Earthquake early warning-oriented high-rise building earthquake response real-time prediction method, device and equipment

Abstract

The invention provides an earthquake early warning-oriented high-rise building earthquake response real-time prediction method, device and equipment. The method comprises the steps of obtaining strong vibration data, a target high-rise building model and building attributes of the target high-rise building model; extracting seismic oscillation parameters and seismic parameters from the seismic initial waveform of the strong vibration data; the seismic oscillation parameters comprise peak acceleration, peak velocity, accumulated absolute velocity, important duration and Arias intensity; the earthquake parameters comprise earthquake magnitude, epicentral distance and site parameters; and inputting the seismic oscillation parameters, the seismic parameters and the building attributes into the high-rise building seismic response real-time prediction model to obtain a seismic response, and performing early warning according to the seismic response. According to the method, the device and the equipment provided by the invention, the earthquake response of the high-rise building can be efficiently and accurately predicted in real time, important technical support is provided for earthquake early warning and disaster risk reduction, and the method, the device and the equipment have relatively good practical value and wide application prospect.

Description

Technical Field

[0001] The present invention relates to the technical field of disaster prevention and mitigation engineering, and particularly to a real-time prediction method, device and equipment for earthquake response of high-rise buildings for earthquake early warning. Background Art

[0002] The real-time prediction of earthquake response of high-rise buildings is of great significance for the seismic performance evaluation of high-rise buildings during earthquakes.

[0003] Due to the high non-linearity of processes such as earthquake wave propagation and building dynamic amplification, it is difficult for traditional PBEE-based (Performance Based Earthquake Engineering) methods based on fixed function forms to capture the complex relationships therein, that is, the prediction accuracy of traditional PBEE-based methods for earthquake response of high-rise buildings is low. Summary of the Invention

[0004] The present invention provides a real-time prediction method, device and equipment for earthquake response of high-rise buildings for earthquake early warning, so as to solve the defect that the prediction accuracy of traditional PBEE-based methods based on fixed function forms for earthquake response of high-rise buildings in the prior art is low.

[0005] The present invention provides a real-time prediction method for earthquake response of high-rise buildings for earthquake early warning, including the following steps: Obtain strong motion data and a target high-rise building model, and obtain the building attributes of the target high-rise building model; the target high-rise building model is constructed based on a high-rise building model with strong motion monitoring data; Extract ground motion parameters and earthquake parameters from the initial earthquake waveform of the strong motion data; the ground motion parameters include peak acceleration, peak velocity, cumulative absolute velocity, significant duration, Arias intensity; the earthquake parameters include magnitude, epicentral distance and site parameters; Input the ground motion parameters, the earthquake parameters and the building attributes into a real-time prediction model for earthquake response of high-rise buildings, obtain the earthquake response output by the real-time prediction model for earthquake response of high-rise buildings, and issue a warning according to the earthquake response; The real-time prediction model for earthquake response of high-rise buildings is constructed based on a deep neural network; The real-time prediction model for earthquake response of high-rise buildings includes multiple fully connected hidden layers, and the number of neurons and the setting of activation functions in each fully connected hidden layer are adjusted based on the characteristics of earthquake response data corresponding to high-rise buildings.

[0006] A real-time prediction method for the seismic response of high-rise buildings for earthquake early warning provided by the present invention, the training steps of the real-time prediction model for the seismic response of high-rise buildings include: Obtain sample ground motion parameters, sample earthquake parameters from a strong motion database, obtain sample building attributes, and the labeled seismic response corresponding to the sample ground motion parameters, the sample earthquake parameters, and the sample building attributes; Obtain an initial real-time prediction model for the seismic response of high-rise buildings; Input the sample ground motion parameters, the sample earthquake parameters, and the sample building attributes into the initial real-time prediction model for the seismic response of high-rise buildings to obtain the predicted seismic response output by the initial real-time prediction model for the seismic response of high-rise buildings; Based on the difference between the predicted seismic response and the labeled seismic response, determine the target loss, and perform parameter iteration on the initial real-time prediction model for the seismic response of high-rise buildings based on the target loss to obtain the real-time prediction model for the seismic response of high-rise buildings.

[0007] A real-time prediction method for the seismic response of high-rise buildings for earthquake early warning provided by the present invention, the steps for obtaining the labeled seismic response include: Based on the urban seismic elastoplastic analysis method, determine the labeled seismic response of the target high-rise building under earthquake ground motion.

[0008] A real-time prediction method for the seismic response of high-rise buildings for earthquake early warning provided by the present invention, the determining the labeled seismic response of the target high-rise building model under earthquake ground motion based on the urban seismic elastoplastic analysis method includes: Based on the multi-degree-of-freedom shear layer model, concentrate the floor mass of the target high-rise building model on a single mass point, and use a trilinear backbone curve and a single-parameter hysteretic model to simulate the nonlinear behavior between floors in the target high-rise building model; Among them, the backbone curve parameters of the trilinear backbone curve are calibrated by a calibration method based on the HAZUS capacity curve database.

[0009] A real-time prediction method for the seismic response of high-rise buildings for earthquake early warning provided by the present invention, the strong motion database is constructed based on earthquake ground motion data with a magnitude between 4 and 8, an epicentral distance less than 200 km, and site parameters between 100 and 900 m / s.

[0010] A real-time prediction method for the seismic response of high-rise buildings for earthquake early warning provided by the present invention, the extracting the ground motion parameters and earthquake parameters from the seismic initial waveform of the strong motion data includes: Adopt a P-wave automatic recognition algorithm to detect the arrival time of the P-wave in the seismic initial waveform of the strong motion data; Intercept the ground motion record in the first 3 seconds after the arrival time of the P wave based on the arrival time of the P wave; Determine the ground motion parameters and the seismic parameters based on the ground motion record in the first 3 seconds.

