Landslide dynamic response judgment method and device based on centrifugal machine vibration table and medium

Abstract

The invention provides a landslide dynamic response judgment method and device based on a centrifugal machine vibration table and a medium, and the method comprises the steps: obtaining the landslide position and other information of a target landslide, building a profile generalization physical model of the target landslide in combination with a preset similarity ratio, setting at least two groups of measurement points with different coordinates in the profile generalization physical model, and determining the landslide dynamic response of the target landslide; and measuring point acceleration sensors in one-to-one correspondence with the measuring points are arranged. According to historical earthquake acceleration time history information, a physical model is generalized by using a vibration section of a centrifuge vibration table, and measurement point model domain acceleration response data is acquired through a measurement point acceleration sensor. And preprocessing the measurement point model domain acceleration response data according to a preset similarity ratio to obtain measurement point prototype domain acceleration response data, and extracting target seismic oscillation characteristic parameters from the measurement point prototype domain acceleration response data. The landslide dynamic response discrimination method based on the centrifuge vibration table provided by the invention can improve the accuracy of quantitatively discriminating the landslide dynamic response.

Description

Technical Field

[0001] The embodiments of the present invention relate to the interdisciplinary field of geological disaster prevention and earthquake engineering, and in particular to a method, device and medium for judging landslide dynamic response based on a centrifuge shaking table. Background Technology

[0002] Landslides are common secondary disasters in seismically active areas, characterized by complex structures, diverse causes, and significant dynamic amplification effects under seismic action. Currently, research on the dynamic response of landslides largely relies on numerical simulations or field monitoring data, which suffers from insufficient experimental verification, low data resolution, and difficulty in reflecting the true response characteristics within the landslide mass.

[0003] Centrifuge shaking table tests can realistically reproduce the effects of gravity fields under controlled conditions, providing an effective means for studying the seismic response mechanism of landslides. However, there is still a lack of effective technical methods for accurately and quantitatively determining the dynamic response of landslides.

[0004] Therefore, improving the accuracy of quantitatively identifying the dynamic response of landslides has become an urgent technical problem to be solved. Summary of the Invention

[0005] In a first aspect, embodiments of the present invention provide a method for determining the dynamic response of a landslide based on a centrifuge vibration table, comprising: Obtain information on the target landslide, including landslide location, landslide scale, landslide boundary, landslide structure layering and physical properties, and historical earthquake acceleration time history. Based on the landslide location information, landslide scale information, landslide boundary information, landslide structure layering and physical property information, a generalized physical model of the target landslide is established. The model scale ratio of the generalized physical model is a preset similarity ratio. At least two sets of measuring points with different coordinates are set in the generalized physical model, and measuring point acceleration sensors corresponding to the measuring points are set one by one. Based on the historical earthquake acceleration time history information, a centrifuge shaking table is used to simulate the vibration of the profile generalized physical model, and the acceleration response data of the measuring point model domain is obtained through the measuring point acceleration sensor. The measuring point model domain acceleration response data includes measuring point model domain acceleration data and timestamp data. Preprocess the acceleration response data of the measurement point model domain according to the preset similarity ratio to obtain the acceleration response data of the measurement point prototype domain. The acceleration response data of the measurement point prototype domain includes the acceleration data of the measurement point prototype domain and the corresponding timestamp data. Target ground motion characteristic parameters are extracted from the acceleration response data of the prototype domain of the measuring point for landslide dynamic response discrimination.

[0006] Optionally, acquiring the acceleration response data of the measurement point model domain through the measurement point acceleration sensor includes: The raw output data of the acceleration response in the model domain of the measuring point is obtained through the measuring point acceleration sensor; The script converts the original output data of the acceleration response of the measurement point model domain into a CSV file and removes the descriptive header text to obtain the acceleration response data of the measurement point model domain.

[0007] Optionally, the step of preprocessing the acceleration response data of the measurement point model domain according to the preset similarity ratio to obtain the acceleration response data of the measurement point prototype domain, wherein the acceleration response data of the measurement point prototype domain includes acceleration data of the measurement point prototype domain and corresponding timestamp data, including: Based on the preset similarity ratio, the acceleration response data of the measurement point model domain is converted into acceleration response data of the measurement point prototype domain to be preprocessed. Robust baseline correction is performed on the preprocessed data of the prototype domain acceleration response of the measurement point to obtain robust time-series data of the prototype domain acceleration response of the measurement point. The robust time-series data of the prototype domain acceleration response of the measurement point includes robust acceleration data of the prototype domain acceleration and corresponding timestamp data. The robust time-series data of the acceleration response of the prototype domain at the measurement point is obtained by Fourier transform, and the robust frequency-domain data of the acceleration response of the prototype domain at the measurement point is obtained by bandpass filtering, and the bandpass-domain data of the acceleration response of the prototype domain at the measurement point is obtained by inverse Fourier transform.

[0008] Optionally, the robust baseline correction of the preprocessed data of the prototype domain acceleration response at the measurement point to obtain robust time-series data of the prototype domain acceleration response at the measurement point includes: The acceleration response data to be preprocessed in the prototype domain of the measuring point is subjected to mean-removal processing to obtain the mean-removed acceleration data in the prototype domain of the measuring point. Linear detrending is performed on the mean-de-trend data of the acceleration in the prototype domain of the measurement point to obtain robust time-series data of the acceleration response in the prototype domain of the measurement point.

[0009] Optionally, the bandpass filtering of the robust frequency domain data of the prototype acceleration response at the measurement point to obtain the bandpass frequency domain data of the prototype acceleration response at the measurement point includes: The robust frequency domain data of the prototype acceleration response at the measurement point is filtered by a Butterworth bandpass filter to obtain the bandpass frequency domain data of the prototype acceleration response at the measurement point.

