strain vector in the elastic principal component subspace.

[0031] Furthermore, the method provided in the application embodiments also includes: The initial residual strain vector is obtained by subtracting the projection component element by element from the original strain vector; the residual projection component of the initial residual strain vector on the elastic principal component subspace is calculated, and the orthogonalized residual strain signal is obtained by subtracting the residual projection component from the initial residual strain vector.

[0032] In this embodiment, when subtracting the projected component element by element from the original strain vector, the values ​​at the first position in the original strain vector are subtracted from the values ​​at the first position in the projected component, according to the data arrangement order in the original strain vector, to obtain the difference at the first position; then, the values ​​at the second position in the original strain vector are subtracted from the values ​​at the second position in the projected component, to obtain the difference at the second position; the remaining positions are subtracted in the same way until all positions have been subtracted. After arranging the subtraction results at all corresponding positions, the initial residual strain vector is formed.

[0033] Next, the residual projection components of the initial residual strain vector on the elastic principal component subspace are calculated. Specifically, the first elastic modal basis vector in the elastic principal component subspace is read first. The initial residual strain vector is multiplied term by term by the corresponding value in this elastic modal basis vector, and all products are summed to obtain the projection coefficient of the initial residual strain vector in the direction of the first elastic modal basis vector. Then, the second elastic modal basis vector is read, and the projection coefficient of the initial residual strain vector in the direction of the second elastic modal basis vector is calculated in the same way. The remaining elastic modal basis vectors are calculated sequentially in the same way to obtain the projection coefficients of the initial residual strain vector relative to all elastic modal basis vectors. Subsequently, each projection coefficient is multiplied term by term by its corresponding elastic modal basis vector, and all products are summed term by term in the corresponding positions to obtain the residual projection components.

[0034] After obtaining the residual projection components, the data at corresponding positions in the initial residual strain vector and the residual projection components are subtracted. Specifically, the value at the first position in the residual projection component is subtracted from the value at the first position in the initial residual strain vector to obtain a new first position result; then, the value at the second position in the initial residual strain vector is subtracted from the value at the second position in the residual projection component to obtain a new second position result; the remaining positions are subtracted in the same way until all positions have been subtracted. After arranging all the subtraction results at corresponding positions, the orthogonalized residual strain signal is obtained.

[0035] Step S400: Extract damage-sensitive features from the orthogonalized residual strain signal, perform deformation state analysis based on the damage-sensitive features, and obtain structural deformation monitoring results.

[0036] In this embodiment, when extracting damage-sensitive features from the orthogonalized residual strain signal, time-domain feature extraction, frequency-domain feature extraction, and spatial-domain feature extraction are performed on the orthogonalized residual strain signal. The time-domain feature extraction is used to obtain peak value, root mean square (RMS), skewness, kurtosis, and zero-crossing rate; the frequency-domain feature extraction is used to obtain the energy percentage within a preset frequency band; and the spatial-domain feature extraction is used to obtain the cross-correlation value and sign consistency index of the residual signals from adjacent measurement points. After completing the above feature extraction, the time-domain features, frequency-domain features, and spatial-domain features are concatenated to form a feature vector, which is then used as the damage-sensitive feature.

[0037] Next, deformation state analysis is performed based on damage-sensitive features. In this process, the damage-sensitive features are input into a pre-trained single-class discriminant model, which compares the current damage-sensitive features with the feature distribution under the condition of structural integrity. When the damage-sensitive features deviate from the feature distribution, it is determined to be an abnormal state. After the abnormal state is determined, damage type matching is performed, and the damage location, damage type, and severity level are output as the structural deformation monitoring results.

[0038] Furthermore, in the method provided in the application embodiments, extracting damage-sensitive features from the orthogonalized residual strain signal further includes: The orthogonalized residual strain signal is subjected to time-domain feature extraction, wherein the time-domain features include peak value, root mean square value, skewness, kurtosis, and zero-crossing rate; the orthogonalized residual strain signal is subjected to frequency-domain feature extraction, wherein the frequency-domain features include the energy percentage within a preset frequency band; the orthogonalized residual strain signal is subjected to spatial-domain feature extraction, wherein the spatial-domain features include the cross-correlation value and sign consistency index of the residual signals of adjacent measuring points; the time-domain features, frequency-domain features, and spatial-domain features are concatenated into a feature vector to obtain the damage-sensitive features.