[0011] The present invention also provides a device for real-time prediction of the seismic response of high-rise buildings for earthquake early warning, including the following units: An acquisition unit, configured to acquire strong motion data and a target high-rise building model, and acquire the building attributes of the target high-rise building model; the target high-rise building model is constructed based on a high-rise building model with strong motion monitoring data; An extraction unit, configured to extract ground motion parameters and seismic parameters from the initial seismic waveform of the strong motion data; the ground motion parameters include peak acceleration, peak velocity, cumulative absolute velocity, significant duration, and Arias intensity; the seismic parameters include magnitude, epicentral distance, and site parameters; A prediction unit, configured to input the ground motion parameters, the seismic parameters, and the building attributes into a real-time prediction model of the seismic response of high-rise buildings, obtain the seismic response output by the real-time prediction model of the seismic response of high-rise buildings, and issue an early warning based on the seismic response; The real-time prediction model of the seismic response of high-rise buildings is constructed based on a deep neural network; The real-time prediction model of the seismic response of high-rise buildings includes multiple fully-connected hidden layers, and the number of neurons and the setting of activation functions in each fully-connected hidden layer are adjusted based on the characteristics of the seismic response data corresponding to high-rise buildings.

[0012] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, it implements the real-time prediction method for the seismic response of high-rise buildings for earthquake early warning as described in any one of the above.

[0013] The present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the real-time prediction method for the seismic response of high-rise buildings for earthquake early warning as described in any one of the above.

[0014] The present invention also provides a computer program product, including a computer program. When the computer program is executed by a processor, it implements the real-time prediction method for the seismic response of high-rise buildings for earthquake early warning as described in any one of the above.

[0015] The real-time prediction method, device and equipment for high-rise building seismic response for earthquake early warning provided by the present invention obtain strong motion data and a target high-rise building model, and obtain the building attributes of the target high-rise building model. Then, ground motion parameters and earthquake parameters are extracted from the initial seismic waveform of the strong motion data. Finally, the ground motion parameters, earthquake parameters and building attributes are input into the real-time prediction model for high-rise building seismic response to obtain the seismic response output by the real-time prediction model for high-rise building seismic response, and an early warning is made based on the seismic response. Predicting the seismic response based on the ground motion parameters, earthquake parameters and building attributes can more comprehensively and finely reflect the impact of the earthquake on high-rise buildings, improving the accuracy and reliability of seismic response prediction. Moreover, the present invention can predict the seismic response of high-rise buildings in real time, efficiently and accurately, providing important technical support for earthquake early warning, building emergency management and disaster risk assessment, and having good practical value and broad application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0017] Figure 1 It is a schematic flowchart of the real-time prediction method for high-rise building seismic response for earthquake early warning provided by the present invention.

[0018] Figure 2 It is a curve graph of the loss function of the training set and the test set provided by the present invention.

[0019] Figure 3 It is a comparison graph of the maximum inter-story drift angle, maximum floor acceleration and top floor acceleration of the model provided by the present invention on the test set with the true values.

[0020] Figure 4 It is a confusion matrix of the measured data of the top floor acceleration of the model provided by the present invention and the prediction results of the present model.

[0021] Figure 5 It is a schematic diagram of the EEWnet network architecture based on DNN provided by the present invention.

[0022] Figure 6 It is a schematic structural diagram of the real-time prediction device for high-rise building seismic response for earthquake early warning provided by the present invention.

[0023] Figure 7 It is a schematic structural diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0024] To make the objectives, technical solutions, and advantages of the present invention clearer, the technical solutions in the present invention will be clearly and completely described below with reference to the accompanying drawings in the present invention. Apparently, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present invention without creative efforts shall fall within the protection scope of the present invention.

[0025] Currently, due to the high nonlinearity of processes such as ground motion propagation and building dynamic amplification, it is relatively difficult for traditional PBEE-based methods based on fixed function forms to capture the complex relationships therein. Deep learning methods can mine complex nonlinear patterns in data and provide an important means to address this challenge. However, at present, the real-time prediction of high-rise building seismic responses based on deep learning is relatively limited.

[0026] To improve the above problems, the present invention provides a real-time prediction method for high-rise building seismic responses for earthquake early warning. Figure 1 It is a schematic flowchart of the real-time prediction method for high-rise building seismic responses for earthquake early warning provided by the present invention. As Figure 1 shown, the method includes step 110, step 120, and step 130.

[0027] Step 110: Obtain strong motion data and a target high-rise building model, and obtain the building attributes of the target high-rise building model; the target high-rise building model is constructed based on a high-rise building model with strong motion monitoring data. Step 120: Extract ground motion parameters and earthquake parameters from the initial earthquake waveform of the strong motion data; the ground motion parameters include peak acceleration, peak velocity, cumulative absolute velocity, significant duration, and Arias intensity; the earthquake parameters include magnitude, epicentral distance, and site parameters. Step 130: Input the ground motion parameters, the earthquake parameters, and the building attributes into a real-time prediction model for high-rise building seismic responses to obtain the seismic response output by the real-time prediction model for high-rise building seismic responses, and issue an early warning based on the seismic response.

[0028] Specifically, obtain strong motion data and a target high-rise building model, and obtain the building attributes of the target high-rise building model. Among them, the strong motion data is the data measured by an acceleration sensor at the location where the high-rise building is located.

[0029] Among them, the target high-rise building model is constructed based on high-rise building models with strong motion monitoring data. Here, 6 high-rise building models with strong motion monitoring data can be selected. Since the target high-rise building model is constructed based on strong motion monitoring data, it can more accurately reflect the true response of high-rise buildings under extreme conditions such as earthquakes. It can be understood that strong motion monitoring data usually comes from sensors installed on high-rise buildings during earthquakes, which can capture and record key parameters such as acceleration and displacement of the building under earthquake action in real time.

