A Smart Testing Method for Simulating the Seismic Reliability of Nursing Bed Structures

Abstract

This invention discloses an intelligent testing method for simulating the seismic reliability of nursing bed structures, belonging to the field of seismic testing technology for medical and nursing equipment. The method performs real-time acquisition of multi-dimensional strain response of the nursing bed structure, obtaining strain time sequences in the first, second, and third directions of key connection nodes under preset simulated seismic excitation; it traverses the strain time sequences to perform joint calculations of energy distribution coefficients in both time and frequency domains, determining structural nonlinear response characteristic factors, structural energy dissipation characteristic factors, and structural stiffness degradation characteristic factors; using these characteristic factors as search anchors, it performs parameter matching of a four-parameter Bouc-Wen hysteresis model in the seismic test parameter space of the nursing bed structure to determine the target test dynamic load spectrum; based on the target test dynamic load spectrum, it performs multi-axis synchronous simulated seismic excitation loading according to a preset set of test conditions to obtain a set of vibration response characteristic values, and after identification, obtains the seismic reliability test results of the nursing bed structure.

Description

Technical Field

[0001] This invention relates to the field of seismic testing technology for medical and nursing equipment, specifically to an intelligent testing method for simulating the seismic reliability of a nursing bed structure. Background Technology

[0002] Seismic reliability testing of nursing bed structures is used to assess the safety and functional retention of medical and nursing equipment under seismic loading. Existing testing methods often employ uniaxial sinusoidal frequency sweeps or unidirectional seismic wave input for excitation, acquiring strain or acceleration responses at measuring points, and then evaluating seismic performance based on linear damage criteria. These methods assume the structural response is within the elastic range, neglecting the nonlinear hysteretic behavior of the nursing bed's connection nodes under multidirectional vibration coupling, resulting in insufficient correlation between simulated seismic loads and actual structural responses. At the data characterization level, conventional tests only extract peak strain or root mean square values ​​in a single direction as characteristic quantities, failing to simultaneously capture the three interrelated damage information types—nonlinear response, energy dissipation, and stiffness degradation—during the time-varying process of multidimensional vibration. This makes subsequent load spectrum design unable to reflect the true evolution path of the structure from elastic to plastic with accompanying stiffness decay. At the load generation level, conventional methods directly apply standard response spectra or historical seismic acceleration records, ignoring the actual hysteretic characteristic parameters of the nursing bed structure, leading to a mismatch between the applied excitation and the structure's dynamic constitutive relationship, often resulting in underloading or overloading, making it difficult for the test results to accurately reflect true seismic reliability. Therefore, how to comprehensively extract multi-dimensional features reflecting the nonlinear behavior of a structure from multi-directional strain responses, and how to generate a load spectrum that matches the actual dynamic characteristics of the structure based on these features, have become key issues in improving the accuracy of seismic testing of nursing bed structures. Summary of the Invention

[0003] This paper presents an intelligent testing method for simulating the seismic reliability of nursing bed structures. It comprehensively extracts nonlinear response, energy dissipation, and stiffness degradation characteristics from multi-dimensional strain response, and uses these characteristics to adaptively match a test load spectrum that matches the actual dynamic hysteresis characteristics of the structure. This solves the problem of distorted seismic reliability assessment caused by the mismatch between the load spectrum and the actual structural response in existing testing methods.

[0004] To achieve the above objectives, the present invention provides the following technical solution: The present invention provides an intelligent testing method for simulating the seismic reliability of a nursing bed structure, comprising: performing real-time acquisition of multi-dimensional strain response of the nursing bed structure to obtain the first-direction strain time sequence, the second-direction strain time sequence, and the third-direction strain time sequence of key connection nodes of the nursing bed structure under a preset simulated seismic excitation; traversing the first-direction strain time sequence, the second-direction strain time sequence, and the third-direction strain time sequence to perform joint calculation of time-frequency dual-domain energy distribution coefficients to determine the structural nonlinear response characteristic factors and structural energy dissipation characteristics. The method employs a combination of structural nonlinear response characteristic factors and structural stiffness degradation characteristic factors. Using these as search anchors, a four-parameter Bouc-Wen hysteresis model is matched within the seismic test parameter space of the nursing bed structure to determine the target dynamic load spectrum. Based on this target dynamic load spectrum, the nursing bed structure undergoes multi-axis synchronous simulated seismic excitation loading according to a preset set of test conditions to obtain a set of vibration response characteristic values. A structural damage identification indicator is then used to identify these vibration response characteristic values, yielding the seismic reliability test results for the nursing bed structure. This method achieves accurate characterization of structural nonlinear behavior, energy dissipation, and stiffness degradation through joint analysis of multi-dimensional strain and dual-domain features. Furthermore, by combining historical model matching and multi-axis loading, it effectively improves the equivalence of simulated seismic tests and the accuracy of damage identification.

[0005] As a preferred technical solution of the present invention, in the real-time acquisition of multi-dimensional strain response, strain gauges are pasted and arranged on the key connection nodes of the nursing bed structure and three-dimensional strain rosette sensors are installed. The three sensitive grids are aligned with the transverse, longitudinal and vertical directions of the structure, respectively. A preset simulated earthquake excitation device is activated to apply low-frequency sinusoidal frequency sweep excitation, and the voltage signals output by the three sensitive grids are recorded synchronously. After differential amplification and analog-to-digital conversion, they are mapped to the strain time sequence in the three directions, thereby ensuring the synchronization and accuracy of multi-directional strain information and providing a reliable data foundation for subsequent feature factor extraction.

[0006] As another preferred technical solution of the present invention, the joint operation of the time-frequency dual-domain energy distribution coefficient specifically includes: performing a short-time Fourier transform on the strain time series in the first direction to obtain the time-spectrum matrix; extracting the amplitude corresponding to the peak frequency of each time window and forming an instantaneous frequency-amplitude curve; performing cubic spline interpolation on the curve and calculating the curve integral to obtain the structural nonlinear response characteristic factor; performing the same short-time Fourier transform and interpolation integration operation on the strain time series in the second direction to obtain the structural energy dissipation characteristic factor; performing the same operation on the strain time series in the third direction to obtain the structural stiffness degradation characteristic factor; and normalizing the structural nonlinear response characteristic factor, structural energy dissipation characteristic factor, and structural stiffness degradation characteristic factor respectively to unify the numerical range of the three characteristic factors to the interval of 0 to 1. Through the joint analysis of the time-frequency dual domain, the strain response in different directions can be transformed into independent nonlinear indices, and the normalization process eliminates the dimensional differences, facilitating high-precision matching and retrieval in the parameter space.

[0007] As a further improvement of the present invention, the process of determining the target dynamic load spectrum using the aforementioned feature factors as search anchor points includes: using the three normalized feature factors as search coordinate values, extracting a preset number of historical test load spectra with the smallest spatial distance from the search coordinate values ​​from the seismic test parameter space of the nursing bed structure; calculating the yield displacement, post-yield stiffness ratio, hysteresis loop shape control parameters, and pinching effect parameters of the four-parameter Bouc-Wen hysteresis model corresponding to each historical test load spectrum; using the weighted average of the four parameters as the initial parameters of the target dynamic load spectrum, and iteratively correcting the initial parameters using the Runge-Kutta numerical integration method until the deviation between the hysteresis loop area corresponding to the corrected parameters and the structural energy dissipation feature factor converges to a preset threshold. The target dynamic load spectrum obtained in this way integrates historical test experience with current structural characteristics, can accurately reproduce the nonlinear hysteresis energy dissipation behavior of the nursing bed structure, and significantly improves the relevance and testing effectiveness of the loading spectrum.

