Real-time seismic oscillation numerical prediction method fusing fracture directivity effect

Abstract

The invention relates to the technical field of real-time earthquake disaster early warning of earthquake engineering, in particular to a real-time earthquake vibration numerical value prediction method fusing a fracture directivity effect. According to the technical scheme, filtering and baseline correction are carried out on real-time observation strong vibration waveforms of an earthquake network; the method comprises the steps that firstly, a ground seismic oscillation waveform is acquired, the ground seismic oscillation waveform is corrected to a bedrock site by using an empirical site effect correction model, then the corrected acceleration record is converted into an energy parameter, a gridding bedrock seismic oscillation energy field is generated through Kriging space interpolation, and a seismic oscillation energy space-time propagation process is simulated based on an RTT model and by using a Monte Carlo method. Furthermore, according to the technical scheme, the current fracture directivity effect parameter is inversed in real time and is applied to correction of a future prediction result, and it is ensured that the prediction result is more accurate.

Description

Technical Field

[0001] The present invention relates to the technical field of real-time earthquake disaster early warning in earthquake engineering, and in particular to a real-time earthquake motion numerical prediction method integrating rupture directional effects. Background Art

[0002] Numerical ground motion prediction, a real-time ground motion simulation technology, has been proposed as a next-generation earthquake early warning (EEW) solution. With the increasing demand for high-precision early warning of complex earthquakes with strong directional effects, reliably predicting directionally dependent ground motions in real time has become a critical issue. Traditional data assimilation-based numerical ground motion prediction methods primarily use optimal interpolation methods. Based on the Gaussian assumption, they dynamically fuse observed data with the background field using a gain matrix to correct the simulation results using the radiative transfer model (RTT). This simulation is effective for small and medium-sized earthquakes or point source models. However, due to the Gaussian assumption, when dealing with large earthquakes or line source models, the simulation results may not conform to the actual spatial distribution of ground motion. Although some methods have been proposed, such as those incorporating nonuniform attenuation models and three-dimensional radiative transfer models, they still do not consider the influence of the directional effects of source rupture on the spatial distribution of ground motion. It is widely recognized that destructive earthquakes exhibit significant directional effects. Specifically, this refers to the phenomenon in which the amplitude of strong ground motions does not completely follow the regular pattern of gradually decreasing with increasing distance from the epicenter during earthquake rupture propagation, resulting in uneven distribution of seismic wave energy in different directions. This effect significantly influences the intensity, frequency content, and duration of ground motions, and is particularly pronounced near faults. The directional effect is controlled and influenced by numerous factors, including the rupture mechanism of the earthquake source, the rupture direction and pattern of the fault, and the rupture velocity of the fault.

[0003] Analysis of rupture directional effects is typically based on ground motion prediction equations (GMPEs) based on point source models or finite fault models. This is quantified by comparing the difference between observed ground motion peaks (e.g., PGA and PGV) and empirically predicted GMPEs. The rupture directional coefficient (Cd) is then used for inversion fitting to obtain information on the direction, velocity, and pattern of the fault rupture. However, traditional inversion analysis of rupture directional effects is limited to post-earthquake rapid warnings and cannot provide a directional correction reference for real-time earthquake early warning technology. Summary of the Invention

[0004] (1) Technical problems solved

[0005] In view of the shortcomings of the existing technology, the present invention provides a real-time numerical prediction method for earthquake motion integrating the directional effect of rupture.

[0006] (2) Technical solution

[0007] To achieve the above-mentioned object, the present invention provides the following technical solution: A real-time numerical prediction method for earthquake motion incorporating the directional effect of fracture of the present invention comprises the following steps:

[0008] Preprocessing steps: filtering and baseline correction of real-time observation waveforms from the seismic network, and correcting ground motions to the bedrock site using a site effect correction model;

[0009] Energy field construction steps: convert the corrected acceleration records into JMA instrument intensity, and obtain the gridded bedrock ground motion energy field through Kriging interpolation;

[0010] Radiative transmission simulation steps: Based on the RTT model and Monte Carlo method, simulate the spatiotemporal propagation of the energy field to obtain the predicted seismic energy field for multiple time steps after time t0;

[0011] Repeat the preprocessing steps to obtain the real-time observed seismic field at (t0+Δt). Using the one-step prediction result at time t0 and the observation result at (t0+Δt), the double-sided rupture line source model is used to establish the inversion equation by nonlinearly minimizing the objective function, and the current rupture directional effect parameters are inverted in real time.