[0039] In this embodiment, after obtaining the orthogonalized residual strain signal, the signal is first segmented according to a uniform sampling frequency and a uniform analysis time window. Specifically, for each strain gauge measurement point, the orthogonalized residual strain signal is truncated into signal segments of uniform length in chronological order, and the sampled values ​​in each signal segment are used as input data for subsequent feature calculations. Assuming there are L sampling points within a certain analysis time window, the orthogonalized residual strain signal within that analysis time window can be represented as a set of sampled values ​​arranged in chronological order. After signal segmentation, time-domain feature extraction, frequency-domain feature extraction, and spatial-domain feature extraction are performed on the orthogonalized residual strain signal within each analysis time window.

[0040] When extracting time-domain features from orthogonalized residual strain signals, the peak value, root mean square (RMS), skewness, kurtosis, and zero-crossing rate are calculated. Specifically, for an orthogonalized residual strain signal within a certain analysis time window, all sampled values ​​are read point by point, and the absolute value of each sampled value is compared. The sampled value with the largest absolute value is taken as the peak value. Then, all sampled values ​​within the analysis time window are squared, and the squared results are summed item by item. The sum is then divided by the number of sampled points (L), and finally, the square root of the result is taken to obtain the RMS. Before calculating skewness and kurtosis, the average value of all sampled values ​​within the analysis time window is calculated. This is done by summing all sampled values ​​item by item and dividing by the number of sampled points (L). The difference between each sampled value and the average value is then calculated to obtain the deviation of each sampled value. The skewness is calculated by first taking the cube of each sampled value, and then taking the cube of all samples. The results are summed item by item and divided by the number of sampling points L, then divided by the cube of the standard deviation to obtain the skewness. The kurtosis is calculated by first raising the deviation of each sample value to the fourth power, then summing all the fourth power results and dividing by the number of sampling points L, then dividing by the cube of the standard deviation to obtain the kurtosis. The standard deviation is obtained by taking the square root of the average of the squared deviations of each sample value. The zero-crossing rate is calculated by comparing the signs of adjacent sample values ​​sequentially according to the sampling order. A zero-crossing is recorded when two adjacent sample values ​​are one positive and one negative. A zero-crossing is also recorded when a sample value is zero and its preceding and following sample values ​​have different signs. The total number of zero-crossings within the analysis time window is counted and then divided by the number of sampling points minus one to obtain the zero-crossing rate. Following the above calculation process, the peak value, root mean square (RMS), skewness, kurtosis, and zero-crossing rate are calculated sequentially for the orthogonalized residual strain signal corresponding to each analysis time window and each strain gauge measurement point, resulting in the time-domain characteristics of the orthogonalized residual strain signal.

[0041] When extracting frequency domain features from orthogonalized residual strain signals, the energy proportion within a preset frequency band is calculated. Specifically, a Discrete Fourier Transform (DFT) is first performed on the orthogonalized residual strain signal within a certain analysis time window to obtain the spectral amplitude corresponding to each discrete frequency point. Then, the spectral amplitude at each frequency point is squared to obtain the spectral energy corresponding to that frequency point. After calculating the spectral energy of all frequency points, the frequency range corresponding to the preset frequency band is determined according to the preset upper and lower limits of the frequency band, and the spectral energy of each frequency point within this range is accumulated item by item to obtain the total energy within the preset frequency band. Then, the spectral energy of all frequency points is accumulated item by item to obtain the total energy of the entire frequency band. Finally, the total energy within the preset frequency band is divided by the total energy of the entire frequency band to obtain the energy proportion within the analysis time window. When multiple strain gauge measurement points need to be processed separately, the DFT, frequency point truncation, energy accumulation, and ratio calculation are repeatedly performed on the orthogonalized residual strain signal of each strain gauge measurement point within its respective analysis time window to obtain the frequency domain features corresponding to all orthogonalized residual strain signals.