[0010] Optionally, the target ground motion characteristic parameters include the ground motion amplification factor; The extraction of target ground motion characteristic parameters from the prototype domain acceleration response data of the measuring point includes: Based on the start and end times entered in the start and end time input boxes in the GUI interface, extract the selected dynamic time interval; The maximum absolute value of the selected dynamic time interval is calculated as the peak acceleration; The peak acceleration of one measuring point is selected from the at least two sets of measuring points as the reference peak acceleration. The ratio of the peak acceleration of the measuring point in the at least two sets of measuring points to the reference peak acceleration is calculated to obtain the seismic amplification factor of all the measuring points.

[0011] Optionally, it also includes: Based on the sliding body boundary information, the coordinates of the measuring points, and the target ground motion characteristic parameters, a uniform interpolation grid map is generated and interpolated.

[0012] Optionally, it also includes: By setting contour lines and color gradients on the uniform interpolation grid, a two-dimensional distribution cloud map of the target seismic characteristic parameters is obtained.

[0013] Secondly, embodiments of the present invention also provide a landslide dynamic response discrimination device based on a centrifuge vibration table, comprising: The information acquisition unit is suitable for acquiring information on the target landslide, including landslide location, landslide scale, landslide boundary, landslide structure layering and physical properties, and historical earthquake acceleration time history. The modeling unit is adapted to establish a profile generalized physical model of the target landslide based on the landslide location information, landslide scale information, landslide boundary information, landslide structure layering and physical property composition information, and a preset similarity ratio. At least two sets of measuring points with different coordinates are set in the profile generalized physical model, and measuring point acceleration sensors corresponding to the measuring points are set one by one. The measuring point model domain acceleration response unit is adapted to simulate the vibration of the profile generalized physical model using a centrifuge shaking table based on the historical earthquake acceleration time history information, and to acquire measuring point model domain acceleration response data through the measuring point acceleration sensor. The measuring point model domain acceleration response data includes measuring point model domain acceleration data and timestamp data. The preprocessing unit is adapted to preprocess the acceleration response data of the measurement point model domain according to the preset similarity ratio to obtain the acceleration response data of the measurement point prototype domain, wherein the acceleration response data of the measurement point prototype domain includes the acceleration data of the measurement point prototype domain and the corresponding timestamp data. The target ground motion feature extraction unit is adapted to extract target ground motion feature parameters from the acceleration response data of the prototype domain of the measuring point.

[0014] Thirdly, embodiments of the present invention also provide a storage medium storing a program for landslide dynamic response discrimination based on a centrifuge vibration table, so as to implement the landslide dynamic response discrimination method based on a centrifuge vibration table as described in the first aspect above.

[0015] The landslide dynamic response discrimination method based on a centrifuge shaking table provided by this invention obtains the landslide location information, landslide scale information, landslide boundary information, landslide structural layering and physical property information, and historical earthquake acceleration time history information of the target landslide; establishes a profile generalized physical model of the target landslide based on the landslide location information, landslide scale information, landslide boundary information, landslide structural layering and physical property information, the model scaling ratio of the profile generalized physical model is a preset similarity ratio, and sets at least two sets of measuring points with different coordinates in the profile generalized physical model, and sets a measuring point acceleration sensor corresponding to each measuring point. Based on historical earthquake acceleration time history information, a generalized physical model is constructed by simulating the seismic profile using a centrifuge shaking table. Accelerometer data of the measured point model domain is acquired using accelerometer sensors. This measured point model domain acceleration response data includes both measured point model domain acceleration data and timestamp data. The measured point model domain acceleration response data is preprocessed according to a preset similarity ratio to obtain measured point prototype domain acceleration response data, which includes both measured point prototype domain acceleration data and corresponding timestamp data. Target ground motion characteristic parameters are extracted from the measured point prototype domain acceleration response data for landslide dynamic response discrimination. Thus, the method provided by this embodiment of the invention can combine landslide data, historical earthquake acceleration time history information, and experimental data to systematically construct a quantitative discrimination method for the dynamic response of landslides in fault zones, thereby improving the accuracy of quantitative landslide dynamic response discrimination. Attached Figure Description

[0016] Figure 1 This is a schematic flowchart of a landslide dynamic response discrimination method based on a centrifuge vibration table provided in an embodiment of the present invention; Figure 2 This is a landslide geological profile diagram provided in the embodiment of the present invention for the landslide dynamic response discrimination method based on a centrifuge vibration table; Figure 3 This is a profile generalized physical model of the landslide dynamic response discrimination method based on a centrifuge vibration table provided in this embodiment of the invention; Figure 4 This is a schematic diagram of a cross-sectional generalized physical model of the landslide dynamic response discrimination method based on a centrifuge vibration table provided in an embodiment of the present invention; Figure 5 This is a schematic diagram of the measurement point layout for the landslide dynamic response discrimination method based on a centrifuge vibration table provided in an embodiment of the present invention; Figure 6This is a time-frequency verification diagram of the landslide dynamic response discrimination method based on a centrifuge vibration table provided in the embodiments of the present invention; Figure 7 This is a two-dimensional distribution cloud of the landslide dynamic response discrimination method based on a centrifuge shaking table provided in this embodiment of the invention. Figure 1 Schematic diagram. Detailed Implementation

[0017] 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, and 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.

[0018] Please refer to Figures 1-3 , Figure 1 This is a flowchart illustrating a landslide dynamic response discrimination method based on a centrifuge vibration table provided in an embodiment of the present invention. Figure 2 This is a landslide geological profile diagram provided in the embodiment of the landslide dynamic response discrimination method based on a centrifuge shaking table according to the present invention. Figure 3 This is a profile generalized physical model of the landslide dynamic response discrimination method based on a centrifuge vibration table provided in the embodiments of the present invention.