[0012] Repeat the radiation transfer simulation steps to obtain the multi-step prediction results at time (t0+Δt);

[0013] Directional inversion step: Utilizing the difference between the previous prediction field and the current observation field, the directional parameters are inverted through the double-sided rupture line source model. The multi-step prediction results at time (t0+Δt) of the repeated radiation transfer simulation steps are corrected, and the corrected prediction results are used as the real-time ground motion intensity.

[0014] Prediction correction step: Use the directivity coefficient obtained by inversion to correct the multi-step prediction result at time (t0+Δt) of repeatedly executing the radiation transfer simulation step, and use the corrected prediction result as the real-time seismic intensity to achieve real-time warning.

[0015] Preferably, in the preprocessing step, the site effect correction model is: F S，B =ln(F lin )+ln(F nl );

[0016] Among them, F S，B F represents the site amplification effect in natural logarithmic units; lin represents the linear component of site amplification; F nl Represents the nonlinear component of site amplification.

[0017] Further preferably, in the energy field construction step, the JMA intensity calculation formula is:

[0018] The acceleration vector amplitude after frequency-dependent weight filtering is defined as A(t), A c Defined as satisfying A(t)>A c The duration is 0.3s.

[0019] Again preferably, in the radiation transfer simulation step, the two-dimensional energy density evolution equation is:

[0020]

[0021] Where f(x, t: θ) represents the energy density in the direction of θ at the spatial position x at time t, V0 is the energy propagation speed, h0 is the intrinsic absorption coefficient of the medium during wave propagation, and g0 is the scattering coefficient of the wave propagation direction transformed from to.

[0022] Preferably, in the directional inversion step, the directional parameter is:

[0023]

[0024] Among them, V r / V S is the Mach number (V r is the rupture velocity, V S is the shear wave velocity in the earthquake source area), is the angle between the ray leaving the earthquake source and the direction of rupture propagation, k represents the proportion of rupture on the sub-fault in one direction, and if k>0.5, the direction is the main direction of rupture.

[0025] Further preferably, in the preprocessing step, the linear component F of the site model lin Describes the linear soil response under V s30 Scaling of ground motion as follows:

[0026]

[0027] Where c describes the V s30 Scaling, V c is the limit speed, V ref The magnification factor is 1 (V s30 for bedrock sites with a velocity of 760 m / s);

[0028] The nonlinear term F of the site model nl The linear field amplification has been modified to reduce the amplification of strong vibration levels. The function form of Fnl is as follows:

[0029]

[0030] Among them, f1, f2 and f3 are model coefficients, PGA rThe median peak horizontal acceleration of the reference rock layer is Vs30 = 760 m / s, and the coefficient f3 is set to 0.1 g;

[0031] Where the coefficient f2 is the period and V s30 The function is as follows:

[0032] f2=f4[exp{f5(min(V S30 ,760)-360)}-exp{f5(760-360)}];

[0033] Where f4 and f5 are periodic correlation coefficients.

[0034] Again preferably, the radiation transfer simulation step is difficult to estimate the energy propagation direction in practical applications, and the energy density distribution F is estimated without the propagation direction θ. int (x, t), F int (x, t) is calculated from f(x, t: θ):

[0035] Among them, F int (x, t) is the energy density at the unit grid point x at time t.

[0036] Preferably, the energy density F in the RTT model of the radiation transfer simulation step is proportional to the square of the earthquake amplitude, and the relationship between F and the JMA instrument intensity can be expressed as:

[0037] Among them, I JMMA (x, t) represents the JMA intensity at time t at the unit grid point x, and C is a coefficient independent of x and t.