[0030] Here, the magnitude is a measure of the energy released by an earthquake, the epicentral distance is the distance from the observation point to the earthquake epicenter, and the site parameters usually refer to the parameters used to reflect the spatial distribution and dynamic characteristics of recent sediments or soil layers covering the bedrock.

[0031] Among them, the building attributes include the number of floors, height, structural type, and construction year, etc., and the embodiments of the present invention do not make specific limitations on this. Common building structural types include brick-concrete structure (mostly used in low-rise or multi-story residential buildings), frame structure (mostly used in mid-high and high-rise buildings), frame-shear wall structure (used to increase the stiffness and integrity of buildings), shear wall structure (mostly used in mid-high and high-rise buildings), and steel structure (used in super high-rise buildings).

[0032] After obtaining the strong motion data, ground motion parameters and earthquake parameters can be extracted from the initial earthquake waveform of the strong motion data. Among them, the ground motion parameters include Peak Ground Acceleration (PGA), Peak Ground Velocity (PGV), Cumulative Absolute Velocity (CAV), Significant Duration, Arias intensity, etc. Among them, the earthquake parameters include magnitude, epicentral distance, and site parameters.

[0033] Among them, PGA is the maximum absolute value of the acceleration of the surface mass point movement during the earthquake vibration. PGV is the maximum value of the velocity describing the ground vibration, usually in meters per second (m / s). It is a representation of the amplitude of ground motion and is often used together with PGA to evaluate the intensity and damage potential of ground motion. CAV is the cumulative amount of the absolute value of the ground mass point velocity during the earthquake vibration. It can reflect the total energy of the ground motion and the influence of the duration on the building. Significant Duration is a measure of the duration of the ground motion and is used to reflect the continuous action time of the ground motion. Arias intensity is a parameter used to quantify the energy size of the ground motion, and Arias intensity can reflect the total energy and energy distribution of the ground motion.

[0034] Finally, input the ground motion parameters, earthquake parameters, and building attributes into the real-time prediction model for the seismic response of high-rise buildings to obtain the seismic response output by the real-time prediction model for the seismic response of high-rise buildings, and issue an early warning based on the seismic response.

[0035] Here, the seismic response includes key response indicators such as inter-story drift ratio and floor acceleration, and the embodiments of the present invention do not make specific limitations thereto.

[0036] For example, a floor acceleration (such as 0.5 or 1.5 m / s 2 ) can be used as the threshold for the floor acceleration early warning of high-rise buildings. If the floor acceleration of a high-rise building exceeds this threshold, an early warning is issued.

[0037] It should be noted that in view of the characteristics of the real-time prediction of the seismic response of high-rise buildings, the model architecture and optimization strategy are designed, the key parameters of the model are adjusted in combination with the distribution characteristics of the building response data, and then based on the non-linear relationship between the ground motion parameters, earthquake parameters, and building attributes, the network depth, the number of neurons, and the type of activation function of the model are optimized. Under the condition of low signal-to-noise ratio seismic records, the generalization ability of the model is improved through regularization methods and data augmentation techniques. Finally, a comparative analysis method is used to determine the network hyperparameter combination suitable for real-time prediction of seismic response based on the prediction accuracy and physical rationality of the key response indicators of high-rise buildings.

[0038] The method provided by the embodiments of the present invention obtains strong motion data and a target high-rise building model, obtains the building attributes of the target high-rise building model, and then extracts the ground motion parameters and earthquake parameters from the seismic initial waveform of the strong motion data; finally, inputs the ground motion parameters, earthquake parameters, and building attributes into the real-time prediction model for the seismic response of high-rise buildings to obtain the seismic response output by the real-time prediction model for the seismic response of high-rise buildings, and issues an early warning based on the seismic response. Predicting the seismic response based on the ground motion parameters, earthquake parameters, and building attributes can more comprehensively and finely reflect the impact of earthquakes on high-rise buildings, improving the accuracy and reliability of seismic response prediction; moreover, the present invention can predict the seismic response of high-rise buildings in real time, efficiently, and accurately, providing important technical support for earthquake early warning and reducing disaster risks, and having good practical value and broad application prospects.

[0039] Based on the above embodiments, the training steps of the real-time prediction model for the seismic response of high-rise buildings include: Step 210, obtain sample ground motion parameters, sample earthquake parameters from the strong motion database, obtain sample building attributes, and the labeled seismic response corresponding to the sample ground motion parameters, the sample earthquake parameters, and the sample building attributes; Step 220, obtain the initial real-time prediction model for the seismic response of high-rise buildings; Step 230: Input the sample ground motion parameters, the sample earthquake parameters, and the sample building attributes into the initial real-time prediction model for high-rise building earthquake response to obtain the predicted earthquake response output by the initial real-time prediction model for high-rise building earthquake response. Step 240: Determine the target loss based on the difference between the predicted earthquake response and the labeled earthquake response, and perform parameter iteration on the initial real-time prediction model for high-rise building earthquake response based on the target loss to obtain the real-time prediction model for high-rise building earthquake response.

[0040] Specifically, in order to better obtain the real-time prediction model for high-rise building earthquake response, the training can be carried out based on the following steps: First, obtain the sample ground motion parameters, the sample earthquake parameters from the strong motion database, and obtain the sample building attributes, as well as the labeled earthquake response corresponding to the sample ground motion parameters, the sample earthquake parameters, and the sample building attributes. Then, obtain the initial real-time prediction model for high-rise building earthquake response. Among them, the strong motion database is constructed based on the ground motion data with a magnitude between 4 and 8, an epicentral distance less than 200 km, and a site parameter between 100 - 900 m / s.