[0042] When extracting spatial domain features from orthogonalized residual strain signals, the cross-correlation value and sign consistency index of the residual signals of adjacent measuring points are calculated. Specifically, the combination of adjacent measuring points is first determined according to the arrangement order of the strain gauge array or the preset adjacency relationship. Then, for each group of adjacent measuring points, the corresponding two groups of orthogonalized residual strain signals within the same analysis time window are read. The cross-correlation value is calculated by multiplying the sampled values ​​of the first and second measuring points at the same sampling position one by one, and then summing all the product results one by one. If necessary, the result is divided by the number of sampling points or normalized according to their respective signal energy to obtain the cross-correlation value of the residual signals of the group of adjacent measuring points. The sign consistency index is calculated by comparing the signs of the sampled values ​​of the group of adjacent measuring points at each sampling position. When the two sampled values ​​are both positive or both negative, they are considered to be consistent. When the two sampled values ​​are one positive and one negative, they are considered to be inconsistent. The number of consistent signs in all sampling positions is counted, and then the number of consistent signs is divided by the total number of sampling points L to obtain the sign consistency index of the residual signals of the group of adjacent measuring points. Following the above calculation process, after calculating the cross-correlation value and sign consistency index for all adjacent measurement point combinations, the spatial domain characteristics corresponding to the orthogonalized residual strain signal are obtained.

[0043] After calculating the time-domain, frequency-domain, and spatial-domain features, these features are concatenated into a feature vector. Specifically, the time-domain features are first arranged in a preset order, with peak value, root mean square (RMS), skewness, kurtosis, and zero-crossing rate written sequentially. Then, the corresponding frequency-domain features within the same analysis time window are written, i.e., the energy percentages within the preset frequency bands are arranged sequentially. Subsequently, the spatial-domain features are written, i.e., the cross-correlation values ​​and sign consistency indices are arranged sequentially according to the numbering order of adjacent measurement point combinations. When all the above features are arranged in a fixed order, a complete feature vector is formed, and this feature vector is used as the damage-sensitive feature.

[0044] Furthermore, in the method provided in the application embodiments, extracting damage-sensitive features from the orthogonalized residual strain signal, performing deformation state analysis based on the damage-sensitive features, and obtaining structural deformation monitoring results, further includes: The damage-sensitive features are input into a pre-trained single-class discrimination model, wherein the single-class discrimination model only learns the feature distribution under the condition of structural integrity. When the damage-sensitive features deviate from the feature distribution, it is judged as an abnormal state. Damage type matching is performed on the abnormal state, and the damage location, damage type and severity level are output as the structural deformation monitoring result.

[0045] In this embodiment, damage-sensitive features are first input into a pre-trained single-class discriminant model, and the data input is completed according to the consistent feature arrangement order during the training phase. The single-class discriminant model is trained only on feature samples under the structural integrity state, and its training result corresponds to the feature distribution range under the structural integrity state. Therefore, after the current damage-sensitive feature is input, the single-class discriminant model judges the distance relationship and distribution relationship between the current damage-sensitive feature and the feature distribution. When the current damage-sensitive feature falls within the feature distribution range, the current structural state is determined to be consistent with the integrity state; when the current damage-sensitive feature exceeds the feature distribution range, the current structural state is determined to be an abnormal state. According to this process, the abnormality judgment result of the current structural state is obtained by first inputting the damage-sensitive feature into the single-class discriminant model.

[0046] After determining that the current structural state is abnormal, damage type matching is performed. Specifically, the orthogonalized residual strain signal corresponding to the current abnormal state and its extracted damage-sensitive features are read and compared item by item with pre-established typical damage mode features. The damage type corresponding to the current abnormal state is determined according to the matching results. After the damage type is determined, the strain gauge number corresponding to the orthogonalized residual strain signal that caused the abnormal state is used to call the pre-established mapping relationship between the strain gauge number and the three-dimensional model of the high-rise building structure. The component position and floor position corresponding to the abnormal strain response are located in the three-dimensional model of the high-rise building structure, thereby determining the component and floor position where the damage occurred, and obtaining the damage location. Then, the cumulative energy value of the orthogonalized residual strain signal is calculated, and the severity level of the damage is determined according to the magnitude of the cumulative energy value. In this process, each sample value of the orthogonalized residual strain signal corresponding to the current abnormal state is squared point by point and accumulated in time order to obtain the cumulative energy value corresponding to the abnormal state. Then, the cumulative energy value is compared with the pre-set severity level threshold range to determine the severity level of the current damage. After determining the location, type, and severity of the damage, a visual damage map and text report containing the location, type, and severity of the damage are generated, and the location, type, and severity of the damage are output as the structural deformation monitoring results.

[0047] Furthermore, the method provided in the application embodiments also includes: Multiple sets of environmental response data of high-rise building structures in the intact state stage are collected, and the environmental response data are sequentially processed by orthogonal residual strain signal extraction and damage-sensitive feature extraction to obtain a set of intact state feature vectors; a single-class support vector machine is used to train the set of intact state feature vectors to obtain a single-class discrimination model.