[0038] Further preferably, the minimization objective function in the directional inversion step is expressed as:

[0039]

[0040] in, The energy density is predicted for the step at time t0 in the radiative transfer simulation step, is the observed energy density at time (t0+Δt), M is the number of strong motion stations, is a random variable used to account for the uncertainty of source mislocation, terrain effects, and radiation patterns in Monte Carlo simulations of energy propagation.

[0041] Again preferably, in the prediction correction step, the multi-step prediction result at time (t0+Δt) of repeatedly executing the radiation transfer simulation is corrected using the directional inversion step and can be expressed as:

[0042]

[0043] in The multi-step predicted energy density for the unit grid is, is the directivity coefficient of the unit grid with respect to the azimuth angle.

[0044] (3) Beneficial effects

[0045] Compared with the existing technology, the present invention provides a real-time numerical prediction method for ground motion that integrates the directional effect of rupture, which has the following beneficial effects:

[0046] This technical solution combines static kriging interpolation techniques with radiative transfer theory (RTT) to accurately predict ground motion fields and provides a new technical approach for earthquake early warning systems. This method is particularly suitable for large earthquakes or linear source models with significant rupture directionality, overcoming the limitations of traditional methods under the Gaussian assumption.

[0047] Improve forecast accuracy

[0048] Breaking through the limitations of the Gaussian assumption: The static Kriging interpolation method is used to avoid the influence of the Gaussian assumption, effectively quantify the spatial variability of the earthquake intensity field, and improve prediction accuracy. The intensity prediction accuracy within 20 seconds in advance is improved by more than 15%.

[0049] Real-time inversion of rupture directional effect: The current rupture directional effect parameters are inverted in real time through the double-sided rupture line source model and applied to the correction of future prediction results, making the prediction results more consistent with the actual situation. The real-time inversion results are highly consistent with the post-earthquake inversion results, which can provide a reference for post-earthquake rapid reporting.

[0050] Realize real-time requirements

[0051] Efficient computing: Based on the RTT model and utilizing the Monte Carlo method, this invention can rapidly simulate the spatiotemporal propagation of ground motion energy on a gridded basis, meeting the needs of real-time early warning. Each simulation step takes approximately 0.12 seconds, fully meeting the real-time requirements of earthquake early warning systems (EEW).

[0052] Optimize inversion efficiency: Using the nonlinear least squares inversion method, the inversion time does not exceed 0.2 seconds, ensuring the response speed of the system.

[0053] Enhanced user experience and reliability

[0054] Improved site effect correction model: The site effect correction model of this technical solution includes Vs30 scaling and nonlinear components, which can more accurately reflect the seismic characteristics under different geological conditions and reduce errors.

[0055] Improve the accuracy of instrument intensity calculations: use a 0.3-second sliding window to match seismic duration, and use frequency weighting to focus on the engineering frequency band to reduce high-frequency noise interference.

[0056] Adapting to complex environments

[0057] Multi-scenario applicability: This technical solution uses pre-earthquake exploration data to consider the scaling effect of soil layers on ground motion. It is not limited to standard bedrock soil layers and can be extended to any soil layer conditions.

[0058] Timely response capability: After being turned on, this technical solution can continuously monitor seismic activity without the need for startup operations. It can adapt to the switching between earthquake-free and earthquake-prone scenarios at any time, and can meet the timely release requirements of the EEW system.

[0059] In summary, this technical solution solves many technical bottlenecks in traditional seismic motion numerical prediction methods through a series of innovative technical solutions, providing users with a more accurate, real-time and reliable earthquake early warning solution with broad application prospects and significant social benefits. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 This is a schematic diagram of the prediction process framework of the present invention;

[0061] Figure 2 This is a schematic diagram of the real-time intensity field prediction of the Mw7.0 Kumamoto earthquake in Japan in the reference embodiment of the present invention;

[0062] Figure 3 Schematic diagram comparing the observed intensity and predicted intensity of the Mw7.0 Kumamoto earthquake in Japan in a reference embodiment of the present invention;

[0063] Figure 4 This is a schematic diagram of the real-time main rupture directional inversion of the Kumamoto earthquake source in the reference embodiment of the present invention. DETAILED DESCRIPTION

[0064] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0065] See also Figure 1-4 The present invention provides a real-time numerical prediction method for earthquake motion integrating the rupture directivity effect, comprising the following steps:

[0066] Preprocessing steps: filtering and baseline correction of real-time observation waveforms from the seismic network, and correcting ground motions to the bedrock site using a site effect correction model;

[0067] Energy field construction steps: convert the corrected acceleration records into JMA instrument intensity, and obtain the gridded bedrock ground motion energy field through Kriging interpolation;

[0068] Radiative transmission simulation steps: Based on the RTT model and Monte Carlo method, simulate the spatiotemporal propagation of the energy field to obtain the predicted seismic energy field for multiple time steps after time t0;

[0069] Repeat the preprocessing steps to obtain the real-time observed seismic field at (t0+Δt), and use the one-step prediction result at time t0 and the observation result at (t0+Δt);

[0070] Repeat the radiation transfer simulation steps to obtain the multi-step prediction results at time (t0+Δt);

[0071] Directional inversion steps: Utilize the difference between the previous prediction field and the current observation field, invert the directional parameters through the double-sided rupture line source model, establish the inversion equation by minimizing the objective function, and invert the current rupture directional effect parameters in real time;

[0072] Prediction correction step: Use the directivity coefficient obtained by inversion to correct the multi-step prediction result at time (t0+Δt) of repeatedly executing the radiation transfer simulation step, and use the corrected prediction result as the real-time seismic intensity to achieve real-time warning.

[0073] This technical solution achieves accurate prediction of seismic motion fields by combining static Kriging interpolation technology and radiative transfer theory (RTT), and provides a new technical means for earthquake early warning systems. This method is particularly suitable for processing large earthquakes or line source models with significant rupture directional effects, and solves the limitations of traditional methods under the Gaussian assumption.

[0074] Principle of preprocessing step

[0075] The noise of the observed waveform is eliminated by filtering and baseline correction, and the ground seismic motion is converted to the bedrock site using the site effect correction model. The model expression is: F S，B =ln(F lin )+ln(F nl );

[0076] Among them, F S，B F represents the site amplification effect in natural logarithmic units; lin represents the linear component of site amplification; F nl represents the nonlinear component of site amplification;

[0077] Linear component F lin : Quantify the linear amplification of ground motion by site shear wave velocity, the formula is:

[0078]

[0079] Where c describes the V s30 Scaling, V c is the limit speed, V ref For the site condition with a magnification of 1, V s30 760m / s;

[0080] Nonlinear component F nl : Correct the nonlinear effect of soil under strong earthquake, the formula is:

[0081]

[0082] Among them, f1, f2 and f3 are model coefficients, PGA r V is the median value of the peak horizontal acceleration of the reference rock layer s30 =760m / s bedrock, the coefficient f3 is set to 0.1g;

[0083] Where the coefficient f2 is the period and V s30 The function is as follows:

[0084] f2=f4[exp{f5(min(V S30 ,760)-360)}-exp{f5(760-360)}];

[0085] Where f4 and f5 are periodic correlation coefficients.

[0086] Energy field construction principle

[0087] The corrected acceleration is converted to JMA intensity using the following formula:

[0088] The acceleration vector amplitude after frequency-dependent weight filtering is defined as A(t), A c Defined as satisfying A(t)>A c The duration is 0.3s;

[0089] Kriging interpolation was used to generate a 3km×3km gridded energy field, and spatial autocorrelation was used to estimate the data of unmeasured points to improve the continuity of the field distribution.

[0090] Principles of Radiative Transfer Simulation

[0091] Based on the RTT model to simulate energy propagation, the two-dimensional energy density evolution equation is:

[0092]

[0093] Where f(x, t: θ) represents the energy density in the direction θ at the spatial position x at time t, V0 is the energy propagation velocity, h0 is the intrinsic absorption coefficient of the medium during wave propagation, and g0 is the scattering coefficient of the wave propagation direction transformed from to;

[0094] Monte Carlo simulation: through 10 6 Each particle traces the energy path, taking into account medium absorption (h0) and scattering (g0).