[0041] Here, the parameters of the initial real-time prediction model for high-rise building earthquake response can be set in advance or randomly generated. The embodiments of the present invention do not make specific limitations on this.

[0042] Among them, the initial real-time prediction model for high-rise building earthquake response is constructed based on a deep neural network (Deep Neural Networks, DNN). The initial real-time prediction model for high-rise building earthquake response includes multiple fully connected hidden layers. The number of neurons and the setting of the activation function in each fully connected hidden layer are adjusted based on the characteristics of the earthquake response data corresponding to high-rise buildings.

[0043] The number of neurons in each fully connected hidden layer is a key parameter, which directly affects the complexity and learning ability of the model. In practical applications, the setting of the number of neurons is usually adjusted based on the characteristics of the earthquake response data corresponding to high-rise buildings. For example, if the earthquake response data contains rich non-linear features, then it may be necessary to increase the number of neurons in the hidden layer to improve the fitting ability of the model.

[0044] The activation function is used to introduce non-linear characteristics, enabling the network to learn complex function mappings. In the real-time prediction model for high-rise building earthquake response, an appropriate activation function (such as ReLU, Sigmoid, or Tanh, etc.) can be selected to enhance the non-linear expression ability of the model. The selection of these activation functions will be weighed based on the characteristics of the earthquake response data and the training effect of the model.

[0045] After obtaining the sample ground motion parameters, sample earthquake parameters, and sample building attributes, the sample ground motion parameters, sample earthquake parameters, and sample building attributes can be input into the initial real-time prediction model for high-rise building earthquake response to obtain the predicted earthquake response output by the initial real-time prediction model for high-rise building earthquake response.

[0046] Finally, based on the difference between the predicted earthquake response and the labeled earthquake response, the target loss is determined, and the parameters of the initial real-time prediction model for high-rise building earthquake response are iterated based on the target loss. The initial real-time prediction model for high-rise building earthquake response after completing the parameter iteration is used as the real-time prediction model for high-rise building earthquake response.

[0047] It can be understood that the greater the difference between the predicted earthquake response and the labeled earthquake response, the greater the target loss; the smaller the difference between the predicted earthquake response and the labeled earthquake response, the smaller the target loss.

[0048] It should be noted that during the training process, network hyperparameter analysis is carried out on the initial real-time prediction model for high-rise building earthquake response, the parameters used are analyzed, and optimal parameters are determined. By changing the model depth, the number of neurons in each hidden layer, the batch size, the learning rate, and the type of optimizer, observe their effects on the real-time prediction results of the earthquake response of high-rise buildings to determine the optimal parameters.

[0049] Specifically, the determined optimal parameters are mainly: Batch size = 512, the activation function uses Leakyrelu, the optimizer uses the Adam optimizer, the learning rate is set to 0.0015, and the Dropout rate = 0.1.

[0050] Furthermore, the steps of training the real-time prediction model for high-rise building earthquake response in the embodiments of the present invention include: all sample data are reasonably divided into a training set, a test set, and a validation set. When training, the training set and the validation set are used, and the test set is used as the data set for testing the prediction ability of the trained model. The prediction accuracy of the model is measured by the Mean Absolute Error (MAE), Mean Squared Error (MSE), Root Mean Square Error (RMSE), and Coefficient of Determination (R 2 ). Based on the optimal model, the entire data set is trained, and the number of training iterations is reasonably designed. When the model can achieve the convergence of the validation error and the mean squared errors of the training set and the validation set both decrease after multiple iterations, it indicates that the training effect has been achieved.

[0051] In one example, the embodiments herein provide and obtain through training a specific implementation of a real-time prediction model for the seismic response of high-rise buildings: The data used is divided according to the ratio of training set: test set: validation set = 0.8:0.1:0.1, and the prediction accuracy of the model is measured by the mean absolute error, mean square error, root mean square error, and coefficient of determination. Figure 2 This is the loss function curve graph of the training set and test set provided by the present invention. As Figure 2 shown, during the training process, as the number of training iterations increases, the loss values of both the training set and the test set gradually decrease, the prediction performance of the model is continuously improved, and the loss curve of the training set is close to that of the test set, indicating that there is no obvious overfitting phenomenon in the model. Thus, a real-time prediction model for the seismic response of high-rise buildings is established.

[0052] In a specific implementation method, the trained high-rise building seismic response prediction model is evaluated, and the prediction results of the model on the maximum inter-story drift ratio (MIDR), maximum peak floor acceleration (MPFA), and peak top-floor acceleration (PTFA) in the test set are analyzed emphatically. Figure 3 This is the comparison graph of the maximum inter-story drift ratio, maximum peak floor acceleration, and peak top-floor acceleration of the model provided by the present invention with the true values in the test set. As Figure 3 shown, the comparison graph of the predicted values and the target values of the three predicted seismic responses: the maximum inter-story drift ratio, the maximum peak floor acceleration, and the peak top-floor acceleration in the test set. It further quantitatively shows the comprehensive evaluation indexes of the model, including the mean absolute error, mean square error, root mean square error, and coefficient of determination. It can be seen from the results that the model shows high prediction accuracy in all indexes in the test set. Specifically, in the prediction of the maximum inter-story drift ratio by the model, the MAE is 0.291, the RMSE is 0.379, and the R 2 reaches 0.90, indicating that the model can better capture the inter-story deformation characteristics of the structure. For the maximum peak floor acceleration and the peak top-floor acceleration, the R 2 reaches 0.96 and 0.95, further proving the excellent performance of the model in acceleration prediction. At the same time, the model only takes 0.001632 s to complete the prediction, which can meet the requirements of real-time early warning.