[0048] Furthermore, the method provided in the application embodiments also includes: During training, each intact state feature vector is orthogonalized to make it orthogonal to the elastic principal component subspace.

[0049] In this embodiment, multiple sets of environmental response data of the high-rise building structure in its intact state are first collected. Specifically, data is collected in the initial stage of structural monitoring when no cracks, plastic deformation, or connection abnormalities have appeared. The collection period can be set to 7 to 30 consecutive days, the sampling frequency can be set to 50 Hz to 200 Hz, the time window length for each collection can be set to 10 s to 60 s, and the overlap rate between adjacent time windows can be set to 50%. After synchronous sampling at each measuring point in the strain gauge array, the measured strain values ​​of all strain gauge measuring points within the same time window are combined into a set of environmental response data, and an original strain vector with the same number of measuring points is constructed. Subsequently, orthogonalized residual strain signal extraction and damage-sensitive feature extraction are sequentially performed on each set of environmental response data. Specifically, the projection component of the original strain vector in the elastic principal component subspace is first calculated, and the projection component is subtracted from the original strain vector to obtain the initial residual strain vector; then the residual projection component of the initial residual strain vector in the elastic principal component subspace is calculated, and the residual projection component is subtracted from the initial residual strain vector to obtain the corresponding orthogonalized residual strain signal. After obtaining the orthogonalized residual strain signal, time-domain, frequency-domain, and spatial-domain features are extracted according to a unified time window. The time-domain features include peak value, root mean square (RMS), skewness, kurtosis, and zero-crossing rate; the frequency-domain features include the energy percentage within a preset frequency band; and the spatial-domain features include the cross-correlation value and sign consistency index of the residual signals from adjacent measurement points. These features are then concatenated in a fixed order to form a set of intact state feature vectors corresponding to the current environmental response data. After processing all environmental response data group by group using the same procedure, multiple sets of intact state feature vectors are obtained and then summarized to form a complete set of intact state feature vectors.

[0050] After forming the set of intact state feature vectors, each intact state feature vector is orthogonalized to make it orthogonal to the elastic principal component subspace. Specifically, the orthogonalized residual strain signal corresponding to the intact state feature vector is first read, and then the residual projection component of the signal on the elastic principal component subspace is checked. When the ratio of the residual projection component to the total energy of the signal is greater than 1% to 5%, the projection subtraction process is repeated until the ratio of the residual projection component to the total energy of the signal is no higher than this range. Then, based on the processed signal, the time domain features, frequency domain features, and spatial domain features are re-extracted, and the corresponding intact state feature vectors are regenerated. After completing the above processing, a set of intact state feature vectors that are orthogonal to the elastic principal component subspace is obtained. Subsequently, a single-class support vector machine is used to train the set of intact state feature vectors. In this training process, all intact state feature vectors are first input into the single-class support vector machine as single-class samples. Then, the kernel function type and training parameters are set. The kernel function can be a radial basis function, and the kernel function parameters can be set to 0.01 to 1. The abnormal sample proportion parameter can be set to 0.01 to 0.1. Before training, each dimension of the features is normalized. During training, the single-class support vector machine learns the feature distribution boundary of the structurally intact state based on the distribution of all intact state feature vectors in the feature space, thus forming a single-class discriminant model used to determine whether subsequent input features deviate from the feature distribution. The result of the training phase is the single-class discriminant model. In the usage phase, the current damage-sensitive features are input into the single-class discriminant model, which then determines whether the current structural state is an abnormal state.

[0051] Furthermore, the method provided in the application embodiments also includes: Mechanical feature templates for multiple typical damage modes are predefined. The typical damage modes include bending plastic hinge, shear crack and bond slip. Each template corresponds to a set of spatial distribution patterns and sign rules of strain residuals. When an abnormal state is determined, the similarity between the spatial distribution of the current orthogonalized residual strain signal and each mechanical feature template is calculated. The damage type corresponding to the mechanical feature template with the highest similarity is selected as the damage type output for the current abnormal state.