[0095] The relationship between energy density and JMA intensity is:

[0096] Among them, I JMA (x, t) represents the JMA intensity at time t at the unit grid point x, and C is a coefficient that is independent of x and t;

[0097] Simplified estimation: When there is no propagation direction, the energy density can be quickly estimated using the following formula:

[0098]

[0099] Among them, F int (x, t) is the energy density at the unit grid point x at time t.

[0100] The preprocessing steps are repeated to obtain the real-time observed seismic field at (t0+Δt). The one-step prediction result at time t0 and the observation result at (t0+Δt) are used. The double-sided rupture line source model is used to establish the inversion equation through the nonlinear minimization objective function to invert the current rupture directional effect parameters in real time.

[0101] Directional inversion principle

[0102] The objective function minimized in repeated preprocessing steps is expressed as:

[0103]

[0104] in, The energy density is predicted for the step at time t0 in the radiative transfer simulation step, is the observed energy density at time (t0+Δt), M is the number of strong motion stations, is a random variable used to account for the uncertainty of source misalignment, terrain effects, and radiation patterns in Monte Carlo simulations of energy propagation;

[0105] Repeat the radiation transfer simulation steps to obtain the multi-step prediction results at time (t0+Δt);

[0106] By using the difference between the previous prediction field and the current observation field, the directional parameters are inverted by the double-sided rupture line source model. The multi-step prediction results at time (t0+Δt) of the repeated radiation transfer simulation steps are corrected. The directional parameters inverted by the double-sided rupture line source model are:

[0107]

[0108] Among them, V r / V S is the Mach number (V r is the rupture velocity, V S is the shear wave velocity in the earthquake source area), is the angle between the ray leaving the earthquake source and the direction of rupture propagation, k represents the proportion of rupture on the sub-fault in one direction, and if k>0.5, the direction is the main direction of rupture.

[0109] Prediction Correction Principle

[0110] The multi-step prediction result at time (t0+Δt) in the repeated execution of the radiative transfer simulation step in the directional inversion step can be expressed as:

[0111]

[0112] in The multi-step predicted energy density for the unit grid is, is the directivity coefficient of the unit grid with respect to the azimuth angle.

[0113] Working principle of the optimal technical solution

[0114] Site Effect Correction Optimization

[0115] Two-component decoupling: the linear component describes the scaling of soil under small strains, and the nonlinear component dynamically attenuates strong earthquake amplification;

[0116] Real-time site effect correction: The site model is combined with the measured PGA to dynamically adjust the model parameters and construct a frequency domain scaling curve to achieve real-time correction of recorded ground motions after an earthquake occurs; JMA intensity is transmitted in real time

[0117] Time window optimization: A 0.3s sliding time window is used to match the seismic motion history, obtaining the earthquake intensity field distribution in real time and realizing the dynamic evolution of the seismic motion field, eliminating the static limitation of using only the seismic motion peak value;

[0118] Frequency weighting: 0.1-10Hz bandpass filtering focuses on the engineering frequency band to reduce high-frequency noise interference.

[0119] RTT simulation acceleration

[0120] Particle sparsification: Allocate particle distribution according to the real-time unit particle capacity, reduce the density in the far field and increase the density in the near field, thereby increasing the near-field simulation accuracy;

[0121] GPU parallel computing: Single-step simulation takes ≤ 0.12s, meeting real-time requirements.

[0122] Directional inversion acceleration

[0123] Nonlinear least squares inversion method, inversion time ≤ 0.2s;

[0124] Prior constraints: the statistical value of rupture velocity (2.0-3.5 km / s), the rupture segment ratio (0.5), and the main rupture direction are the direction with the strongest directional effect.

[0125] Reference Examples

[0126] Taking the Kumamoto earthquake in Japan on April 16, 2016 as an example, a real-time ground motion numerical prediction method integrating the rupture directionality effect was used to simulate earthquake early warning. Figure 2 The JMA intensity field is observed at real-time stations after an earthquake, and the intensity fields predicted 5, 10, and 20 seconds afterward. The left column shows the observed intensity field, and the right three columns show the intensity predictions 5, 10, and 20 seconds afterward. The simulation model is set as a bedrock soil layer. The two-dimensional spatial model is gridded with 3 km × 3 km rectangular cells. The total number of Monte Carlo simulation particles is 10^6, and the simulation unit time step is 1 second. Data from the K-net and KiK-net networks of the Japan Institute for Disaster Prevention Science and Technology are used, including three-component acceleration records from 217 strong earthquake stations. The initial filtering range is 0.1–10 Hz, and baseline correction is performed before filtering. To meet the real-time requirements of the EEW system, a nonlinear least squares method is used to improve inversion efficiency when inverting the directional effect parameters.