[0019] As shown in the figure, the method provided by the embodiment of the present invention includes the following steps: In step S1, the following information is obtained: landslide location, landslide scale, landslide boundary, landslide structure layering and physical property composition, and historical earthquake acceleration time history.

[0020] Information on landslide location, scale, boundary conditions, structural layering, and physical properties can be obtained through unmanned aerial vehicle (UAV) photogrammetry and on-site geological surveys, utilizing remote sensing data and field investigations. Historical earthquake acceleration time-history information can be obtained by retrieving past earthquake data.

[0021] In step S2, a generalized physical model of the target landslide is established based on the landslide location information, landslide scale information, landslide boundary information, landslide structure layering and physical property information, and a preset similarity ratio. At least two sets of measuring points with different coordinates are set in the generalized physical model, and a measuring point acceleration sensor corresponding to each measuring point is set.

[0022] The profile generalized physical model is a generalized physical model of a representative landslide profile. The preset similarity ratio, based on the similarity theorem and centrifugal acceleration, ensures that the profile generalized physical model and the landslide simultaneously meet geometric, dynamic, and temporal similarity criteria. The specific value can be preset. In some embodiments, the preset similarity ratio can be 40, meaning the ratio of the landslide to the generalized physical model is 40.

[0023] The landslide location information is a geographical location, such as Zhang County, Gansu Province. The landslide boundary information is the boundary of the landslide area. In a specific embodiment, the landslide boundary information can be recorded by establishing a coordinate system and using coordinates.

[0024] Please refer to the information on landslide structure stratification and physical properties. Figure 2 ,like Figure 2 As shown, layer 4 is sandstone bedrock D3-C1, and layer 5 is residual colluvial layer Q4. del Layer 6 is landslide deposit Q4 dl Fault 1 is the Zhangxian Fault, slip surface 2 is the main slip surface, slip surface 3 is the secondary slip surface, and layer 5 is the residual colluvial layer Q4. del and the landslide deposits Q4 of layer 6 dl The main constituent material is loess, so it is considered as a single medium during modeling. Landslide scale information includes the thickness, size, and other scale information of each structure of the landslide.

[0025] Please refer to the simplified physical model of the cross-section. Figure 3 As shown in the figure, in some embodiments, barite powder, quartz sand, gypsum, glycerin, and water can be used as similar materials for the landslide layer 4 sandstone bedrock D3-C1 and layer 5 residual colluvial layer Q4. del and the landslide deposits Q4 of layer 6 dl Three media, fault 1, and the following mass percentage ratios were designed for each: Layer 4 Sandstone Bedrock D3-C1: Quartz Sand 48% : Barite Powder 34.5% : Gypsum 10% : Glycerin 2% : Water 5.5%; Layer 5 Residual Colour Layer Q4 del and the landslide deposits Q4 of layer 6 dl The composition of the first layer is as follows: Quartz sand 52% : Barite powder 34.5% : Gypsum 6% : Glycerin 2% : Water 5.5%; Fault 1: Quartz sand 53% : Barite powder 35% : Gypsum 2% : Glycerin 2% : Water 8%. The model is then constructed using a layered masonry method within the mold box.

[0026] Please refer to Figure 4 , Figure 4This is a schematic diagram of a profile generalized physical model for landslide dynamic response discrimination based on a centrifuge vibration table, provided in an embodiment of the present invention. In some embodiments, before establishing the profile generalized physical model, a generalized model of the profile generalized physical model can be established based on the landslide location information, landslide scale information, landslide boundary information, landslide structural layering and physical property information, and a preset similarity ratio. In this model, generalized layer 4 is a sandstone bedrock layer, generalized layer 5 is a residual slope deposit and landslide deposit layer, and fault 1 is the Zhangxian Fault. Then, a model is established based on the generalized model. Figure 3 The cross-section shown is a generalized physical model.

[0027] In some embodiments, to comprehensively capture the dynamic response characteristics and failure mechanisms of landslides under seismic loading, coordinates of measuring points are set in the profile-generalized physical model. At least one measuring point is set at the bedrock layer at the bottom of the model or on the shaking table surface to collect the raw time history data of the input seismic waves, which serves as the reference denominator for subsequent calculation of the seismic motion amplification factor. At least three measuring points are set along the elevation direction of the main landslide profile, such as the toe, middle, and crest of the slope, to capture the "bottom-up amplification" effect of seismic waves along the slope height. Local measuring points are densely deployed at key structural locations such as slip zones, fault zones, or interfaces between soft and hard rock layers to capture stress concentration and discontinuous deformation signals during shear failure.

[0028] In a preferred embodiment, 15 measuring points can be set up, with one single-axis accelerometer at each measuring point. Please refer to [reference needed] for details. Figure 5 , Figure 5 This is a schematic diagram of the measurement point layout for the landslide dynamic response discrimination method based on a centrifuge shaking table provided in this embodiment of the invention. As shown in the figure, the bedrock reference group measurement points, including A3, A5, and A9, are arranged in the bedrock at the bottom of the model to record the seismic motion input from the bedrock; the fault zone densified group measurement points, including A6, A7, A8, and A4, are densely arranged along both sides of the hanging wall and footwall of the fault zone to monitor the blocking and amplification effect of fault slippage on wave propagation; the landslide response group measurement points, including A1 and A10 at the top of the slope, A11 and A14 in the middle of the slope, and A13 and A15 at the toe of the slope, are arranged at different elevations inside and on the surface of the landslide body to form a monitoring grid covering the entire slope surface, used to generate a two-dimensional distribution cloud map. In this way, it can be ensured that the collected data covers the complete spatial characteristics of the landslide dynamic response, including both the overall amplification law and the local damage details.