[0052] In this embodiment, mechanical characteristic templates for various typical damage modes are predefined. Specifically, based on the three-dimensional model of the high-rise building structure, the location of the strain gauge array, and the stress relationship of the components, three typical damage modes—flexural plastic hinge, shear crack, and bond slip—are established, and degradation parameters are applied to the components corresponding to each damage mode. Among them, the bending stiffness at the end of the component corresponding to the flexural plastic hinge is set to 50% of the initial value, the shear stiffness in the region corresponding to the shear crack is set to 40% of the initial value, and the force transmission stiffness of the connection section corresponding to the bond slip is set to 30% of the initial value. Under each typical damage mode, strain response data were acquired using the same sampling conditions as the monitoring phase. The sampling frequency was set to 100Hz, and the analysis time window was set to 20s. Each strain gauge measurement point corresponded to 2000 sampling points within a single time window. Then, the 2000 residual sampling values ​​for each strain gauge measurement point within that time window were summed point-by-point and divided by 2000 to obtain the average residual value for that measurement point. These residual values ​​were then arranged in the order of strain gauge numbers to form a template vector for the corresponding damage mode. Finally, based on the template vector and the spatial coordinates of each strain gauge measurement point in the 3D model, the spatial distribution morphology and symbol of the damage mode were determined. The process involves several steps. First, the average residual value of each measuring point in the template vector is mapped one-to-one with the spatial coordinates of the corresponding measuring point. Then, according to the component orientation, floor position, and the order of adjacent measuring points, the average residual value of each measuring point is mapped in the three-dimensional model to obtain the distribution of the residual value along the component length direction, both sides of the cross section, and between upper and lower floors. Subsequently, the difference and ratio of the average residual values ​​of adjacent measuring points are calculated to obtain the variation range and direction of the residual value along the spatial position. Based on this, it is determined whether the residual value has a concentrated increase, abrupt change, or gradual decay distribution characteristic in a local area, and this distribution characteristic is determined as the spatial distribution pattern. Regarding the sign pattern, firstly, determine the sign of the average residual value of each measuring point in the template vector, and then statistically analyze the combination relationship of positive and negative signs according to the order of adjacent measuring points. When adjacent measuring points show one positive and one negative sign with similar absolute values, it is recorded as a tension-compression corresponding distribution. When adjacent measuring points have the same sign and their values ​​continuously increase or decrease, it is recorded as a continuous distribution in the same direction. When the signs of local adjacent measuring points suddenly change and the corresponding values ​​change synchronously, it is recorded as a local reverse sign abrupt change distribution. Finally, fix the sign arrangement results of each measuring point and its adjacent relationship to form the sign pattern corresponding to the typical damage mode. After completing the above processing, mechanical characteristic templates corresponding to bending plastic hinges, shear cracks, and bond slip are obtained respectively.

[0053] When an abnormal state is identified, the orthogonalized residual strain signal corresponding to the current abnormal state is read, and the similarity between the current orthogonalized residual strain signal and each mechanical feature template is calculated. Specifically, the orthogonalized residual strain signals of all strain gauge measurement points are read within the current analysis time window; when the sampling frequency is 100Hz and the analysis time window is 20s, each measurement point corresponds to 2000 sampling points. Then, the 2000 sampled values ​​at each measuring point are summed point-by-point and divided by 2000 to obtain the average residual value of that measuring point within the current time window. These residual values ​​are then arranged in strain gauge number order to form the spatial distribution vector corresponding to the current abnormal state. Next, this spatial distribution vector is multiplied item by item with the data at the corresponding position in the first mechanical feature template, and all products are summed to obtain the first product sum. Then, the data at each position of the current spatial distribution vector are squared, summed, and the square root is taken to obtain the first length value. The data at each position of the first mechanical feature template are squared, summed, and the square root is taken to obtain the second length value. Finally, the first product sum is divided by the product of the first and second length values ​​to obtain the similarity between the current abnormal state and the first mechanical feature template. Following the same calculation process, the similarity between the current abnormal state and the remaining mechanical feature templates is calculated, and all similarities are compared. The mechanical feature template with the highest similarity is selected, and the typical damage mode corresponding to this mechanical feature template is determined as the damage type of the current abnormal state. After completing the above processing, the damage type corresponding to the current abnormal state is obtained.