[0008] As another preferred embodiment of the present invention, the step of multi-axis synchronous simulated seismic excitation loading includes: decomposing the target test dynamic load spectrum into transverse load components, longitudinal load components, and vertical load components, and inputting them respectively into the servo valve control module of the electro-hydraulic servo actuator, controlling the electro-hydraulic servo actuator to apply multi-axis synchronous seismic excitation to the nursing bed structure according to each component; recording the acceleration response time series data, displacement response time series data, and force response time series data of the nursing bed structure under excitation, and combining them to form the vibration response characteristic value set. Through spectral decomposition and multi-axis synchronous loading, the multi-dimensional seismic motion coupling effect is realistically reproduced, so that the measured vibration response characteristic value set can comprehensively reflect the dynamic characteristics of the structure.

[0009] In another preferred embodiment of the present invention, the process of using a structural damage identification indicator to identify the vibration response feature value set and obtain the seismic reliability test results of the nursing bed structure includes: performing multi-window singular spectrum decomposition on the acceleration response time series data, displacement response time series data, and force response time series data respectively; setting the length of the first window to one-third of the total length, the length of the second window to one-sixth of the total length, and the length of the third window to one-tenth of the total length; constructing trajectory matrices and performing singular value decomposition on each; extracting the feature vectors corresponding to the first three singular values ​​of each trajectory matrix and reconstructing them by averaging to obtain principal component components; subtracting the principal component components from the original data to obtain the residual components; and then... The modal confidence criterion is calculated by combining the principal component components with the theoretical modal shapes obtained from the finite element model of the nursing bed structure. The calculation method involves using the square of the dot product of the test modal shape and the theoretical modal shape as the numerator, and the product of their respective dot products as the denominator, taking the ratio as the modal confidence score. Modal confidence scores below a preset threshold are marked as potential damage areas. Potential damage areas are then merged into connected components based on spatial adjacency. The total number of nodes in each merged connected component is counted as the damage count, and this count is divided by the total number of critical connection nodes in the nursing bed structure to obtain the distribution density. This damage count and distribution density are used as the seismic reliability test results of the nursing bed structure. Multi-window singular spectrum decomposition can effectively separate the principal components and noise residues in the response signal, improving the robustness of modal parameter identification. Combining the modal confidence criterion and connected component statistics, automatic location and quantitative assessment of structural damage can be achieved, providing objective criteria for the seismic reliability of the nursing bed.

[0010] The technical effects and advantages provided by the present invention in the above technical solution are as follows: A triaxial strain gauge array was used to locate key connection nodes of the nursing bed, simultaneously acquiring transverse, longitudinal, and vertical strain time sequences. Short-time Fourier transform and cubic spline interpolation integration were performed on each sequence to obtain the corresponding structural nonlinear response characteristic factors, structural energy dissipation characteristic factors, and structural stiffness degradation characteristic factors. Through this joint calculation of energy distribution coefficients in both time and frequency domains, the nonlinearity, energy absorption capacity, and stiffness attenuation trend of connection nodes under multi-directional vibration coupling can be quantified into three-dimensional characteristic coordinates with unified dimensions. This overcomes the limitation of single-directional peak extraction, which can only reflect local elastic strain, and allows for accurate characterization of the complete hysteretic behavior of the structure under multi-axis excitation. Using the three characteristic factors as retrieval coordinates, the historical test load spectrum with the smallest spatial distance was extracted from the seismic test parameter space of the nursing bed structure. Based on this, the yield displacement, post-yield stiffness ratio, hysteresis loop shape control parameters, and pinching effect parameters of the four-parameter Bouc, Wen hysteresis model were obtained. The weighted average of historical load spectra is used as the initial value of the target test dynamic load spectrum. This is then iteratively corrected using Runge and Kutta numerical integration until the deviation between the hysteresis loop area and the structural energy dissipation characteristic factor converges. This matching process directly controls the generation of the load spectrum to the characteristics of the measured nonlinear hysteresis system, replacing the externally preset loading waveform that is detached from the actual hysteretic characteristics of the structure. This allows multi-axis synchronous seismic excitation to track the stiffness degradation and dissipation changes of the structure after it enters the plastic phase in real time. The resulting set of vibration response characteristic values ​​can more realistically induce potential damage, improving the ability of the nursing bed seismic reliability test results to reproduce the actual strong earthquake response. Attached Figure Description

[0011] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0012] Figure 1 This is a flowchart of an intelligent testing method for simulating the seismic reliability of nursing bed structures. Figure 2 This is a flowchart of the strain measurement method for key connection nodes of a nursing bed; Figure 3 This is a flowchart of the feature factor extraction and normalization process for multi-directional strain signals; Figure 4 These are the three-dimensional strain time-series response curves of key connection nodes in the structure of the nursing bed; Figure 5 It is the instantaneous frequency-amplitude curve of the triaxial strain time series; Figure 6 These are the multi-axis acceleration response time-series data of the nursing bed structure; Figure 7 It is the result of multi-window singular spectrum decomposition of the time series data of the structural acceleration response of the nursing bed. Detailed Implementation

[0013] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0014] See Figure 1 This invention provides an intelligent testing method for simulating the seismic reliability of a nursing bed structure, comprising: performing real-time acquisition of multi-dimensional strain response of the nursing bed structure to obtain the first-direction strain time sequence, the second-direction strain time sequence, and the third-direction strain time sequence of key connection nodes of the nursing bed structure under a preset simulated seismic excitation; performing joint calculation of time-frequency dual-domain energy distribution coefficients on the first-direction strain time sequence, the second-direction strain time sequence, and the third-direction strain time sequence to determine the structural nonlinear response characteristic factor, the structural energy dissipation characteristic factor, and the structural stiffness degradation characteristic factor; using the structural nonlinear response characteristic factor, the structural energy dissipation characteristic factor, and the structural stiffness degradation characteristic factor as search anchor points, performing four-parameter Bouc-Wen hysteresis model parameter matching in the seismic test parameter space of the nursing bed structure to determine the target test dynamic load spectrum; and performing multi-axis synchronous simulated seismic excitation loading on the nursing bed structure according to a preset test condition set based on the target test dynamic load spectrum to obtain a set of vibration response characteristic values, and using a structural damage identification indicator to identify the set of vibration response characteristic values ​​to obtain the seismic reliability test results of the nursing bed structure.

[0015] Example 1:

[0016] In specific implementation, please refer to Figure 2When attaching strain gauges to key connection nodes of the nursing bed structure, the surfaces of these nodes are ground and cleaned to remove oil and rust. Cyanoacrylate adhesive is used to adhere the strain gauges to the surface of the key connection nodes, and pressure is applied for curing. A silicone rubber protective layer is then applied to the strain gauge surfaces to prevent electromagnetic interference and mechanical damage. A triaxial strain gauge sensor is installed at the key connection node. This sensor contains three sensitive grids arranged at 45 or 60 degrees to each other, and the substrate of the triaxial strain gauge sensor is fixed to the surface of the key connection node with adhesive. The first sensitive grid of the triaxial strain gauge sensor is aligned with the transverse direction of the nursing bed structure, the second sensitive grid with the longitudinal direction, and the third sensitive grid with the vertical direction. During alignment, a level and angle measuring tools are used to ensure that the deviation of the first sensitive grid axis from the transverse direction, the second sensitive grid axis from the longitudinal direction, and the third sensitive grid axis from the vertical direction are all within a preset allowable angle.