[0053] In a specific implementation method, the CESMD database provides valuable building monitoring data, and also provides the basic attributes of the monitored building and the layout positions of the monitoring sensors. For this reason, in this paper, 6 high-rise buildings with a relatively large number of measured data (a total of 74 groups of data) are selected from this database as the case analysis objects for verification. Considering that sensors are arranged on the top floors of the buildings, in the embodiments of the present invention, the monitoring results of the top floors are selected to verify the proposed method. In this paper, the acceleration values of 0.5 and 1.5 m / s 2 are taken as the thresholds for high-rise acceleration early warning. When the acceleration is greater than the acceleration threshold, people will feel uncomfortable and very uncomfortable. Figure 4 is the confusion matrix of the measured top-floor acceleration data of the model provided by the present invention and the prediction results of this model. As Figure 4 shown, <Threshold means less than the threshold, ≥Threshold means greater than the threshold, True means measured, predicted means predicted, and the confusion matrix is the accuracy of the predicted classification. By comparing the measured data with the prediction results of the model in the embodiments of the present invention, it can be seen from the confusion matrix and relevant evaluation indicators that the model adopted in the embodiments of the present invention performs excellently in the high-rise building acceleration earthquake early warning, with a relatively high accuracy and precision, up to 94.59% at most. Thus, it is determined that the method can preferably predict the earthquake response of high-rise buildings.

[0054] In summary, Figure 5 is the schematic diagram of the DNN-based network architecture EEWnet provided by the present invention. As Figure 5 shown, the network optimization adopts an adaptive optimization algorithm, combined with a learning rate adjustment strategy to improve the convergence efficiency and prediction accuracy of the model. By adjusting the network architecture, activation function, and optimization method, a deep neural network EEWnet (Earthquake Early Warning net) suitable for real-time prediction of high-rise building earthquake response is constructed. Among them, EEWnet is the real-time prediction model of high-rise building earthquake response, and EDPs (Engineering Demand Parameters) are the earthquake responses of high-rise buildings.

[0055] The above real-time prediction model and method for high-rise building structure earthquake response based on deep learning provided in this embodiment establish a real-time prediction model for high-rise building earthquake response and are verified and tested. Thus, it is determined that the method can preferably predict the earthquake response of high-rise buildings, providing an important reference basis for the earthquake early warning of high-rise buildings.

[0056] Based on the above embodiments, the steps for obtaining the label earthquake response include: Based on the urban seismic elastoplastic analysis method, determine the seismic response of the target high-rise building under the action of ground motion.

[0057] Specifically, based on the seismic elastoplastic analysis method, determine the seismic response of the target high-rise building under the action of ground motion.

[0058] Specifically, based on the multi-degree-of-freedom shear layer model, concentrate the floor mass of the target high-rise building model on a single mass point, and use the trilinear backbone curve and the single-parameter hysteretic model to simulate the nonlinear behavior between floors in the target high-rise building model. Among them, the backbone curve parameters of the trilinear backbone curve are calibrated by a calibration method based on the HAZUS capacity curve database.

[0059] That is, use an open-source earthquake damage framework based on the urban seismic elastoplastic analysis method to calculate the seismic response of high-rise buildings. This model uses a multi-degree-of-freedom shear layer model to simulate the building, concentrates the mass of each floor of the building on a single mass point, the backbone curve of the spring between floors in the multi-degree-of-freedom model uses the trilinear backbone curve recommended in the HAZUS report, the hysteretic model uses a single-parameter hysteretic model, and uses a calibration method based on the HAZUS capacity curve database to calibrate its backbone curve parameters.

[0060] Based on the above embodiments, step 120 includes: Step 121, use a P-wave automatic recognition algorithm to detect the arrival time of the P-wave in the seismic initial waveform of the strong motion data; Step 122, based on the arrival time of the P-wave, intercept the first 3 seconds of ground motion records after the arrival time of the P-wave; Step 123, based on the first 3 seconds of ground motion records, determine the ground motion parameters and the seismic parameters.

[0061] Specifically, use a P-wave automatic recognition algorithm to detect the arrival time of the P-wave in the seismic initial waveform of the strong motion data. For example, use the P PHASE P ICKER algorithm to detect the arrival time of the P-wave in the seismic initial waveform of the strong motion data. The P PHASE P ICKER algorithm is a P-wave phase arrival automatic picking algorithm for single-component acceleration or broadband velocity records. This algorithm converts the seismic signal into the response domain of a single-degree-of-freedom (SDOF) oscillator with viscous damping and tracks the rate of change of the dissipated damping energy to pick up the P-wave phase. The SDOF oscillator has a short natural period (about 0.01 seconds) and a high damping ratio (60%), avoiding resonance. The damping energy is zero at the initial stage of the signal, close to zero before the arrival of the P-wave, and increases rapidly when the P-wave arrives. Through this energy change, the P PHASE PICKER The algorithm can accurately detect the arrival time of the P-wave phase under low signal-to-noise ratio conditions. Using this method to process the ground motion records, after identifying the P-wave, select the first 3 s of the ground motion records and calculate their ground motion parameters, including PGA, significant duration (in the embodiments of the present invention, the time interval between 5% - 95% Arias intensity is selected as the index of the duration, that is D S5-95 ), Arias intensity, CAV, PGV, etc.