[0127] Compared with the traditional optimal interpolation method, this technical solution predicts the instrumental seismic intensity values ​​of all stations. Figure 3 The intensity time history curves for three stations at different distances from the fault in this simulation are shown. The solid black line represents the measured intensity, while the solid gray line and dashed gray line represent the 5-second predicted earthquake intensity obtained using the proposed method and the traditional optimal interpolation method, respectively. For ease of comparison, the predicted intensity curves are offset by 5 seconds to coincide with the observed intensity curves. It can be seen that the traditional optimal interpolation method overestimates and underestimates the intensity curves for stations in the near-field rupture direction and perpendicular to the fault, respectively. However, the numerical ground motion prediction method, which incorporates the rupture directionality effect, produces an instrumental intensity curve that is more consistent with the observed intensity curve. Figure 4The proposed method provides a simultaneous inversion of the real-time primary rupture direction. Furthermore, the simulation time per unit step of this technical solution is approximately 0.12 seconds, fully meeting the real-time requirements of the EEW system. Therefore, this technical solution has been validated through numerical simulation experiments to validate the effectiveness of numerical ground motion prediction that incorporates the effects of rupture directionality.

[0128] Detailed workflow

[0129] Data preprocessing

[0130] 0.1-10Hz filtering and polynomial baseline correction;

[0131] Site effect correction to V s30 =760m / s bedrock, obtain bedrock acceleration.

[0132] Energy field generation

[0133] Calculate JMA intensity (T = 0.3s time window);

[0134] Kriging interpolation generates a 3km×3km grid energy field.

[0135] First radiative transfer simulation

[0136] Monte Carlo simulation (10 6 Particles) predict the energy field 5s, 10s, and 20s after time t0.

[0137] Real-time data iteration

[0138] Obtain (t0+Δt) observation data, repeat preprocessing and energy field construction;

[0139] Compare the t0 prediction with the (t0+Δt) observation and invert the directional parameter Cd.

[0140] Forecast revision and output

[0141] Use Cd to correct the (t0+Δt) multi-step prediction results;

[0142] Send the revised intensity forecast, with a single iteration taking ≤ 0.12s.

[0143] Loop Iteration

[0144] Observation data is continuously updated to dynamically track the directional evolution of the rupture until the earthquake ends.

[0145] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A real-time numerical prediction method for earthquake motion integrating rupture directionality effect, characterized in that: The following steps are involved: Preprocessing steps: filtering and baseline correction of real-time observation waveforms from the seismic network, and correcting ground motions to the bedrock site using a site effect correction model; Energy field construction steps: convert the corrected acceleration records into JMA instrument intensity, and obtain the gridded bedrock ground motion energy field through Kriging interpolation; Radiative transmission simulation steps: Based on the RTT model and Monte Carlo method, simulate the spatiotemporal propagation of the energy field to obtain the predicted seismic energy field for multiple time steps after time t0; Repeat the preprocessing steps to obtain the real-time observed seismic field at (t0+Δt). Using the one-step prediction result at time t0 and the observation result at (t0+Δt), the double-sided rupture line source model is used to establish the inversion equation by nonlinearly minimizing the objective function, and the current rupture directional effect parameters are inverted in real time. Repeat the radiation transfer simulation steps to obtain the multi-step prediction results at time (t0+Δt); Directional inversion step: using the difference between the previous predicted field and the current observed field, the directional parameters are inverted through the double-sided rupture line source model; Prediction correction step: Use the directivity coefficient obtained by inversion to correct the multi-step prediction result at time (t0+Δt) of repeatedly executing the radiation transfer simulation step, and use the corrected prediction result as the real-time seismic intensity to achieve real-time warning.