[0017] In some embodiments, the preset allowable angle is set to ±1 degree.

[0018] A preset simulated seismic excitation device was activated to apply a low-frequency sinusoidal sweep excitation to the nursing bed structure. The preset simulated seismic excitation device was an electro-hydraulic servo seismic simulation vibration table. The excitation signal output by the vibration table was a sinusoidal sweep signal with a frequency linearly increasing from 0.2 Hz to 5 Hz, a sweep duration of 60 seconds, and an acceleration amplitude maintained at 0.1g. The vibration table surface and the nursing bed structure base were rigidly connected by bolts, and the connection stiffness was checked by pre-tightening force.

[0019] The voltage signals output from three sensitive grids are recorded synchronously. Each of the three sensitive grids is connected to an independent Wheatstone bridge circuit, using a half-bridge or full-bridge configuration to compensate for temperature effects. The bridge output is connected to a differential amplifier, which differentially amplifies the voltage signal to suppress common-mode noise. The differential amplifier gain is set to 1000. The differentially amplified voltage signal is input to a synchronous analog-to-digital converter (ADC). The ADC sampling frequency is set to 1000 Hz, with a resolution of 16 bits. All channels are synchronously acquired on the rising edge of the same sampling clock. After analog-to-digital conversion, the voltage signal is mapped into a first-direction strain time sequence, a second-direction strain time sequence, and a third-direction strain time sequence based on the sensitivity coefficient of the triaxial strain gauge sensor, the bridge excitation voltage, and the differential amplifier gain. The mapping uses the following relationship: the sampling time in the first-direction strain time sequence... strain value Obtained from the following formula:

[0020] in, Sampling time The output voltage value of the analog-to-digital converter corresponding to the first sensitive gate is in volts; The sensitivity coefficient of the triaxial strain gauge sensor is provided in the sensor's factory calibration report, and the unit is microstrain per volt. The excitation voltage for the bridge circuit is 2.5 volts DC. The gain of the differential amplifier is set to 1000. The second-direction strain time series and the third-direction strain time series use the same mapping relationship, only differing in the formula. Replace with the voltage value corresponding to the second sensitive gate. The voltage value corresponding to the third sensitive gate .

[0021] See Figure 4 The figure shows the time-series strain patterns in the first, second, and third directions, collected by a triaxial strain gauge at a key connection node of the nursing bed structure in Example 1, as a function of time. The horizontal axis represents time, ranging from 0 to 60 seconds, corresponding to the low-frequency sinusoidal sweep excitation process applied by the preset simulated earthquake excitation device; the vertical axis represents strain value, in microstrain (με), with a value range of approximately -150 to 150 microstrains. The curves use solid lines, dashed lines, and dotted lines to represent the strain time-series in the first, second, and third directions, respectively, and the legend clearly identifies each strain time sequence.

[0022] From the curve trends, the strain timing in all three directions exhibits amplitude modulation characteristics that gradually increase over time. In the initial stage (0-5 seconds), the amplitude rapidly increases from a small value to approximately 90-100 microstrains, followed by a scanning stage (approximately 5-60 seconds) where the frequency increases linearly and the amplitude increases slowly. This corresponds to the sinusoidal sweep frequency signal in Example 1, where the vibration table output frequency increases linearly from 0.2 Hz to 5 Hz. The vibration period of the three-directional strain curves gradually shortens, reflecting the rising excitation frequency. The amplitudes of the three-directional strain timing remain basically synchronized throughout the frequency sweep process, and the waveform trends are highly consistent, indicating that the deformation responses of key connection nodes in the transverse, longitudinal, and vertical directions are coordinated and stable, meeting the orthogonal arrangement and precise alignment requirements of the sensitive grids of the three-directional strain flower sensor in Example 1. Furthermore, the curves show no obvious nonlinear distortion or abrupt changes, indicating that the nursing bed structure did not experience significant damage or stiffness degradation during this stage, satisfying the accuracy requirements for real-time strain signal acquisition and subsequent structural feature factor extraction in Example 1.

[0023] Example 2:

[0024] In specific implementation, please refer to Figure 3When performing a short-time Fourier transform on the strain time series in the first direction, a Hanning window was selected as the window function for the short-time Fourier transform. The window function length was set to 256 sampling points, the overlap length between adjacent time windows was set to 128 sampling points, and the number of discrete points in the frequency domain was set to 512. The strain value at each sampling time in the strain time series in the first direction was then processed. After multiplying with the window function, a Fast Fourier Transform is performed to obtain the local spectrum corresponding to each time window. The local spectra of all time windows are arranged in order of their start times to form the time-spectrum matrix of the strain time series in the first direction. In the time-spectrum matrix, the row index corresponds to the frequency interval, the column index corresponds to the start time of the time window, and the matrix element values ​​are the amplitudes of the strain signal at the corresponding frequency and within the corresponding time window.

[0025] To extract the amplitude corresponding to the peak frequency of each time window in the time-spectrum matrix, iterate through each column of the matrix and find the row index of the column containing the maximum amplitude. The frequency corresponding to that row index is the peak frequency of that time window, and the maximum value is the amplitude corresponding to that peak frequency. Arrange the amplitudes corresponding to the peak frequencies of each time window in chronological order of the start time of the time window to form an instantaneous frequency-amplitude curve. The horizontal axis of the instantaneous frequency-amplitude curve represents the start time of the time window, and the vertical axis represents the amplitude corresponding to the peak frequency.

[0026] When performing cubic spline interpolation on the instantaneous frequency-amplitude curve, the starting time of the time window in the curve is used as the interpolation node, and the amplitude corresponding to the peak frequency is used as the function value at the interpolation node, constructing a cubic spline interpolation function with continuous second derivative. Interpolation is performed within the time range of the instantaneous frequency-amplitude curve at a preset interpolation step size, which is set to one-tenth of the original sampling interval. After interpolation, the instantaneous frequency-amplitude curve is numerically integrated from the starting time to the ending time. The integral is calculated using the composite Simpson formula, and the integration result is used as the numerical value of the structural nonlinear response characteristic factor.

[0027] The same short-time Fourier transform and interpolation integration operations as those for the first direction strain time series are performed on the second-direction strain time series. In the short-time Fourier transform, the window function type, window function length, overlap length, and number of frequency domain discrete points are exactly the same as when processing the first-direction strain time series. The steps for extracting the instantaneous frequency-amplitude curve, performing cubic spline interpolation, and calculating the curve integral are exactly the same as when processing the first-direction strain time series. The final integral result is used as the numerical value of the structural energy dissipation characteristic factor. The same process is repeated for the third-direction strain time series to obtain the numerical value of the structural stiffness degradation characteristic factor.