[0029] In step S3, based on the historical earthquake acceleration time history information, a centrifuge shaking table is used to simulate the vibration of the profile generalized physical model, and the acceleration response data of the measuring point model domain is obtained through the measuring point acceleration sensor. The measuring point model domain acceleration response data includes measuring point model domain acceleration data and timestamp data.

[0030] The sensors are mainly deployed in the slip zone, slip body, and bedrock of the profile generalized physical model.

[0031] In some embodiments, to address the issues of inconsistent raw data file formats output by centrifugal vibration table test sensors and the presence of a large amount of redundant header information in the files, and to ensure that subsequent algorithms can directly read standardized numerical data, the step of acquiring the acceleration response data of the measurement point model domain through the measurement point acceleration sensor includes: In step S31, the raw output data of the acceleration response in the measurement point model domain is acquired through the measurement point acceleration sensor; In step S32, the original output data of the acceleration response of the measurement point model domain is converted into a CSV file using a script, and the descriptive text header is removed to obtain the acceleration response data of the measurement point model domain.

[0032] Specifically, a self-written Python script can be used for automated batch processing. By calling the Microsoft Excel application programming interface (COM interface, win32com.client), Excel runs silently in the background. This program can automatically identify all .txt files in the current folder and, leveraging Excel's powerful text parsing capabilities, batch save each .txt file as a standard .csv file. This process requires no manual intervention, achieving automated conversion from raw heterogeneous text data to structured CSV data.

[0033] Open the .csv files generated in the previous step again via the Excel COM interface. Precisely delete the first N rows of each file. These N rows typically contain descriptive text information such as instrument model, sampling parameters, and test conditions, which are redundant for numerical analysis. In this way, all .csv files are cleaned into a standard table format containing only valid time-history numerical data, eliminating the tediousness and error-proneness of manual cleanup.

[0034] In step S4, the acceleration response data of the measurement point model domain is preprocessed according to the preset similarity ratio to obtain the acceleration response data of the measurement point prototype domain. The acceleration response data of the measurement point prototype domain includes the acceleration data of the measurement point prototype domain and the corresponding timestamp data.

[0035] The prototype domain refers to the original object to be studied and modeled; in this invention, it is the target landslide. The model domain is an abstract and simplified system constructed for studying the prototype domain; in this invention, it is a generalized physical model of the target landslide profile. Preprocessing the acceleration response data of the measurement point model domain according to a preset similarity ratio yields the acceleration response data of the measurement point prototype domain. This involves multiplying the time in the acceleration data of the measurement point model domain and the corresponding timestamp data by a preset similarity ratio, dividing the acceleration by the preset similarity ratio, and dividing the sampling frequency by the preset similarity ratio to obtain the time, acceleration, and sampling frequency in the acceleration data of the measurement point prototype domain and the corresponding timestamp data.

[0036] In this way, unifying the data to the prototype domain can eliminate scaling effects. This transformation unifies all subsequent signal processing within the prototype domain, ensuring the accuracy of physical meaning and the consistency of results. Specifically, the time prototype is the model time multiplied by the similarity ratio, and the acceleration response data of the measurement point in the model domain is obtained by dividing the acceleration data of the measurement point in the similarity ratio.

[0037] In some embodiments, in order to eliminate scaling effects, instrument zero-point drift, low-frequency trends, and high-frequency noise while preprocessing to obtain the prototype domain acceleration data and corresponding timestamp data of the measurement point, in step S4, the measurement point model domain acceleration response data is preprocessed according to the preset similarity ratio to obtain the prototype domain acceleration response data of the measurement point. The prototype domain acceleration response data of the measurement point includes the prototype domain acceleration data and the corresponding timestamp data, including: In step S41, the acceleration response data of the measurement point model domain is converted into acceleration response data of the measurement point prototype domain to be preprocessed according to the preset similarity ratio.

[0038] In step S42, the robust baseline correction of the preprocessed data of the prototype domain acceleration response of the measurement point is used to obtain robust time-series data of the prototype domain acceleration response of the measurement point. The robust time-series data of the prototype domain acceleration response of the measurement point includes robust acceleration data of the prototype domain and corresponding timestamp data.

[0039] Robust baseline correction refers to a data preprocessing method that uses robust algorithms to fit the true baseline of data containing noise, outliers, or interference. Robust time-series data of the prototype domain acceleration response refers to the data obtained after robust baseline correction of the prototype domain acceleration response data to be preprocessed. Robust baseline correction can eliminate zero-point drift and low-frequency linear trends in the signal, ensuring baseline stability. In some embodiments, the prototype domain acceleration response data to be preprocessed may first undergo mean-reduction processing to obtain mean-reduction acceleration data, and then linear detrending may be performed on the mean-reduction acceleration data to obtain robust time-series acceleration response data. Specifically, one approach is to call the `detrend(acc0, type='linear')` function from the Python scientific computing library `scipy.signal` to perform linear detrending. This function internally calculates the best-fit line by minimizing the sum of squared residuals. In other embodiments, a polynomial fitting algorithm can be used, performing the fitting operation through the least squares method, and then subtracting the fitted trend term. Specifically, first, a linear trend equation `y = kt + b` is constructed, where `k` is the slope, `b` is the intercept, and `t` is the time series index. Then, the principle of least squares is used to solve for the equation that minimizes the sum of squared errors. The minimum parameters k and b are then used to subtract the trend term from the original sequence to obtain the final corrected acceleration sequence. .