[0062] Based on any of the above embodiments, a real-time prediction method for the seismic response of high-rise buildings for earthquake early warning is as follows: The first step is the training step of the real-time prediction model for the seismic response of high-rise buildings, including: Obtain sample ground motion parameters, sample seismic parameters from the strong motion database, and obtain sample building attributes, as well as the labeled seismic responses corresponding to the sample ground motion parameters, sample seismic parameters, and sample building attributes; the strong motion database is constructed based on ground motion data with a magnitude between 4 - 8, an epicentral distance less than 200 km, and site parameters between 100 - 900 m / s; Obtain the initial real-time prediction model for the seismic response of high-rise buildings; Input the sample ground motion parameters, sample seismic parameters, and sample building attributes into the initial real-time prediction model for the seismic response of high-rise buildings to obtain the predicted seismic response output by the initial real-time prediction model for the seismic response of high-rise buildings; Based on the difference between the predicted seismic response and the labeled seismic response, determine the target loss, and perform parameter iteration on the initial real-time prediction model for the seismic response of high-rise buildings based on the target loss to obtain the real-time prediction model for the seismic response of high-rise buildings.

[0063] Among them, the steps for obtaining the labeled seismic response include: Based on the urban seismic elastoplastic analysis method, determine the labeled seismic response of the target high-rise building under ground motion. For example, based on the multi-degree-of-freedom shear layer model, concentrate the floor masses of the target high-rise building model on a single mass point, and use the trilinear backbone curve and single-parameter hysteretic model to simulate the nonlinear behavior between floors in the target high-rise building model. Among them, the backbone curve parameters of the trilinear backbone curve are calibrated by a calibration method based on the HAZUS capacity curve database.

[0064] The second step is to obtain strong motion data and the target high-rise building model, and obtain the building attributes of the target high-rise building model. The target high-rise building model is constructed based on a high-rise building model with strong motion monitoring data.

[0065] Among them, the real-time prediction model for high-rise building seismic response is constructed based on a deep neural network. The real-time prediction model for high-rise building seismic response includes multiple fully connected hidden layers. The number of neurons and the setting of activation functions in each fully connected hidden layer are adjusted based on the characteristics of the seismic response data corresponding to the high-rise building.

[0066] In the third step, the P-wave automatic recognition algorithm is used to detect the arrival time of the P-wave in the seismic initial waveform of the strong motion data. Then, based on the arrival time of the P-wave, the first 3 seconds of the ground motion record after the arrival time of the P-wave is intercepted. Finally, based on the first 3 seconds of the ground motion record, the ground motion parameters and seismic parameters are determined. Among them, the ground motion parameters include peak acceleration, peak velocity, cumulative absolute velocity, significant duration, and Arias intensity; the seismic parameters include magnitude, epicentral distance, and site parameters. In the fourth step, the ground motion parameters, seismic parameters, and building attributes are input into the real-time prediction model for high-rise building seismic response, and the seismic response output by the real-time prediction model for high-rise building seismic response is obtained, and an early warning is made according to the seismic response.

[0067] Next, the real-time prediction device for high-rise building seismic response for earthquake early warning provided by the present invention will be described. The real-time prediction device for high-rise building seismic response for earthquake early warning described below can be mutually referred to the real-time prediction method for high-rise building seismic response for earthquake early warning described above.

[0068] Based on any of the above embodiments, the present invention provides a real-time prediction device for high-rise building seismic response for earthquake early warning. Figure 6 is a schematic structural diagram of the real-time prediction device for high-rise building seismic response for earthquake early warning provided by the present invention, as Figure 6 shown, the device includes: An acquisition unit 610, configured to acquire strong motion data and a target high-rise building model, and acquire the building attributes of the target high-rise building model; the target high-rise building model is constructed based on a high-rise building model with strong motion monitoring data. An extraction unit 620, configured to extract ground motion parameters and seismic parameters from the seismic initial waveform of the strong motion data; the ground motion parameters include peak acceleration, peak velocity, cumulative absolute velocity, significant duration, and Arias intensity; the seismic parameters include magnitude, epicentral distance, and site parameters. A prediction unit 630, configured to input the ground motion parameters, the seismic parameters, and the building attributes into a real-time prediction model for high-rise building seismic response, obtain the seismic response output by the real-time prediction model for high-rise building seismic response, and make an early warning according to the seismic response. The real-time prediction model for high-rise building seismic response is constructed based on a deep neural network. The real-time prediction model for the seismic response of high-rise buildings includes multiple fully-connected hidden layers. The number of neurons and the settings of activation functions in each fully-connected hidden layer are adjusted based on the characteristics of the seismic response data corresponding to the high-rise buildings.

[0069] The device provided by the embodiment of the present invention acquires strong ground motion data and a target high-rise building model, and acquires the building attributes of the target high-rise building model. Then, it extracts ground motion parameters and seismic parameters from the initial seismic waveform of the strong ground motion data. Finally, it inputs the ground motion parameters, seismic parameters, and building attributes into the real-time prediction model for the seismic response of high-rise buildings to obtain the seismic response output by the real-time prediction model for the seismic response of high-rise buildings, and issues an early warning based on the seismic response. Predicting the seismic response based on the ground motion parameters, seismic parameters, and building attributes can more comprehensively and precisely reflect the impact of earthquakes on high-rise buildings, improving the accuracy and reliability of seismic response prediction. Moreover, the present invention can predict the seismic response of high-rise buildings in real time, efficiently, and accurately, providing important technical support for earthquake early warning and reducing disaster risks, and having good practical value and broad application prospects.

[0070] Based on any of the above embodiments, it further includes a training unit, and the training unit is specifically used for: Acquire sample ground motion parameters, sample seismic parameters from a strong ground motion database, and acquire sample building attributes, as well as the labeled seismic response corresponding to the sample ground motion parameters, the sample seismic parameters, and the sample building attributes; Acquire an initial real-time prediction model for the seismic response of high-rise buildings; Input the sample ground motion parameters, the sample seismic parameters, and the sample building attributes into the initial real-time prediction model for the seismic response of high-rise buildings to obtain the predicted seismic response output by the initial real-time prediction model for the seismic response of high-rise buildings; Based on the difference between the predicted seismic response and the labeled seismic response, determine the target loss, and perform parameter iteration on the initial real-time prediction model for the seismic response of high-rise buildings based on the target loss to obtain the real-time prediction model for the seismic response of high-rise buildings.