[0028] The structural nonlinear response characteristic factor, structural energy dissipation characteristic factor, and structural stiffness degradation characteristic factor are normalized to unify their numerical range to the interval of 0 to 1. The normalization process uses a maximum-minimum normalization method. Taking the structural nonlinear response characteristic factor as an example, the normalization formula is as follows:

[0029] in, This represents the numerical value of the structural nonlinear response characteristic factor before normalization; This represents the minimum value of the structural nonlinear response characteristic factor in historical test data under the same test environment of the nursing bed structure. This minimum value is extracted from the pre-established seismic test parameter space of the nursing bed structure. This represents the maximum value of the structural nonlinear response characteristic factor in historical test data under the same test environment of the nursing bed structure. This maximum value is extracted from the pre-established seismic test parameter space of the nursing bed structure. This represents the normalized structural nonlinear response characteristic factor, with a value ranging from 0 to 1. The structural energy dissipation characteristic factor and the structural stiffness degradation characteristic factor use the exact same normalization method, only differing in the formula... , , and The corresponding values ​​were replaced with the structural energy dissipation characteristic factor and the structural stiffness degradation characteristic factor. After the attribution process, three characteristic factors with values ​​uniformly ranging from 0 to 1 were obtained.

[0030] See Figure 5The horizontal axis in the figure represents the start time of the time window in seconds, and the vertical axis represents the amplitude of the peak frequency of the strain signal within the corresponding time window in microstrain (με). The figure shows the instantaneous frequency-amplitude curves of the first, second, and third directions measured by a three-dimensional strain gauge at the key connection nodes of the nursing bed structure, represented by blue solid dots, red dashed squares, and green dotted triangles, respectively. The curves generally show a slow upward trend over time, reflecting the gradual increase in the strain response amplitude of the key nodes of the nursing bed structure under the preset simulated seismic excitation, indicating the process of structural strain accumulation and enhanced nonlinear response. Within the interval from 0 to approximately 10 seconds, the amplitude fluctuations in the three directions are relatively stable and similar, with amplitudes ranging from approximately 70 to 100 με. As time progresses to around 40 seconds, the amplitude shows obvious intermittent peaks, especially in the second direction, where the amplitude curve exhibits more significant peak fluctuations, with an increased peak frequency and amplitude, reaching a maximum of nearly 135 με, indicating a rapid increase in the structural energy dissipation characteristic factor corresponding to this direction. The amplitude curve in the third direction is generally lower than that in the second direction and slightly lower than that in the first direction, but the trends are similar, with frequent fluctuations in local peaks.

[0031] Example 3:

[0032] In practical implementation, when using structural nonlinear response characteristic factors, structural energy dissipation characteristic factors, and structural stiffness degradation characteristic factors as search coordinate values, the normalized value of the structural nonlinear response characteristic factors is recorded as the first search coordinate value. The normalized structural energy dissipation characteristic factor value is recorded as the second coordinate value for retrieval. The normalized structural stiffness degradation characteristic factor value is recorded as the third coordinate value for retrieval. By retrieving the first coordinate value 1. Retrieve the second coordinate value and retrieving the third coordinate value Construct a three-dimensional search coordinate vector .

[0033] When extracting a preset number of historical test load spectra with the smallest distance from the search coordinate value space from the seismic test parameter space of the nursing bed structure, the seismic test parameter space of the nursing bed structure is a pre-constructed structured database. Each record in the database corresponds to a set of historical test data, including the historical characteristic factor set and historical test load spectrum data obtained from that historical test. Each record also stores the four-parameter Bouc-Wen hysteresis model parameters identified from that historical test load spectrum. The historical characteristic factor set consists of normalized historical structural nonlinear response characteristic factor values, normalized historical structural energy dissipation characteristic factor values, and normalized historical structural stiffness degradation characteristic factor values, recorded in the form of a three-dimensional vector. For the search coordinate vector... With the first in the parameter space The spatial distance between the historical feature factor vectors of each record is measured using Euclidean distance. A preset number of 5 is set based on the fact that, in the commonly used k-nearest neighbor search, selecting 5 neighboring samples achieves a balance between local estimation accuracy and computational cost. Furthermore, cross-validation of historical test data for nursing bed structures shows that the mean prediction deviation of the 5 neighboring samples is lower than the mean prediction deviation corresponding to larger or smaller preset numbers. All records are traversed from the seismic test parameter space of the nursing bed structure, and the spatial distance corresponding to each record is calculated. Records are sorted in ascending order of spatial distance, and the historical test load spectra corresponding to the top 5 records with the smallest spatial distances are extracted.

[0034] When calculating the yield displacement, post-yield stiffness ratio, hysteresis loop shape control parameters, and pinching effect parameters of the four-parameter Bouc-Wen hysteresis model corresponding to each historical test load spectrum, the stored four-parameter Bouc-Wen hysteresis model parameters are directly read from the five extracted historical test load spectrum records. In each record, the yield displacement of the four-parameter Bouc-Wen hysteresis model is denoted as the parameter. The stiffness ratio after yielding is denoted as a parameter. The hysteresis loop shape control parameter is denoted as parameter. The pinching effect parameter is denoted as parameter. .

[0035] When the four parameters of the historical test load spectrum are weighted and averaged to serve as the initial parameters for the target test dynamic load spectrum, the weighted average calculation formula is expressed as:

[0036] in, The index of the extracted historical test load spectrum, with values ​​ranging from 1 to... integers, The preset quantity is 5. Indicates the first The parameters to be averaged corresponding to the historical test load spectra are sequentially the yield displacement parameters. Post-yield stiffness ratio parameter Hysteresis loop shape control parameters and pinching effect parameters As Substituting into the formula and performing weighted average calculation, we obtain the initial yield displacement, initial post-yield stiffness ratio, initial hysteresis loop shape control parameters, and initial pinching effect parameters of the target test dynamic load spectrum. Indicates assigning the first Weighting factors for historical test load spectrum parameters. Indicates the first Historical feature factor vector and retrieval coordinate vector of historical test load spectrum records The spatial distance between them. Let be a very small positive number, taking the value of This is used to prevent division by zero errors when the spatial distance is zero. The value is based on the lower limit of the effective number of bits of a computer's double-precision floating-point number, which ensures the numerical stability of the weight calculation without significantly changing the reciprocal weight relationship of the spatial distance.

[0037] When iteratively correcting the initial parameters using the Runge-Kutta numerical integration method, a sinusoidal sweep frequency displacement loading sequence is set as the standard input displacement time history. The amplitude of the sinusoidal sweep frequency displacement loading sequence is 1.2 times the elastic limit displacement of the nursing bed structure, the frequency range is 0.1 Hz to 2 Hz, and the sweep duration is 40 seconds. The initial yield displacement, initial post-yield stiffness ratio, initial hysteresis loop shape control parameters, and initial pinching effect parameters of the target test dynamic load spectrum are substituted into the differential equation of the four-parameter Bouc-Wen hysteresis model. The fourth-order Runge-Kutta numerical integration method is used to solve the differential equation, with a numerical integration step size set to 0.001 seconds. The hysteresis force values ​​at each time point are obtained, and combined with the input displacement sequence to construct a force-displacement hysteresis curve. The area of ​​the hysteresis loop enclosed by the force-displacement hysteresis curve is calculated and denoted as the area. The structural energy dissipation characteristic factor is compared with the reference elastic energy. The product of these factors is used as the target area, with reference to the elastic energy. The maximum elastic energy of the equivalent linear system of the nursing bed structure under the elastic limit displacement is taken as half of the product of the elastic limit displacement and the elastic limit restoring force in a single cycle. The area deviation is calculated. ,in This is the normalized structural energy dissipation characteristic factor. When the area deviation... When the area deviation exceeds a preset threshold, the central difference method is used to calculate the numerical gradient of the four parameters with respect to the hysteresis loop area. The values ​​of the four parameters are adjusted according to the gradient descent direction with a preset step size. The Runge-Kutta numerical integration is then performed again to solve the hysteresis loop and calculate the area deviation. The preset threshold is set to 0.01, based on the fact that the standard deviation of repeatability tests of energy dissipation indicators in engineering is usually controlled within 1%. The iteration process continues until the area deviation is reached. The test terminates when the value is less than or equal to a preset threshold for the first time. The parameter combination at the time of termination is used as the corrected parameters, and the target test dynamic load spectrum is determined by the corrected parameters.