[0040] This approach not only effectively removes the DC component and linear drift in the acceleration time history, ensuring the accuracy of the integral displacement, but also, through the dual-path design of the detrend function call and the polynomial fitting algorithm, ensures the robustness, cross-platform consistency, and high accuracy of the baseline correction, while improving the stability of the algorithm in engineering applications. Ensuring robustness means that even in rudimentary environments such as field laptops or embedded industrial PCs, where direct calls to a single function could easily lead to program crashes due to incomplete or incompatible scientific computing libraries, the dual-path design ensures that the program can still complete complex baseline correction tasks without interruption, even in extremely rudimentary software environments. Ensuring cross-platform consistency means that the underlying mathematical core of both paths is completely consistent, both based on the principle of minimizing the sum of squared errors. This ensures that the algorithm results are not dependent on a specific operating system or a specific third-party library version. Regardless of the platform on which it runs, the detrended waveform output is uniquely determined, avoiding data "black box" bias caused by differences in underlying library implementations. High accuracy means that, compared to the detrending method of "connecting the first and last data points," which is easily affected by noise from the first and last data points and leads to incorrect overall baseline correction, the dual paths of this invention are based on least squares fitting of full-time data. This method considers every point in the time series. The algorithm finds a straight line that minimizes the sum of the deviations of all global points, which can automatically "smooth out" local random noise interference. Even if there are severe oscillations at the beginning and end of the data, the fitted trend line can still accurately reflect the true zero-point drift direction, thus achieving higher physical accuracy than simple geometric methods.

[0041] In step S43, the robust time-series data of the prototype acceleration response of the measurement point is Fourier transformed to obtain robust frequency-domain data of the prototype acceleration response of the measurement point, the robust frequency-domain data of the prototype acceleration response of the measurement point is bandpass filtered to obtain bandpass frequency-domain data of the prototype acceleration response of the measurement point, and the bandpass frequency-domain data of the prototype acceleration response of the measurement point is inverse Fourier transformed to obtain acceleration response data of the prototype acceleration response of the measurement point.

[0042] To filter out high-frequency noise and unwanted low-frequency components in the signal, in some embodiments, the robust frequency domain data of the prototype domain acceleration response can be filtered using a Butterworth bandpass filter to obtain the prototype domain acceleration response bandpass frequency domain data. Specifically, a fourth-order Butterworth bandpass filter is preferred. The filter performs filtering between the lower and upper limits of the prototype domain frequency. The `filtfilt` function uses zero-phase filtering to avoid signal phase distortion. If the scipy library is unavailable or an error occurs during Butterworth filtering, the program will automatically fall back to the frequency domain windowing method based on the Fast Fourier Transform. This method first performs a Fast Fourier Transform (FFT) on the robust time-series acceleration response data of the measurement point prototype domain, which is a time-series signal, to transform it to the frequency domain, obtaining the robust frequency domain acceleration response data of the measurement point prototype domain. Then, all frequency components in the robust frequency domain acceleration response data of the measurement point prototype domain that fall outside the preset prototype domain frequency band are directly set to zero, thus obtaining the bandpass frequency domain acceleration response data of the measurement point prototype domain. Finally, an Inverse Fast Fourier Transform (IFFT) is performed on the bandpass frequency domain acceleration response data of the measurement point prototype domain to transform the signal back to the time domain, obtaining the acceleration response data of the measurement point prototype domain. In this way, this dual-path filtering mechanism ensures stable and accurate bandpass filtering results even under specific software environments or data conditions.

[0043] In some embodiments, all preprocessed prototype domain acceleration response data can be output to a specified directory with CSV column names and structure similar to the original files, facilitating direct reading by subsequent analysis tools. This ensures seamless data processing and standardization of the final data.

[0044] In a preferred embodiment, a Butterworth filter is used in the preprocessing stage because it has the flattest frequency response in the passband, effectively avoiding signal distortion. The order is set to 4 (FILTER_ORDER = 4). A lower order leads to poor filtering, while a higher order introduces phase delay and computational instability; 4 is the optimal balance. The cutoff frequency is set to the prototype domain of 0.1 Hz to 15.0 Hz (F_LOW_PROTO = 0.1, F_HIGH_PROTO = 15.0). This band covers the majority of the dominant frequency components of destructive seismic waves while effectively filtering out sensor temperature drift (<0.1 Hz) in the low-frequency band and electromagnetic noise (>15 Hz) in the high-frequency band. The interpolation strategy uses radial basis function (RBF) interpolation. The kernel type is set to "Linear," but in another preferred embodiment, it can also be set to "Multiquadric." This allows the linear kernel, compared to the Gaussian kernel, to better maintain the overall trend of landslide response with elevation variation when dealing with sparse measurement points (e.g., only 15 sensors), avoiding non-physical local oscillations (Runge phenomenon). The code defaults to "rbf-linear."

[0045] Please refer to Figure 6 , Figure 6 This is a time-frequency verification diagram of the landslide dynamic response discrimination method based on a centrifuge vibration table provided in this embodiment of the invention. In some embodiments, to verify the effectiveness of the preprocessing step, this invention also performs verification through joint time-frequency domain verification. Specifically, time-domain verification compares the original waveform in the preprocessed data of the prototype domain acceleration response of the measurement point with the preprocessed waveform in the preprocessed prototype domain acceleration response data of the measurement point. If the baseline of the preprocessed waveform returns to the zero axis, i.e., there is no drift, and the spiky high-frequency random noise is smoothed, and the occurrence time of the main shock peak (PGA) does not shift, then the time-domain verification passes. Frequency-domain verification performs a Fast Fourier Transform (FFT) on the preprocessed waveform. If the amplitude in the high-frequency band above 15Hz should be attenuated to near zero, and the low-frequency components below 0.1Hz are effectively suppressed, then the filter parameter settings are reasonable, and the frequency-domain verification passes.