[0071] Based on any of the above embodiments, it further includes a labeled seismic response acquisition unit, and the labeled seismic response acquisition unit is specifically used for: Based on the urban seismic elastoplastic analysis method, determine the labeled seismic response of the target high-rise building under the action of ground motion.

[0072] Based on any of the above embodiments, the prediction unit 630 is specifically used for: Based on the multi-degree-of-freedom shear layer model, concentrate the floor mass of the target high-rise building model on a single mass point, and use the trilinear backbone curve and the single-parameter hysteretic model to simulate the nonlinear behavior between floors in the target high-rise building model; Among them, the backbone curve parameters of the trilinear backbone curve are calibrated by a calibration method based on the HAZUS capacity curve database.

[0073] Based on any of the above embodiments, the strong motion database is constructed based on ground motion data with a magnitude between 4 and 8, an epicentral distance less than 200 km, and site parameters between 100 and 900 m / s.

[0074] Based on any of the above embodiments, the extraction unit 620 is specifically configured to: Adopt a P-wave automatic recognition algorithm to detect the arrival time of the P-wave in the initial seismic waveform of the strong motion data; Based on the arrival time of the P-wave, intercept the first 3 seconds of ground motion records after the arrival time of the P-wave; Based on the first 3 seconds of ground motion records, determine the ground motion parameters and the seismic parameters.

[0075] Figure 7 is a schematic structural diagram of the electronic device provided by the present invention, as Figure 7As shown in the figure, the electronic device may include: a processor 710, a communications interface 720, a memory 730, and a communication bus 740. Among them, the processor 710, the communications interface 720, and the memory 730 complete communication with each other through the communication bus 740. The processor 710 may call the logical instructions in the memory 730 to execute a real-time prediction method for the earthquake response of high-rise buildings for earthquake early warning. The method includes: obtaining strong motion data and a target high-rise building model, and obtaining the building attributes of the target high-rise building model; the target high-rise building model is constructed based on a high-rise building model with strong motion monitoring data; extracting ground motion parameters and earthquake parameters from the initial earthquake waveform of the strong motion data; the ground motion parameters include peak acceleration, peak velocity, cumulative absolute velocity, significant duration, and Arias intensity; the earthquake parameters include magnitude, epicentral distance, and site parameters; inputting the ground motion parameters, the earthquake parameters, and the building attributes into a real-time prediction model for the earthquake response of high-rise buildings to obtain the earthquake response output by the real-time prediction model for the earthquake response of high-rise buildings, and making an early warning based on the earthquake response; the real-time prediction model for the earthquake response of high-rise buildings is constructed based on a deep neural network; the real-time prediction model for the earthquake response of high-rise buildings includes multiple fully connected hidden layers, and the number of neurons and the setting of the activation function in each fully connected hidden layer are adjusted based on the characteristics of the earthquake response data corresponding to the high-rise building.

[0076] In addition, when the logical instructions in the above-mentioned memory 730 are implemented in the form of software functional units and sold or used as an independent product, they can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROMs, Read-Only Memories), random access memories (RAMs, Random Access Memories), magnetic disks, or optical discs that can store program codes.

[0077] On the other hand, the present invention also provides a computer program product, which includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the real-time prediction method for the earthquake response of high-rise buildings for earthquake early warning provided by the above-mentioned various methods. The method includes: acquiring strong motion data and a target high-rise building model, and acquiring the building attributes of the target high-rise building model; the target high-rise building model is constructed based on the high-rise building model with strong motion monitoring data; extracting ground motion parameters and earthquake parameters from the initial earthquake waveform of the strong motion data; the ground motion parameters include peak acceleration, peak velocity, cumulative absolute velocity, significant duration, Arias intensity; the earthquake parameters include magnitude, epicentral distance and site parameters; inputting the ground motion parameters, the earthquake parameters and the building attributes into a real-time prediction model for the earthquake response of high-rise buildings to obtain the earthquake response output by the real-time prediction model for the earthquake response of high-rise buildings, and making an early warning according to the earthquake response; the real-time prediction model for the earthquake response of high-rise buildings is constructed based on a deep neural network; the real-time prediction model for the earthquake response of high-rise buildings includes multiple layers of fully connected hidden layers, and the number of neurons and the setting of activation functions in each layer of fully connected hidden layers are adjusted based on the characteristics of the earthquake response data corresponding to the high-rise building.

[0078] In yet another aspect, the present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it is implemented to execute the real-time prediction method for the earthquake response of high-rise buildings for earthquake early warning provided by the above-mentioned various methods. The method includes: acquiring strong motion data and a target high-rise building model, and acquiring the building attributes of the target high-rise building model; the target high-rise building model is constructed based on the high-rise building model with strong motion monitoring data; extracting ground motion parameters and earthquake parameters from the initial earthquake waveform of the strong motion data; the ground motion parameters include peak acceleration, peak velocity, cumulative absolute velocity, significant duration, Arias intensity; the earthquake parameters include magnitude, epicentral distance and site parameters; inputting the ground motion parameters, the earthquake parameters and the building attributes into a real-time prediction model for the earthquake response of high-rise buildings to obtain the earthquake response output by the real-time prediction model for the earthquake response of high-rise buildings, and making an early warning according to the earthquake response; the real-time prediction model for the earthquake response of high-rise buildings is constructed based on a deep neural network; the real-time prediction model for the earthquake response of high-rise buildings includes multiple layers of fully connected hidden layers, and the number of neurons and the setting of activation functions in each layer of fully connected hidden layers are adjusted based on the characteristics of the earthquake response data corresponding to the high-rise building.