[0038] Example 4:

[0039] In practical implementation, when the target test dynamic load spectrum is decomposed into lateral load components, longitudinal load components, and vertical load components, the target test dynamic load spectrum is a two-dimensional data sequence containing the force-time correspondence, where the measurement point position corresponds to the coordinate position of the connection point between the electro-hydraulic servo actuator and the bed frame on the nursing bed structure. The decomposition process is performed based on the local coordinate system of the nursing bed structure. The local coordinate system takes the center of the bed frame plane as the origin, the lateral direction is parallel to the minor axis of the bed frame, the longitudinal direction is parallel to the major axis of the bed frame, and the vertical direction is perpendicular to the bed frame plane and upwards. For the force vector at each sampling moment in the target test dynamic load spectrum, the force vector is orthogonally projected onto the three coordinate axes of the local coordinate system. The force components projected onto the lateral coordinate axis constitute the lateral load component sequence, the force components projected onto the longitudinal coordinate axis constitute the longitudinal load component sequence, and the force components projected onto the vertical coordinate axis constitute the vertical load component sequence. The projection calculation at each sampling moment maintains the vector composition relationship between the magnitude of the force vector and the magnitudes of the three components.

[0040] When the lateral, longitudinal, and vertical load components are input to the servo valve control module of the electro-hydraulic servo actuator, each load component corresponds to a control channel of an electro-hydraulic servo actuator. The lateral load component sequence is converted into a voltage command signal, with an amplitude range set from -10 volts to +10 volts. The conversion ratio factor is determined by the force sensor calibration coefficient of the electro-hydraulic servo actuator. The voltage command signal undergoes closed-loop regulation using the PID control algorithm built into the servo valve control module. The proportional gain of the PID control algorithm is set to 1.2, the integral time constant to 0.05 seconds, and the derivative time constant to 0.01 seconds. The servo valve control module outputs a current drive signal to the electro-hydraulic servo valve, controlling the displacement and output force of the hydraulic cylinder piston rod. The longitudinal and vertical load components use the same conversion ratio factor and PID control algorithm parameters, and are input to the corresponding servo valve control module through their respective electro-hydraulic servo actuator control channels.

[0041] When the electro-hydraulic servo actuators apply multi-axis synchronous seismic excitation to the nursing bed structure according to the lateral, longitudinal, and vertical load components, the installation directions of the three electro-hydraulic servo actuators are aligned with the lateral, longitudinal, and vertical directions of the local coordinate system of the nursing bed structure, respectively. The three electro-hydraulic servo actuators are connected to the loading points of the nursing bed structure via ball joints, with the loading points located on the reinforcing ribs at the four corners of the bed frame. Synchronous control of the multi-axis synchronous seismic excitation is achieved through a distributed clock synchronization module. This module sends a unified sampling trigger clock signal to the servo valve control modules of the three electro-hydraulic servo actuators. The trigger clock signal frequency is set to 1000 Hz, and the standard time reference is provided by the global positioning system timing module. The three electro-hydraulic servo actuators synchronously update their force output values ​​at the rising edge of the trigger clock signal, with a synchronization deviation of no more than 10 microseconds.

[0042] When recording the acceleration, displacement, and force response time-series data of the nursing bed structure under multi-axis synchronous seismic excitation, triaxial accelerometers, laser displacement sensors, and three-component force sensors were deployed at key connection nodes of the nursing bed structure. The triaxial accelerometer's range was set to ±5g, and its sensitivity to 100 mV / g; the laser displacement sensor's range was set to ±50 mm, and its resolution to 1 μm; the three-component force sensors' range was set to ±10 kN, and its accuracy class to 0.1. The output signals of all sensors were acquired through a synchronous data acquisition system, with a uniform sampling frequency of 1000 Hz, and the sampling synchronization deviation between channels was controlled within 100 nanoseconds. The acquired lateral, longitudinal, and vertical acceleration data were arranged in chronological order to form acceleration response time-series data; the lateral, longitudinal, and vertical displacement data were arranged in chronological order to form displacement response time-series data; and the lateral, longitudinal, and vertical force data were arranged in chronological order to form force response time-series data. Acceleration response time series data, displacement response time series data, and force response time series data are combined into a three-dimensional array as a set of vibration response feature values. The first dimension of the three-dimensional array represents the time sampling point, the second dimension represents the measurement point location index, and the third dimension represents the response type, which includes three categories: acceleration, displacement, and force.

[0043] When performing multi-window singular spectrum decomposition on acceleration response time-series data, displacement response time-series data, and force response time-series data, the multi-window singular spectrum decomposition operation is performed on the acceleration response time-series data, displacement response time-series data, and force response time-series data in the vibration response eigenvalue set, respectively. Taking acceleration response time-series data as an example, the window length set is set to include a first window length, a second window length, and a third window length. The first window length is one-third of the total length of the acceleration response time-series data, the second window length is one-sixth of the total length of the acceleration response time-series data, and the third window length is one-tenth of the total length of the acceleration response time-series data. For each window length, a trajectory matrix is ​​constructed from the acceleration response time-series data according to the window length. Each column of the trajectory matrix is ​​a data segment extracted from the acceleration response time-series data according to the window length, and one sampling point slides between columns. Singular value decomposition (SVD) is performed on the trajectory matrix corresponding to each window length. The left singular vectors corresponding to the first three largest singular values ​​of each trajectory matrix are extracted as eigenvectors. The three eigenvectors obtained from the first, second, and third window lengths are averaged, and the averaged eigenvectors are reconstructed as principal component components of the acceleration response time series data. The principal component components are subtracted from the acceleration response time series data to obtain the residual components. The same multi-window singular spectrum decomposition operation is applied to the displacement response time series data and the force response time series data to obtain the principal component components and residual components of the displacement response time series data, and the principal component components and residual components of the force response time series data, respectively.

[0044] When calculating the modal confidence criterion using the principal component components and the theoretical mode shapes obtained from the finite element model based on the nursing bed structure, the finite element model based on the nursing bed structure is established during the nursing bed structure design stage. In the finite element model, the bed frame, guardrails, legs, and hinged connections of the nursing bed structure are all discretized using three-dimensional solid elements. Material properties are assigned based on the measured elastic modulus and density of the aluminum alloy profiles used. Key connection nodes use spring-damped elements to simulate the contact stiffness and damping characteristics of bolted connections. Eigenvalue analysis is performed on the finite element model to calculate the first six modal frequencies and corresponding mode shapes. The BlockLanczos method is used for eigenvalue analysis, and the solution frequency range is set to 1 Hz to 200 Hz. The principal component components of the acceleration response time series data are used as input, and the test mode shapes of the nursing bed structure are calculated using the random subspace identification method. The test mode shape vector and the theoretical mode shape vector are substituted into the modal confidence criterion calculation formula for calculation. The modal confidence criterion calculation formula is expressed as:

[0045] in, This represents the first principal component calculated from the acceleration response time series data. Column vector of test mode shapes, The value of is a positive integer, and the maximum value is determined by the stable mode order output by the random subspace identification method; This represents the first result calculated using a finite element model based on the structure of the nursing bed. Column vectors of theoretical mode shapes, The value of is a positive integer from 1 to 6, corresponding to the first six modes obtained by eigenvalue analysis; superscript This represents the transpose operation of a vector. The numerator is the square of the dot product of the test mode shape vector and the theoretical mode shape vector, and the denominator is the product of the self-dot product of the test mode shape vector and the self-dot product of the theoretical mode shape vector. The ratio of the numerator to the denominator is the result of the modal confidence criterion calculation. , The value of is between 0 and 1, with a value closer to 1 indicating higher consistency between the two mode shapes. The principal component components of the displacement response time series data and the principal component components of the force response time series data are also calculated using the same random subspace identification method and modal confidence criterion formula, respectively, and compared with the theoretical mode shapes to obtain the calculated results of the modal confidence criterion for the displacement response and the modal confidence criterion for the force response.