[0046] In step S5, target ground motion characteristic parameters are extracted from the acceleration response data of the prototype domain of the measuring point for landslide dynamic response discrimination.

[0047] The target ground motion characteristic parameter can be the ground motion amplification factor.

[0048] In some embodiments, step S5, extracting target ground motion characteristic parameters from the prototype domain acceleration response data of the measuring point, includes: In step S51, the selected dynamic time interval is extracted based on the start time and end time entered in the start time and end time input boxes in the GUI interface.

[0049] In step S52, the maximum absolute value of the selected dynamic time interval is calculated as the peak acceleration.

[0050] In step S53, the peak acceleration of one measuring point is selected from the at least two sets of measuring points as the reference peak acceleration, and the ratio of the peak acceleration of the measuring point to the reference peak acceleration is calculated for each of the at least two sets of measuring points to obtain the seismic amplification factor for all the measuring points.

[0051] In some embodiments, the peak acceleration of the measuring point closest to the seismic source can be selected as the reference peak acceleration.

[0052] In this way, a graphical user interface (GUI) program is used to integrate data loading, visualization, feature calculation, and data export functions. The GUI interface provides "start time" and "end time" input boxes, allowing users to dynamically extract time-series data as needed. All subsequent PGA and FFT calculations will be based on the user-defined extraction interval.

[0053] In some embodiments, the benchmark can be the peak acceleration of a sensor measuring point located in the bedrock layer at the bottom of the model. For each other measuring point i inside the landslide model, the seismic amplification factor Ki is defined as the ratio of the peak acceleration of that measuring point to the benchmark peak acceleration.

[0054] As can be seen, the landslide dynamic response discrimination method based on centrifuge shaking table provided by the present invention obtains the landslide location information, landslide scale information, landslide boundary information, landslide structural layering and physical property information, and historical earthquake acceleration time history information of the target landslide; establishes a profile generalized physical model of the target landslide based on the landslide location information, landslide scale information, landslide boundary information, landslide structural layering and physical property information, the model scaling ratio of the profile generalized physical model is a preset similarity ratio, and sets at least two sets of measuring points with different coordinates in the profile generalized physical model, and sets a measuring point acceleration sensor corresponding to each measuring point. Based on historical earthquake acceleration time history information, a generalized physical model is constructed by simulating the seismic profile using a centrifuge shaking table. Accelerometer data of the measured point model domain is acquired using accelerometer sensors. This measured point model domain acceleration response data includes both measured point model domain acceleration data and timestamp data. The measured point model domain acceleration response data is preprocessed according to a preset similarity ratio to obtain measured point prototype domain acceleration response data, which includes both measured point prototype domain acceleration data and corresponding timestamp data. Target ground motion characteristic parameters are extracted from the measured point prototype domain acceleration response data for landslide dynamic response discrimination. Thus, the method provided by this embodiment of the invention can combine landslide data, historical earthquake acceleration time history information, and experimental data to systematically construct a quantitative discrimination method for the dynamic response of landslides in fault zones, thereby improving the accuracy of quantitative landslide dynamic response discrimination.

[0055] In some embodiments, the landslide dynamic response discrimination method based on a centrifuge vibration table provided by the present invention further includes: In step S6, a uniform interpolation grid map is generated and interpolated based on the sliding body boundary information, the coordinates of the measuring point, and the target ground motion characteristic parameters.

[0056] The sliding body boundary information refers to the sequence of boundary point coordinates on the XY plane. The grid density of the uniform interpolation grid can be set by the user, and the number of grid points in the Y direction is adaptively adjusted according to the aspect ratio of the slope to ensure interpolation accuracy.

[0057] In some embodiments, the method for generating and interpolating a uniform interpolation grid can be linear. In other embodiments, the method can be nearest neighbor. In still other embodiments, the method can be cubic or based on scipy.interpolate.griddata. Furthermore, the method for generating and interpolating a uniform interpolation grid also includes RBF interpolation, which has excellent smoothness and local responsiveness.

[0058] When processing scattered sensor data points, radial basis function interpolation exhibits excellent smoothness and local response capabilities, enabling it to better capture the nonlinear spatial variation characteristics of the seismic amplification factor. Based on the user-selected method, the sensor coordinates and corresponding seismic amplification factor values ​​are used as input and interpolated onto an automatically generated grid map to obtain continuous interpolation results.

[0059] In some embodiments, to show only the interior area of ​​the landslide, the slope outline can be masked. Eliminating invalid interpolation results outside the slope profile improves the accuracy and interpretability of the contour map. Specifically, a geometric path is constructed based on the input slope profile data. For each point in the interpolation grid, if the point is located outside the slope profile, its corresponding interpolation result, grid_zi, is set to an invalid value.

[0060] This allows for a more intuitive observation of the response.

[0061] In a preferred embodiment, the interpolation grid is set to 700×N (Nx=700) to ensure that the fineness of the subsequent two-dimensional distribution cloud map of the target seismic characteristic parameters is sufficient to reflect the subtle changes in the shear zone.

[0062] Please refer to Figure 7 , Figure 7 This is a two-dimensional distribution cloud of the landslide dynamic response discrimination method based on a centrifuge shaking table provided in this embodiment of the invention. Figure 1 Schematic diagram. In some embodiments, the landslide dynamic response discrimination method based on a centrifuge vibration table provided by the present invention further includes: By setting contour lines and color gradients on the uniform interpolation grid, a two-dimensional distribution cloud map of the target seismic characteristic parameters is obtained.