[0079] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. A person of ordinary skill in the art can understand and implement it without creative work.

[0080] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the essence of the above technical solution, or the part that contributes to the prior art, can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0081] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments or perform equivalent replacements for some of the technical features. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A real-time prediction method for earthquake response of high-rise buildings for earthquake early warning, characterized in that: include: Acquire strong vibration data and a target high-rise building model, and acquire building properties of the target high-rise building model; The target high-rise building model is constructed based on a high-rise building model with strong vibration monitoring data; Extracting ground motion parameters and earthquake parameters from the initial waveform of the strong motion data; the ground motion parameters include peak acceleration, peak velocity, cumulative absolute velocity, important duration, and Arias intensity; the earthquake parameters include magnitude, epicenter distance, and site parameters; Inputting the seismic parameters, the earthquake parameters and the building attributes into a real-time prediction model for earthquake response of a high-rise building, obtaining an earthquake response output by the real-time prediction model for earthquake response of a high-rise building, and issuing an early warning according to the earthquake response; The real-time prediction model for earthquake response of high-rise buildings is constructed based on a deep neural network; The real-time prediction model for earthquake response of high-rise buildings comprises multiple layers of fully connected hidden layers, and the number of neurons and the setting of activation functions of each fully connected hidden layer are adjusted based on the characteristics of earthquake response data corresponding to the high-rise buildings.

2. The real-time prediction method for earthquake response of high-rise buildings for earthquake early warning according to claim 1 is characterized in that: The training steps of the real-time prediction model for earthquake response of high-rise buildings include: Obtaining sample seismic parameters and sample earthquake parameters from a strong motion database, and obtaining sample building attributes, and labeled seismic responses corresponding to the sample seismic parameters, the sample earthquake parameters, and the sample building attributes; Obtain the initial real-time prediction model of high-rise building earthquake response; Inputting the sample seismic motion parameters, the sample earthquake parameters and the sample building attributes into the initial high-rise building seismic response real-time prediction model to obtain the predicted seismic response output by the initial high-rise building seismic response real-time prediction model; Based on the difference between the predicted seismic response and the labeled seismic response, a target loss is determined, and based on the target loss, parameter iteration is performed on the initial high-rise building seismic response real-time prediction model to obtain the high-rise building seismic response real-time prediction model.

3. The real-time prediction method for earthquake response of high-rise buildings for earthquake early warning according to claim 2 is characterized in that: The step of obtaining the label seismic response comprises: Based on the urban seismic elastoplastic analysis method, the labeled seismic response of the target high-rise building under the action of ground motion is determined.

4. The real-time prediction method for earthquake response of high-rise buildings for earthquake early warning according to claim 3 is characterized in that: The method of determining the labeled seismic response of the target high-rise building model under the action of ground motion based on the urban seismic elastoplastic analysis method includes: Based on a multi-degree-of-freedom shear layer model, the floor mass of the target high-rise building model is concentrated at a single mass point, and a trilinear skeleton line and a single-parameter hysteresis model are used to simulate the nonlinear behavior between floors in the target high-rise building model; The skeleton line parameters of the trilinear skeleton line are calibrated by a calibration method based on the HAZUS capability curve database.

5. The real-time prediction method for earthquake response of high-rise buildings for earthquake early warning according to claim 2 is characterized in that: The strong motion database is constructed based on seismic data with magnitudes between 4 and 8, epicenter distances less than 200 km, and site parameters between 100 and 900 m / s.

6. The real-time prediction method for earthquake response of high-rise buildings for earthquake early warning according to any one of claims 1 to 5, characterized in that: The extracting of ground motion parameters and earthquake parameters from the initial earthquake waveform of the strong motion data comprises: Using a P-wave automatic identification algorithm to detect the P-wave arrival time in the initial earthquake waveform of the strong vibration data; Based on the arrival time of the P wave, intercept the first 3 seconds of the earthquake record after the arrival time of the P wave; Based on the first 3 seconds of the seismic motion record, the seismic motion parameters and the earthquake parameters are determined.

7. A real-time prediction device for earthquake response of high-rise buildings for earthquake early warning, characterized in that: include: An acquisition unit, used to acquire strong vibration data and a target high-rise building model, and acquire building properties of the target high-rise building model; The target high-rise building model is constructed based on a high-rise building model with strong vibration monitoring data; An extraction unit, used to extract ground motion parameters and earthquake parameters from the initial waveform of the strong motion data; the ground motion parameters include peak acceleration, peak velocity, cumulative absolute velocity, important duration, and Arias intensity; the earthquake parameters include magnitude, epicenter distance, and site parameters; A prediction unit, used for inputting the seismic parameters, the earthquake parameters and the building attributes into a real-time prediction model for earthquake response of a high-rise building, obtaining an earthquake response output by the real-time prediction model for earthquake response of a high-rise building, and making an early warning according to the earthquake response; The real-time prediction model for earthquake response of high-rise buildings is constructed based on a deep neural network; The real-time prediction model for earthquake response of high-rise buildings comprises multiple layers of fully connected hidden layers, and the number of neurons and the setting of activation functions of each fully connected hidden layer are adjusted based on the characteristics of earthquake response data corresponding to the high-rise buildings.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the real-time prediction method for earthquake response of high-rise buildings for earthquake early warning as described in any one of claims 1 to 6 is implemented.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the real-time prediction method for earthquake response of high-rise buildings for earthquake early warning as described in any one of claims 1 to 6 is implemented.

Images