[0054] After determining the damage type, the component and floor where the damage occurred are located in the 3D model of the high-rise building structure based on the strain gauge number corresponding to the orthogonalized residual strain signal that caused the abnormal state. The cumulative energy value of the orthogonalized residual strain signal is then calculated to determine the severity level. In this process, the orthogonalized residual strain signal of each strain gauge measurement point within the current analysis time window is squared and accumulated. Assuming that the residual value of a measurement point at 2000 sampling points is r1 to r2000, the cumulative energy value of that measurement point is the sum of the squares of r1 and r2 up to r2000. Then, the cumulative energy values ​​of all strain gauge measurement points located on the same component and the same floor are accumulated to obtain the cumulative energy value of the component and the cumulative energy value of the floor. Combining the pre-established strain gauge number-component number-floor number-3D coordinate mapping relationship, the component with the largest cumulative energy value and its floor are selected as the damage location. After determining the damage location, the cumulative energy values ​​of all measuring points are summed along the measuring point dimension to obtain the total cumulative energy value of the current abnormal state, which is then compared with a pre-set threshold range. A total cumulative energy value less than 0.50 is considered minor damage, between 0.50 and 1.50 is considered moderate damage, and greater than 1.50 is considered severe damage. After determining the damage location, damage type, and severity level, the corresponding components and floors are highlighted in the 3D model, generating a visualized damage map containing the damage location, damage type, and severity level. Simultaneously, a text report recording the abnormal time, strain gauge number, damage location, damage type, severity level, and total cumulative energy value is generated, thus obtaining the structural deformation monitoring results.

[0055] In summary, the embodiments of this application have at least the following technical effects: This application deploys strain gauge arrays at key stress-bearing components of high-rise building structures to acquire raw strain signals in real time; establishes a linear elastic finite element model corresponding to the high-rise building structure; extracts a set of elastic modal basis vectors from the linear elastic finite element model; and uses the linear space spanned by the set of elastic modal basis vectors as the elastic principal component subspace; calculates the projection component of the raw strain signal in the elastic principal component subspace, and subtracts the projection component from the raw strain signal to obtain the orthogonalized residual strain signal; extracts damage-sensitive features from the orthogonalized residual strain signal, and performs deformation state analysis based on the damage-sensitive features to obtain structural deformation monitoring results. This invention solves the technical problem of accurately identifying local damage and deformation in high-rise building structures in existing technologies. By performing elastic principal component separation on the strain signal and extracting damage-sensitive features from the orthogonalized residual strain signal, it achieves the technical effect of improving the accuracy of damage and deformation monitoring in high-rise building structures.

[0056] Example 2 is based on the same inventive concept as the strain gauge array-based high-rise building structure deformation monitoring method in the previous examples, such as... Figure 2 As shown, this application provides a high-rise building structure deformation monitoring system based on strain gauge arrays. The system and method embodiments in this application are based on the same inventive concept. The system includes: The signal acquisition module 11 is used to deploy strain gauge arrays at key stress-bearing parts of the high-rise building structure to acquire the original strain signals of the high-rise building structure under environmental loads and extreme loads in real time; the reference model establishment module 12 is used to establish a linear elastic finite element model corresponding to the high-rise building structure, extract the elastic modal basis vector set from the linear elastic finite element model, and use the linear space spanned by the elastic modal basis vector set as the elastic principal component subspace; the calculation module 13 is used to calculate the projection component of the original strain signal in the elastic principal component subspace, and subtract the projection component from the original strain signal to obtain the orthogonalized residual strain signal; the deformation state analysis module 14 is used to extract damage-sensitive features from the orthogonalized residual strain signal, perform deformation state analysis based on the damage-sensitive features, and obtain the structural deformation monitoring results.

[0057] Furthermore, the system is also used to implement the following functions: Based on the design drawings of the high-rise building structure, a linear elastic finite element model is established, wherein the material parameters in the linear elastic finite element model include the elastic modulus and Poisson's ratio; modal analysis is performed on the linear elastic finite element model to extract the first N overall deformation modes, where N is a preset positive integer, and the theoretical strain values ​​of each strain gauge measurement point position under each mode are recorded to form an elastic modal basis vector set; the elastic modal basis vector set is solidified and stored to construct an elastic principal component subset space.

[0058] Furthermore, the system is also used to implement the following functions: The original strain signal is represented as the original strain vector composed of the measured values ​​of each strain gauge point in the strain gauge array at the current moment; the inner product of the original strain vector and each elastic modal basis vector in the elastic principal component subspace is calculated in turn to obtain the projection coefficients corresponding to each elastic modal basis vector, thus obtaining the projection coefficient set; each projection coefficient in the projection coefficient set is multiplied by the corresponding elastic modal basis vector and then summed to obtain the projection component.