[0063] Visualization can be achieved using the `matplotlib.pyplot.contourf` and `contour` functions to plot a two-dimensional distribution cloud map of the seismic amplification factor using color fills and contour lines. The color gradient of the cloud map visually represents the magnitude of the amplification factor. Users can flexibly set the contour line level, color theme, color scale range, and font size to meet different visualization needs. In some embodiments, the locations of the original sensor data points can be overlaid on the cloud map for easy result verification. Thus, by visually interpreting the generated two-dimensional distribution cloud map, combined with the color gradient and contour line distribution, the areas within the landslide body with the strongest dynamic response under seismic loading can be identified intuitively and with high precision. These high-response areas typically correspond to the parts of the landslide most prone to damage and instability, providing crucial, quantitative scientific evidence for slope stability assessment, risk zoning, and the formulation of seismic reinforcement measures.

[0064] In some embodiments, the two-dimensional distribution cloud map is saved as a high-quality image file, and a detailed data table containing the original values ​​of each sensor point and the interpolated results can be exported.

[0065] In some embodiments, to prevent the interpolation algorithm from producing false "artifacts," the interpolation results are verified using a "cloud map-scatter point overlay verification method" after generating the two-dimensional distribution cloud map. Specifically, on the generated two-dimensional distribution cloud map, the actual measurements from the original sensor are overlaid as colored dots (Scatter Points) at the corresponding coordinate positions. If the dot color blends perfectly with the background ("Blend in"), it indicates that the interpolation result accurately reflects the actual measurements; if obvious color abrupt changes occur (such as the "bull's eye" phenomenon), it suggests that the RBF smoothing parameter needs to be adjusted or the kernel function needs to be changed. In this way, this visual verification method allows users to identify interpolation deviations at a glance, ensuring the credibility of the final published two-dimensional distribution cloud map.

[0066] In one specific embodiment, such as Figure 7 As shown, the horizontal axis represents the horizontal distance in millimeters, the left vertical axis represents the vertical elevation in millimeters, and the right vertical axis represents the acceleration amplification factor.

[0067] In one specific embodiment, please refer to Figure 3 , Figure 5 Continue to refer to Figure 7 The target landslide is the landslide in Zhang County, Gansu Province, as shown in the figure. The measuring points are arranged as follows. Figure 5 As shown, the simplified physical model of the cross-section is as follows: Figure 3 As shown, the centrifuge shaking table uses a generalized physical model based on the seismic profile of historical earthquake acceleration time history information, and sets a gravitational acceleration of 0.45. The target ground motion characteristic parameter is the ground motion amplification factor. The obtained data are as follows: ; Among them, the NO column represents Figure 5 The measurement points in the diagram have X and Y columns representing the horizontal and vertical coordinates. The peak ground acceleration (PGA) is calculated over a selected dynamic time interval. The PGA of measurement point A3 is selected as the baseline PGA, and the seismic amplification factor for each measurement point is calculated accordingly. The final result is as follows: Figure 7 The two-dimensional distribution cloud map shown.

[0068] The present invention also provides a landslide dynamic response discrimination device based on a centrifuge vibration table, comprising: The information acquisition unit is suitable for acquiring information on the target landslide, including landslide location, landslide scale, landslide boundary, landslide structure layering and physical properties, and historical earthquake acceleration time history.

[0069] The modeling unit is adapted to establish a profile generalized physical model of the target landslide based on the landslide location information, landslide scale information, landslide boundary information, landslide structure layering and physical property information, and a preset similarity ratio. At least two sets of measuring points with different coordinates are set in the profile generalized physical model, and measuring point acceleration sensors corresponding to the measuring points are set one by one.

[0070] The measuring point model domain acceleration response unit is adapted to simulate the vibration of the profile generalized physical model using a centrifuge shaking table based on the historical earthquake acceleration time history information, and to acquire measuring point model domain acceleration response data through the measuring point acceleration sensor. The measuring point model domain acceleration response data includes measuring point model domain acceleration data and timestamp data.

[0071] The preprocessing unit is adapted to preprocess the acceleration response data of the measurement point model domain according to the preset similarity ratio to obtain the acceleration response data of the measurement point prototype domain, wherein the acceleration response data of the measurement point prototype domain includes the acceleration data of the measurement point prototype domain and the corresponding timestamp data. The target ground motion feature extraction unit is adapted to extract target ground motion feature parameters from the acceleration response data of the prototype domain of the measuring point.

[0072] The present invention also provides a storage medium storing a program for landslide dynamic response discrimination based on a centrifuge vibration table, so as to implement the landslide dynamic response discrimination method based on a centrifuge vibration table described above.

[0073] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is accorded the widest scope consistent with the principles and novel features disclosed herein.

[0074] While the embodiments of the present invention have been disclosed above, the present invention is not limited thereto. Any person skilled in the art can make various changes and modifications without departing from the spirit and scope of the present invention; therefore, the scope of protection of the present invention should be determined by the scope defined in the claims.

Claims

1. A method for judging the dynamic response of landslides based on a centrifuge shaking table, characterized in that, include: Obtain information on the target landslide, including landslide location, landslide scale, landslide boundary, landslide structure layering and physical properties, and historical earthquake acceleration time history. Based on the landslide location information, landslide scale information, landslide boundary information, landslide structure layering and physical property information, and a preset similarity ratio, a profile generalized physical model of the target landslide is established. At least two sets of measuring points with different coordinates are set in the profile generalized physical model, and a measuring point acceleration sensor corresponding to each measuring point is set. Based on the historical earthquake acceleration time history information, a centrifuge shaking table is used to simulate the vibration of the profile generalized physical model, and the acceleration response data of the measuring point model domain is obtained through the measuring point acceleration sensor. The measuring point model domain acceleration response data includes measuring point model domain acceleration data and timestamp data. Preprocess the acceleration response data of the measurement point model domain according to the preset similarity ratio to obtain the acceleration response data of the measurement point prototype domain. The acceleration response data of the measurement point prototype domain includes the acceleration data of the measurement point prototype domain and the corresponding timestamp data. Target ground motion characteristic parameters are extracted from the acceleration response data of the prototype domain of the measuring point for landslide dynamic response discrimination.