[0046] When marking the locations of modes whose modal confidence scores fall below a preset threshold as potential damage areas, the preset threshold is set to 0.85. This threshold is based on a common criterion used in structural dynamics to evaluate modal consistency; a score below 0.85 indicates a significant difference between the test and theoretical modal shapes. The modal confidence score calculation results for all test and theoretical modal orders are iterated. For each result below 0.85, the node with the largest absolute value of its component in the modal shape vector corresponding to that test modal order is mapped from its physical coordinates in the finite element model to its corresponding location on the nursing bed structure. A circular mark is then made at this location. All marked locations constitute a set of potential damage areas.

[0047] The number and distribution density of potential damage areas are statistically analyzed. When using these numbers and density as the results of the seismic reliability test of the nursing bed structure, the number of potential damage areas is the total number of all marked potential damage areas. The distribution density is calculated by dividing the number of potential damage areas by the total number of critical connection nodes in the nursing bed structure. Critical connection nodes include the union of bolted connections, welded connections, and hinged connections on the nursing bed structure; the total number is pre-calculated from the assembly drawings of the nursing bed structure. The number and distribution density of potential damage areas are used as two independent numerical indicators to constitute the output of the seismic reliability test results for the nursing bed structure.

[0048] See Figure 6 In the figure, the horizontal axis represents time, ranging from 0 to 10 seconds, and the vertical axis represents acceleration, with the unit being g. The legend shows three acceleration response time-series curves, corresponding to the lateral acceleration response time-series data (blue solid line), longitudinal acceleration response time-series data (red dashed line), and vertical acceleration response time-series data (green dotted line) of the nursing bed structure under multi-axis synchronous seismic excitation in Example 4, respectively.

[0049] As shown in the figure, the acceleration responses in all three directions exhibit periodic oscillations over time, with the amplitude showing a certain trend. Specifically, the amplitude of the longitudinal acceleration response time series data is generally greater than that in the lateral and vertical directions, with a maximum peak value close to 0.25g, indicating a stronger vibration response in the longitudinal direction. The amplitude of the lateral acceleration response time series data is the second largest, with a peak value close to 0.2g, and the oscillation waveform is relatively regular. The amplitude of the vertical acceleration response time series data is smaller, with a peak value of approximately 0.15g, and the waveform shows a certain phase difference and amplitude fluctuation.

[0050] From a temporal perspective, the oscillation frequencies in all three directions gradually increase over time. Within the first 0 to 5 seconds, the oscillation period is relatively long, followed by a gradual increase in frequency, reflecting the frequency sweep characteristics of the applied multi-axis synchronous simulated seismic excitation. The amplitude variation shows a certain degree of attenuation or fluctuation, possibly reflecting the dynamic characteristics and energy dissipation behavior of the nursing bed structure.

[0051] Example 5:

[0052] In practical implementation, when performing multi-window singular spectrum decomposition on acceleration response time-series data, the acceleration response time-series data is denoted as a sequence. ,sequence The total length is denoted as Total length The sampling frequency is determined by multiplying the sampling duration by the sampling frequency, which is 1000 Hz, and the sampling duration is the total duration of multi-axis synchronous seismic excitation loading. The first window length is set. Total length of acceleration response timing data One-third, specifically the value ,in This indicates a floor function, where floor function is performed only if the window length is an integer number of sampling points. Set the second window length. Total length of acceleration response timing data One-sixth, specifically the value Set the length of the third window. Total length of acceleration response timing data One-tenth, specifically the value The basis for setting three different window lengths is that, in singular spectrum analysis theory, the window length determines the number of rows in the trajectory matrix. The larger the window length, the more comprehensive the trend information contained in the trajectory matrix; the smaller the window length, the stronger the ability to capture local short-term abrupt changes. The proportions of one-third, one-sixth, and one-tenth cover multiple analysis scales from long-term trends to short-term transients, and the three proportions are non-integer multiples of each other to avoid spectral aliasing effects between different window lengths.

[0053] Using the length of the first window respectively Second window length and the length of the third window When constructing three different trajectory matrices, for the first window length trajectory matrix The construction method is as follows: set the trajectory matrix The number of rows is , number of columns Depend on Confirm. Transfer acceleration response timing data. Fill the trajectory matrix by column In the trajectory matrix The Middle Line number The elements of the column are ,in The value range is from 1 to integers, The value range is from 1 to An integer. For the second window length... Construct the trajectory matrix in the same way trajectory matrix The number of rows is , number of columns trajectory matrix The Middle Line number The elements of the column are For the third window length Construct the trajectory matrix in the same way trajectory matrix The number of rows is , number of columns trajectory matrix The Middle Line number The elements of the column are The multi-window singular spectral decomposition of displacement response time series data and force response time series data uses the exact same window length setting method and trajectory matrix construction method.

[0054] When performing singular value decomposition on the three trajectory matrices respectively, the trajectory matrices... Perform singular value decomposition, the decomposition form is as follows ,in for A left singular vector matrix of dimension, for A singular value diagonal matrix of dimension 1. for A right singular vector matrix of dimension, with superscript Indicates matrix transpose. Extracts the trajectory matrix. The left singular vectors corresponding to the first three singular values, i.e., the matrix The three column vectors corresponding to the three largest singular values ​​are denoted as the first left singular vector. Second left singular vector and the third left singular vector For the trajectory matrix Perform singular value decomposition and extract the left singular vectors corresponding to the first three singular values, denoted as the first left singular vector. Second left singular vector and the third left singular vector For the trajectory matrix Perform singular value decomposition and extract the left singular vectors corresponding to the first three singular values, denoted as the first left singular vector. Second left singular vector and the third left singular vector The basis for extracting the eigenvectors corresponding to the first three singular values ​​is that, in structural vibration response signals, the first three singular values ​​usually have a cumulative contribution rate of over 95%, containing most of the signal's energy and main modal information. Extracting the first three singular vectors can remove noise interference while preserving the main modal characteristics.

[0055] When reconstructing principal component components by averaging the three eigenvectors, for the first-order eigenvector, the first left singular vector... First left singular vector and the first left singular vector Take the average. Since the lengths of the three left singular vectors are respectively equal to... , and Before averaging, the three left singular vectors are uniformly resampled to a length of [length missing] using linear interpolation. The vectors are resampled using interpolation based on discrete cosine transform. The three resampled vectors of equal length are added element-wise and divided by 3 to obtain the averaged first-order eigenvector. The same averaging method is applied to the second and third-order eigenvectors to obtain the averaged second-order and third-order eigenvectors. The averaged first-order, second-order, and third-order eigenvectors are then reconstructed into time series components using a diagonal averaging method. The reconstruction process of the diagonal averaging method is as follows: for the reconstructed time series at the time index... The value at point, take all values ​​on the diagonal that satisfy the condition. The average value of the elements of the trajectory matrix is ​​used as the reconstructed value. The three reconstructed time series components are added together to obtain the principal component components of the acceleration response time series data. The residual components are obtained by subtracting the values ​​of the principal component components from the acceleration response time series data point by point according to the sampling time.