[0059] Furthermore, the system is also used to implement the following functions: The initial residual strain vector is obtained by subtracting the projection component element by element from the original strain vector; the residual projection component of the initial residual strain vector on the elastic principal component subspace is calculated, and the orthogonalized residual strain signal is obtained by subtracting the residual projection component from the initial residual strain vector.

[0060] Furthermore, the system is also used to implement the following functions: The orthogonalized residual strain signal is subjected to time-domain feature extraction, wherein the time-domain features include peak value, root mean square value, skewness, kurtosis, and zero-crossing rate; the orthogonalized residual strain signal is subjected to frequency-domain feature extraction, wherein the frequency-domain features include the energy percentage within a preset frequency band; the orthogonalized residual strain signal is subjected to spatial-domain feature extraction, wherein the spatial-domain features include the cross-correlation value and sign consistency index of the residual signals of adjacent measuring points; the time-domain features, frequency-domain features, and spatial-domain features are concatenated into a feature vector to obtain the damage-sensitive features.

[0061] Furthermore, the system is also used to implement the following functions: The damage-sensitive features are input into a pre-trained single-class discrimination model, wherein the single-class discrimination model only learns the feature distribution under the condition of structural integrity. When the damage-sensitive features deviate from the feature distribution, it is judged as an abnormal state. Damage type matching is performed on the abnormal state, and the damage location, damage type and severity level are output as the structural deformation monitoring result.

[0062] Furthermore, the system is also used to implement the following functions: Multiple sets of environmental response data of high-rise building structures in the intact state stage are collected, and the environmental response data are sequentially processed by orthogonal residual strain signal extraction and damage-sensitive feature extraction to obtain a set of intact state feature vectors; a single-class support vector machine is used to train the set of intact state feature vectors to obtain a single-class discrimination model.

[0063] Furthermore, the system is also used to implement the following functions: During training, each intact state feature vector is orthogonalized to make it orthogonal to the elastic principal component subspace.

[0064] Furthermore, the system is also used to implement the following functions: Mechanical feature templates for multiple typical damage modes are predefined. The typical damage modes include bending plastic hinge, shear crack and bond slip. Each template corresponds to a set of spatial distribution patterns and sign rules of strain residuals. When an abnormal state is determined, the similarity between the spatial distribution of the current orthogonalized residual strain signal and each mechanical feature template is calculated. The damage type corresponding to the mechanical feature template with the highest similarity is selected as the damage type output for the current abnormal state.

[0065] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0066] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method for monitoring the deformation of high-rise building structures based on strain gauge arrays, characterized in that, The method includes: Strain gauge arrays are deployed at key stress-bearing parts of high-rise building structures to collect raw strain signals in real time. A linear elastic finite element model corresponding to the high-rise building structure is established, an elastic modal basis vector set is extracted from the linear elastic finite element model, and the linear space spanned by the elastic modal basis vector set is used as the elastic principal component subspace. Calculate the projection component of the original strain signal in the elastic principal component subspace, and subtract the projection component from the original strain signal to obtain the orthogonalized residual strain signal; Damage-sensitive features are extracted from the orthogonalized residual strain signal, and deformation state analysis is performed based on the damage-sensitive features to obtain structural deformation monitoring results.

2. The method for monitoring the deformation of high-rise building structures based on strain gauge arrays as described in claim 1, characterized in that, A linear elastic finite element model corresponding to the high-rise building structure is established. An elastic modal basis vector set is extracted from the linear elastic finite element model, and the linear space spanned by the elastic modal basis vector set is used as the elastic principal component subset space, including: Based on the design drawings of the high-rise building structure, a linear elastic finite element model is established, wherein the material parameters in the linear elastic finite element model include the elastic modulus and Poisson's ratio. Modal analysis is performed on the linear elastic finite element model to extract the first N global deformation modes, where N is a preset positive integer. The theoretical strain values ​​of each strain gauge measurement point position under each mode are recorded to form a set of elastic modal basis vectors. The set of elastic modal basis vectors is fixed and stored to construct an elastic principal component subset space.