2. The landslide dynamic response discrimination method based on a centrifuge vibration table as described in claim 1, characterized in that, The acquisition of acceleration response data in the measurement point model domain through the measurement point acceleration sensor includes: The raw output data of the acceleration response in the model domain of the measuring point is obtained through the measuring point acceleration sensor; The script converts the original output data of the acceleration response of the measurement point model domain into a CSV file and removes the descriptive header text to obtain the acceleration response data of the measurement point model domain.

3. The landslide dynamic response discrimination method based on a centrifuge vibration table as described in claim 2, characterized in that, The step involves preprocessing the acceleration response data of the measurement point model domain according to the preset similarity ratio to obtain the acceleration response data of the measurement point prototype domain. The acceleration response data of the measurement point prototype domain includes acceleration data of the measurement point prototype domain and corresponding timestamp data, including: Based on the preset similarity ratio, the acceleration response data of the measurement point model domain is converted into acceleration response data of the measurement point prototype domain to be preprocessed. Robust baseline correction is performed on the preprocessed data of the prototype domain acceleration response of the measurement point to obtain robust time-series data of the prototype domain acceleration response of the measurement point. The robust time-series data of the prototype domain acceleration response of the measurement point includes robust acceleration data of the prototype domain acceleration and corresponding timestamp data. The robust time-series data of the acceleration response of the prototype domain at the measurement point is obtained by Fourier transform, and the robust frequency-domain data of the acceleration response of the prototype domain at the measurement point is obtained by bandpass filtering, and the bandpass-domain data of the acceleration response of the prototype domain at the measurement point is obtained by inverse Fourier transform.

4. The landslide dynamic response discrimination method based on a centrifuge vibration table as described in claim 3, characterized in that, The robust baseline correction of the prototype domain acceleration response data at the measurement point yields robust time-series acceleration response data at the measurement point, including: The acceleration response data to be preprocessed in the prototype domain of the measuring point is subjected to mean-removal processing to obtain the mean-removed acceleration data in the prototype domain of the measuring point. Linear detrending is performed on the mean-de-trend data of the acceleration in the prototype domain of the measurement point to obtain robust time-series data of the acceleration response in the prototype domain of the measurement point.

5. The landslide dynamic response discrimination method based on a centrifuge vibration table as described in claim 4, characterized in that, The bandpass filtering of the robust frequency domain data of the prototype acceleration response at the measurement point yields the bandpass frequency domain data of the prototype acceleration response at the measurement point, including: The robust frequency domain data of the prototype acceleration response at the measurement point is filtered by a Butterworth bandpass filter to obtain the bandpass frequency domain data of the prototype acceleration response at the measurement point.

6. The landslide dynamic response discrimination method based on a centrifuge vibration table as described in claim 1, characterized in that, The target ground motion characteristic parameters include the ground motion amplification factor; The extraction of target ground motion characteristic parameters from the prototype domain acceleration response data of the measuring point includes: Based on the start and end times entered in the start and end time input boxes in the GUI interface, extract the selected dynamic time interval; The maximum absolute value of the selected dynamic time interval is calculated as the peak acceleration; The peak acceleration of one measuring point is selected from the at least two sets of measuring points as the reference peak acceleration. The ratio of the peak acceleration of the measuring point in the at least two sets of measuring points to the reference peak acceleration is calculated to obtain the seismic amplification factor of all the measuring points.

7. The landslide dynamic response discrimination method based on a centrifuge vibration table as described in any one of claims 1-6, characterized in that, Also includes: Based on the sliding body boundary information, the coordinates of the measuring points, and the target ground motion characteristic parameters, a uniform interpolation grid map is generated and interpolated.

8. The landslide dynamic response discrimination method based on a centrifuge vibration table as described in claim 7, characterized in that, Also includes: By setting contour lines and color gradients on the uniform interpolation grid, a two-dimensional distribution cloud map of the target seismic characteristic parameters is obtained.

9. A landslide dynamic response discrimination device based on a centrifuge vibration table, characterized in that, include: The information acquisition unit is suitable for acquiring information on the target landslide, including landslide location, landslide scale, landslide boundary, landslide structure layering and physical properties, and historical earthquake acceleration time history. The modeling unit is adapted to establish a profile generalized physical model of the target landslide based on the landslide location information, landslide scale information, landslide boundary information, landslide structure layering and physical property composition information, and a preset similarity ratio. At least two sets of measuring points with different coordinates are set in the profile generalized physical model, and measuring point acceleration sensors corresponding to the measuring points are set one by one. The measuring point model domain acceleration response unit is adapted to simulate the vibration of the profile generalized physical model using a centrifuge shaking table based on the historical earthquake acceleration time history information, and to acquire measuring point model domain acceleration response data through the measuring point acceleration sensor. The measuring point model domain acceleration response data includes measuring point model domain acceleration data and timestamp data. The preprocessing unit is adapted to preprocess the acceleration response data of the measurement point model domain according to the preset similarity ratio to obtain the acceleration response data of the measurement point prototype domain, wherein the acceleration response data of the measurement point prototype domain includes the acceleration data of the measurement point prototype domain and the corresponding timestamp data. The target ground motion feature extraction unit is adapted to extract target ground motion feature parameters from the acceleration response data of the prototype domain of the measuring point.

10. A storage medium, characterized in that, The storage medium stores a program for landslide dynamic response discrimination based on a centrifuge vibration table, so as to implement the landslide dynamic response discrimination method based on a centrifuge vibration table as described in any one of claims 1-8.

Images