[0056] When calculating the test mode shapes of the nursing bed structure based on principal component analysis, the principal component components of the acceleration response time series data are used as input data, and a covariance-driven stochastic subspace identification algorithm is employed for modal parameter identification. The covariance-driven stochastic subspace identification algorithm constructs a discrete-time state-space model, with the model order increasing from 2 to 40. For each order, eigenvalue decomposition of the system matrix is ​​calculated, and eigenvalues ​​and eigenvectors are extracted. The stability graph method is used to distinguish between true and false modes. Mode points in the stability graph where the characteristic frequency changes by less than 1% between adjacent orders, the damping ratio changes by less than 5% between adjacent orders, and the modal confidence criterion value between adjacent orders is greater than 0.95 are marked as stable points. The eigenvectors corresponding to the stable points are the test mode shape vectors.

[0057] When performing a dot product operation between the test mode shape and the theoretical mode shape, the formula for calculating the modal confidence criterion is expressed as:

[0058] in, This represents the column vector of test mode shapes calculated using a covariance-driven random subspace identification algorithm based on the principal component components of the acceleration response time series data. The dimension is equal to the number of sensor measurement points arranged on the nursing bed structure. The number of measurement points is 12, which corresponds to the installation positions of 12 triaxial acceleration sensors arranged on the key connection nodes of the nursing bed structure. This represents the column vector of theoretical modal shapes calculated based on the finite element model of the nursing bed structure. The dimension is equal to 12, and the dimension value is the same as the column vector of the test mode shape. The dimension values ​​are the same, corresponding to the mode shape components at the nodes in the finite element model that match the coordinates of the sensor placement location; superscript This represents the transpose operation of a vector.

[0059] The square of the dot product is used as the numerator, where the dot product refers to the column vector of the test mode shapes. Transpose and theoretical mode shape column vector The scalar value obtained by multiplication. The product of the self-dot product of the test mode shape and the self-dot product of the theoretical mode shape is used as the denominator. The self-dot product of the test mode shape is... The self-dot product of the theoretical mode shape is The ratio of the numerator to the denominator is used as the calculation result of the modal confidence criterion. , The value ranges from 0 to 1.

[0060] When merging potential damage areas into connected components based on spatial adjacency, the coordinates of all potential damage areas in the three-dimensional space of the nursing bed structure are extracted from the set of potential damage areas marked in the modal confidence criterion calculation. The criterion for determining spatial adjacency is: if the Euclidean distance between the coordinates of two potential damage areas is less than a preset connectivity threshold, the two potential damage areas are considered spatially adjacent. The preset connectivity threshold is set to 50 mm, based on the fact that the minimum spacing of the key connection nodes in the nursing bed structure is measured to be 80 mm. 50 mm is less than the minimum spacing but greater than the size range of a single connection node, which can merge potential damage areas marked multiple times near the same connection node into a single connected component without incorrectly merging potential damage areas of adjacent independent nodes. All potential damage area coordinates are traversed, and an eight-connected component labeling algorithm is used to cluster spatially adjacent coordinates. Each clustering result forms a connected component, and all connected components constitute a connected component set.

[0061] When counting the number of nodes in each connected component in the set of connected components, a node refers to the number of coordinate points of potential damaged regions contained within the connected component; one coordinate point corresponds to one node. Each connected component in the set is traversed, and the total number of coordinate points within that component is counted as the number of nodes in that component. To calculate the sum of the number of nodes in all connected components, the number of nodes in each connected component in the set is summed one by one; the sum is the number of potential damaged regions.

[0062] The distribution density is obtained by dividing the number of potentially damaged areas by the total number of critical connection nodes in the nursing bed structure. The total number of critical connection nodes in the nursing bed structure is determined by the critical connection node list marked in the 3D assembly model of the nursing bed structure. The critical connection node list includes all bolted connections, welded connections, and hinged connections, and the total number is denoted as [missing information]. , The values ​​were obtained statistically based on the actual design drawings of the nursing bed structure. Distribution density The calculation formula is ,in The number of potential damage areas, The value ranges from 0 to 1. The number and distribution density of potential damage areas are output as the seismic reliability test results of the nursing bed structure. The output format includes numerical values ​​and corresponding evaluation level descriptions.

[0063] See Figure 7 In the figure, the horizontal axis represents time, ranging from 0 to 10 seconds, and the vertical axis represents acceleration, in g. The figure plots three curves: acceleration response time series data (blue solid line), principal component (red dashed line), and residual component (green dotted line). The acceleration response time series data exhibits rapid oscillations with amplitudes between ±0.4g, and its envelope shows a low-frequency sinusoidal trend with a peak period of approximately 1 second, corresponding to the frequency sweep characteristics of the preset simulated seismic excitation. The amplitude of the principal component curve is significantly lower than that of the original time series data, with a maximum amplitude of approximately 0.2g, and it highly matches the envelope contour of the acceleration response time series data, indicating that the principal component effectively extracts the main modal information and trend components of the acceleration signal. The residual component curve exhibits random fluctuations within the range of ±0.25g, displaying high-frequency noise and local non-stationary characteristics, indicating that the residual component contains small perturbations and random noise from the test signal.

[0064] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. An intelligent testing method for simulating the seismic reliability of a nursing bed structure, characterized in that, The method includes: Real-time acquisition of multi-dimensional strain response of nursing bed structure is performed to obtain the first direction strain time sequence, the second direction strain time sequence and the third direction strain time sequence of key connection nodes of nursing bed structure under preset simulated seismic excitation; By traversing the strain time series in the first direction, the strain time series in the second direction, and the strain time series in the third direction, the joint calculation of the energy distribution coefficients in the time-frequency dual domain is performed to determine the structural nonlinear response characteristic factor, the structural energy dissipation characteristic factor, and the structural stiffness degradation characteristic factor. Using the structural nonlinear response characteristic factor, structural energy dissipation characteristic factor, and structural stiffness degradation characteristic factor as search anchors, the parameters of the four-parameter Bouc-Wen hysteresis model are matched in the seismic test parameter space of the nursing bed structure to determine the target test dynamic load spectrum. Based on the target test dynamic load spectrum, the nursing bed structure is subjected to multi-axis synchronous simulated seismic excitation loading according to a preset test condition set to obtain a set of vibration response characteristic values. The set of vibration response characteristic values ​​is then identified using a structural damage identification indicator to obtain the seismic reliability test results of the nursing bed structure.

2. The intelligent testing method for simulating the seismic reliability of a nursing bed structure according to claim 1, characterized in that, The real-time acquisition of multi-dimensional strain response of the nursing bed structure obtains the first-direction strain time sequence, the second-direction strain time sequence, and the third-direction strain time sequence of key connection nodes of the nursing bed structure under a preset simulated seismic excitation, including: Strain gauges were attached to the key connection nodes of the nursing bed structure, and triaxial strain gauge sensors were installed at the key connection nodes. The three sensitive grids of the triaxial strain gauge sensor are respectively aligned with the transverse, longitudinal, and vertical directions of the nursing bed structure; A preset simulated earthquake excitation device is activated to apply a low-frequency sinusoidal sweep excitation to the nursing bed structure; The voltage signals output by the three sensitive gates are recorded synchronously, and the voltage signals are mapped into the first direction strain time sequence, the second direction strain time sequence, and the third direction strain time sequence after differential amplification and analog-to-digital conversion.