3. The method for monitoring the deformation of high-rise building structures based on strain gauge arrays as described in claim 1, characterized in that, Calculating the projection components of the original strain signal in the elastic principal component subspace includes: The original strain signal is represented as the original strain vector composed of the measured values ​​of each strain gauge measuring point in the strain gauge array at the current moment; The inner product of the original strain vector and each elastic modal basis vector in the elastic principal component subspace is calculated sequentially to obtain the projection coefficients corresponding to each elastic modal basis vector, thus obtaining the set of projection coefficients. The projection components are obtained by multiplying each projection coefficient in the projection coefficient set by its corresponding elastic modal basis vector and then summing the results.

4. The method for monitoring the deformation of high-rise building structures based on strain gauge arrays as described in claim 3, characterized in that, Also includes: Subtract the projected component element by element from the original strain vector to obtain the initial residual strain vector; The residual projection component of the initial residual strain vector on the elastic principal component subspace is calculated, and the residual projection component is subtracted from the initial residual strain vector to obtain the orthogonalized residual strain signal.

5. The method for monitoring the deformation of high-rise building structures based on strain gauge arrays as described in claim 1, characterized in that, Extracting damage-sensitive features from the orthogonalized residual strain signal includes: The orthogonalized residual strain signal is subjected to time-domain feature extraction, wherein the time-domain features include peak value, root mean square value, skewness, kurtosis and zero-crossing rate; Frequency domain features are extracted from the orthogonalized residual strain signal, wherein the frequency domain features include the energy percentage within a preset frequency band; Spatial domain features are extracted from the orthogonalized residual strain signal, wherein the spatial domain features include the cross-correlation value and sign consistency index of the residual signals of adjacent measuring points; The time-domain features, frequency-domain features, and spatial-domain features are concatenated into a feature vector to obtain the damage-sensitive features.

6. The method for monitoring the deformation of high-rise building structures based on strain gauge arrays as described in claim 1, characterized in that, Damage-sensitive features are extracted from the orthogonalized residual strain signal, and deformation state analysis is performed based on the damage-sensitive features to obtain structural deformation monitoring results, including: The damage-sensitive features are input into a pre-trained single-class discrimination model, wherein the single-class discrimination model only learns the feature distribution under the condition of structural integrity, and when the damage-sensitive features deviate from the feature distribution, it is judged as an abnormal state; The abnormal state is matched for damage type, and the damage location, damage type and severity level are output as the structural deformation monitoring result.

7. The method for monitoring the deformation of high-rise building structures based on strain gauge arrays as described in claim 6, characterized in that, Also includes: Multiple sets of environmental response data of high-rise building structures in the intact state stage are collected, and the environmental response data are sequentially extracted by orthogonalized residual strain signal extraction and damage-sensitive feature extraction to obtain the set of intact state feature vectors. A single-class support vector machine is used to train the set of intact state feature vectors to obtain a single-class discrimination model.

8. The method for monitoring the deformation of high-rise building structures based on strain gauge arrays as described in claim 7, characterized in that, During training, each intact state feature vector is orthogonalized to make it orthogonal to the elastic principal component subspace.

9. The method for monitoring the deformation of high-rise building structures based on strain gauge arrays as described in claim 6, characterized in that, include: Mechanical characteristic templates for multiple typical damage modes are predefined. The typical damage modes include bending plastic hinge, shear crack and bond slip. Each template corresponds to a set of spatial distribution patterns and sign rules of strain residuals. When an abnormal state is determined, the spatial distribution of the current orthogonalized residual strain signal is calculated and its similarity to that of each mechanical feature template is determined. The damage type corresponding to the mechanical feature template with the highest similarity is selected as the damage type output for the current abnormal state.

10. A high-rise building structure deformation monitoring system based on strain gauge array, characterized in that, The system is used to execute the high-rise building structure deformation monitoring method based on strain gauge array as described in any one of claims 1-9, and the system includes: The signal acquisition module is used to deploy strain gauge arrays at key stress-bearing parts of high-rise building structures to acquire raw strain signals in real time. The benchmark model establishment module is used to establish a linear elastic finite element model corresponding to the high-rise building structure, extract the set of elastic modal basis vectors from the linear elastic finite element model, and use the linear space spanned by the set of elastic modal basis vectors as the elastic principal component subspace. The calculation module is used to calculate the projection component of the original strain signal in the elastic principal component subspace, and subtract the projection component from the original strain signal to obtain the orthogonalized residual strain signal. The deformation state analysis module is used to extract damage-sensitive features from the orthogonalized residual strain signal, perform deformation state analysis based on the damage-sensitive features, and obtain structural deformation monitoring results.

Images