2. A real-time numerical prediction method for earthquake motion integrating fracture directionality effect according to claim 1, characterized in that: In the preprocessing step, the site effect correction model is: F S，B =ln(F lin )+ln(F nl ); Among them, F S，B F represents the site amplification effect in natural logarithmic units; lin represents the linear component of site amplification; F nl Represents the nonlinear component of site amplification.

3. The real-time numerical prediction method for earthquake motion integrating fracture directionality effect according to claim 1 is characterized in that: In the energy field construction step, the JMA intensity calculation formula is: The acceleration vector amplitude after frequency-dependent weight filtering is defined as A(t), A c Defined as satisfying A(t)>A c The duration is 0.3s.

4. The real-time numerical prediction method for earthquake motion integrating fracture directionality effect according to claim 1 is characterized in that: In the radiation transfer simulation step, the two-dimensional energy density evolution equation is: Where f(x,t:θ) represents the energy density in the direction of θ at the spatial position x at time t, V0 is the energy propagation speed, h0 is the intrinsic absorption coefficient of the medium during wave propagation, and g0 is the scattering coefficient of the wave propagation direction transformed from to.

5. The real-time numerical prediction method for earthquake motion integrating fracture directionality effect according to claim 1 is characterized in that: In the directional inversion step, the directional parameters are: Among them, V r / V S is the Mach number (V r is the rupture velocity, V S is the shear wave velocity in the source area), θ is the angle between the ray leaving the source and the direction of rupture propagation, and k represents the proportion of rupture on the sub-fault in one direction. If k>0.5, this direction is the main direction of rupture.

6. The real-time numerical prediction method for earthquake motion integrating the directional effect of fracture according to claim 2, characterized in that: In the preprocessing step, the linear component F of the site model lin Describes the linear soil response under V s30 Scaling of ground motion as follows: Where c describes the V s30 Scaling, V c is the limit speed, V ref The magnification factor is 1 (V s30 for bedrock sites with a velocity of 760 m / s); The nonlinear term F of the site model n1 The linear field amplification has been modified to reduce the amplification of strong vibration levels. The function form of Fnl is as follows: Among them, f1, f2 and f3 are model coefficients, PGA r The median peak horizontal acceleration of the reference rock layer is Vs30 = 760 m / s, and the coefficient f3 is set to 0.1 g; Where the coefficient f2 is the period and V s30 The function is as follows: f2=f4[exp{f5(min(V S30 ,760)-360)}-exp{f5(760-360)}]; Among them, f4 and f5 are periodic correlation coefficients.

7. The method for real-time numerical prediction of earthquake motion integrating fracture directionality effect according to claim 4, characterized in that: The radiation transfer simulation step is difficult to estimate the energy propagation direction in practical applications. In the absence of the propagation direction θ, the energy density distribution F is estimated. int (x, t), F int (x, t) is calculated from f(x, t: θ): Among them, F int (x, t) is the energy density at the unit grid point x at time t.

8. The real-time numerical prediction method for earthquake motion incorporating the directional effect of fracture according to claim 4 is characterized in that: The energy density F in the RTT model of the radiation transfer simulation step is proportional to the square of the ground motion amplitude. The relationship between F and the JMA instrument intensity can be expressed as follows: Among them, I JMA (x, t) represents the JMA intensity at time t at the unit grid point x, and C is a coefficient independent of x and t.

9. The real-time numerical prediction method for earthquake motion integrating fracture directionality effect according to claim 1, characterized in that: The minimization objective function in the directional inversion step is expressed as: in, The energy density is predicted for the step at time t0 in the radiative transfer simulation step, is the observed energy density at time (t0+Δt), M is the number of strong motion stations, is a random variable used to account for the uncertainty of source mislocation, terrain effects, and radiation patterns in Monte Carlo simulations of energy propagation.

10. The real-time numerical prediction method for earthquake motion integrating fracture directionality effect according to claim 1, characterized in that: In the prediction correction step, the multi-step prediction result of the repeated execution of the radiation transfer simulation at time (t0+Δt) using the directivity coefficient can be expressed as: in The multi-step predicted energy density for the unit grid is, is the directivity coefficient of the unit grid with respect to the azimuth angle.

Images