3. The intelligent testing method for simulating the seismic reliability of a nursing bed structure according to claim 1, characterized in that, The joint calculation of time-frequency dual-domain energy distribution coefficients by traversing the strain time series in the first direction, the strain time series in the second direction, and the strain time series in the third direction determines the structural nonlinear response characteristic factor, the structural energy dissipation characteristic factor, and the structural stiffness degradation characteristic factor, including: Perform a short-time Fourier transform on the strain time series sequence in the first direction to obtain the time spectrum matrix of the strain time series sequence in the first direction; Extract the amplitude corresponding to the peak frequency of each time window in the time spectrum matrix, and arrange the amplitudes in time order to form an instantaneous frequency-amplitude curve; The nonlinear response characteristic factor of the structure is obtained by calculating the curve integral after performing cubic spline interpolation on the instantaneous frequency-amplitude curve; Repeat the above steps for the structural energy dissipation characteristic factor and the structural stiffness degradation characteristic factor.

4. The intelligent testing method for simulating the seismic reliability of a nursing bed structure according to claim 3, characterized in that, The above steps are repeated for the structural energy dissipation characteristic factor and the structural stiffness degradation characteristic factor, including: Perform the same short-time Fourier transform and interpolation integration operations on the strain time series sequence in the second direction as on the strain time series sequence in the first direction to obtain the energy dissipation characteristic factor of the structure; Perform the same short-time Fourier transform and interpolation integration operations on the third-direction strain time series as on the first-direction strain time series to obtain the structural stiffness degradation characteristic factor; The structural nonlinear response characteristic factor, the structural energy dissipation characteristic factor, and the structural stiffness degradation characteristic factor are normalized so that the numerical range of the three characteristic factors is unified to the interval between 0 and 1.

5. The intelligent testing method for simulating the seismic reliability of a nursing bed structure according to claim 1, characterized in that, The process involves using the structural nonlinear response characteristic factor, structural energy dissipation characteristic factor, and structural stiffness degradation characteristic factor as search anchors, performing four-parameter Bouc-Wen hysteresis model parameter matching in the seismic test parameter space of the nursing bed structure, and determining the target test dynamic load spectrum, including: The structural nonlinear response characteristic factor, structural energy dissipation characteristic factor, and structural stiffness degradation characteristic factor are used as the search coordinate values; Extract a preset number of historical test load spectra that have the smallest spatial distance to the retrieved coordinate values ​​from the seismic test parameter space of the nursing bed structure; Calculate the yield displacement, post-yield stiffness ratio, hysteresis loop shape control parameters, and pinching effect parameters of the four-parameter Bouc-Wen hysteresis model corresponding to each of the historical test load spectra. The four parameters of the historical test load spectrum are weighted and averaged to serve as the initial parameters of the target test dynamic load spectrum. The initial parameters are iteratively corrected using the Runge-Kutta numerical integration method until the deviation between the hysteresis loop area corresponding to the corrected parameters and the structural energy dissipation characteristic factor converges to a preset threshold.

6. The intelligent testing method for simulating the seismic reliability of a nursing bed structure according to claim 1, characterized in that, The process involves performing multi-axis synchronous simulated seismic excitation loading on the nursing bed structure based on the target test dynamic load spectrum according to a preset test condition set, obtaining a set of vibration response characteristic values, and using a structural damage identification indicator to identify the set of vibration response characteristic values ​​to obtain the seismic reliability test results of the nursing bed structure, including: The target test dynamic load spectrum is decomposed into transverse load components, longitudinal load components, and vertical load components; The lateral load component, longitudinal load component, and vertical load component are respectively input to the servo valve control module of the electro-hydraulic servo actuator; The electro-hydraulic servo actuator is controlled to apply multi-axis synchronous seismic excitation to the nursing bed structure according to the lateral load component, longitudinal load component and vertical load component; Record the acceleration response time series data, displacement response time series data, and force response time series data of the nursing bed structure under the multi-axis synchronous seismic excitation, and combine the acceleration response time series data, displacement response time series data, and force response time series data into the vibration response characteristic value set.

7. The intelligent testing method for simulating the seismic reliability of a nursing bed structure according to claim 6, characterized in that, The method of using a structural damage identification indicator to identify the set of vibration response feature values ​​to obtain the seismic reliability test results of the nursing bed structure includes: Multi-window singular spectrum decomposition was performed on the acceleration response time series data, displacement response time series data and force response time series data respectively to obtain the principal component components and residual components of each response data. The modal confidence criterion is used to calculate the principal component components and the theoretical mode shapes obtained from the finite element model based on the nursing bed structure. The locations of modes whose confidence scores are below a preset threshold in the modality confidence criterion calculation are marked as potential damage areas. The number and distribution density of the potential damage areas are counted, and the number and distribution density are used as the seismic reliability test results of the nursing bed structure.

8. The intelligent testing method for simulating the seismic reliability of a nursing bed structure according to claim 7, characterized in that, The step of performing multi-window singular spectrum decomposition on the acceleration response time series data, displacement response time series data, and force response time series data to obtain the principal component components and residual components of each response data includes: When performing multi-window singular spectrum decomposition on the acceleration response time series data, the length of the first window is set to one-third of the total length of the acceleration response time series data, the length of the second window is set to one-sixth of the total length of the acceleration response time series data, and the length of the third window is set to one-tenth of the total length of the acceleration response time series data. Three different trajectory matrices are constructed using the first window length, the second window length, and the third window length, respectively. Singular value decomposition is performed on the three trajectory matrices respectively, and the eigenvectors corresponding to the first three singular values ​​of each trajectory matrix are extracted. The three eigenvectors are averaged and reconstructed into the principal component components. The residual component is obtained by subtracting the principal component from the acceleration response time series data.

9. The intelligent testing method for simulating the seismic reliability of a nursing bed structure according to claim 8, characterized in that, The step of calculating the modal confidence criterion by combining the principal component components with the theoretical mode shapes calculated based on the finite element model of the nursing bed structure includes: The test mode shape of the nursing bed structure is calculated based on the principal component components, and the test mode shape is multiplied by the theoretical mode shape to obtain the multiplication result. The square of the dot product result is used as the numerator, and the product of the self-dot product result of the test mode shape and the self-dot product result of the theoretical mode shape is used as the denominator. The ratio of the numerator to the denominator is used as the calculation result of the modal confidence criterion.

10. The intelligent testing method for simulating the seismic reliability of a nursing bed structure according to claim 9, characterized in that, The method of statistically analyzing the number and distribution density of the potential damage areas, and using these numbers and distribution densities as the seismic reliability test results of the nursing bed structure, includes: The potential damaged regions are merged into connected components according to their spatial adjacency to obtain a set of connected components. The number of nodes contained in each connected component in the set of connected components is counted, and the sum of the number of nodes in all connected components is calculated as the number of potential damage regions. The distribution density is obtained by dividing the number of potential damage areas by the total number of critical connection nodes in the nursing bed structure. The number and distribution density of the potential damage areas are used as the seismic reliability test results of the nursing bed structure.

Images