^thlayer, respectively; A_P^jand A_SV^jrepresent incident displacement amplitudes of inhomogeneous P-waves and SV-waves in the j^thlayer; B_P^jand B_SV^jrepresent reflected displacement amplitudes of inhomogeneous P-waves and SV-waves in the j^thlayer; V_R* is a complex Rayleigh velocity considering the damping effect; l_x, m_X, {tilde over (s)}, and {tilde over (t)} are parameters related to the complex Rayleigh-wave velocity, complex P-wave velocity, and complex S-wave velocity in the j^thlayer; λ* and μ* are complex Lame constants of the j^thsoil layer; x and z are space coordinates; and i is an imaginary unit;

1.5. Calculating the free-field velocities {dot over (U)}_Y^jand free-field stresses τ_XY^jinduced by Love waves with the frequency-domain dynamic stiffness matrix method:

$\begin{matrix} U_{Y}^{j} (x, z, ω) = A_{S H}^{j} e^{i ω \tilde{t} z / V_{L}^{*} - i ω x / V_{L}^{*}} + B_{SH}^{j} e^{- i ω \tilde{t} z / V_{L}^{*} - i ω x / V_{L}^{*}} & (5) \end{matrix}$

$\begin{matrix} {\dot{U}}_{Y}^{j} (x, z, ω) = i ω U_{Y}^{j} (x, z, ω) & (6) \end{matrix}$

$\begin{matrix} τ_{XY}^{j} (x, z, ω) = μ^{*} \frac{\partial U_{Y}^{j} (x, z, ω)}{\partial x} & (7) \end{matrix}$

- - where U_Y^jis a free-field horizontal displacement induced by Love waves in the j^thlayer; A_SH^jrepresents an incident displacement amplitude of inhomogeneous SH-waves in the j^thlayer; B_SH^jrepresents a reflected displacement amplitude of inhomogeneous SH-waves in the j^thlayer; and V_L* is a complex Love velocity considering the damping effect;
- 1.6. Using Equations (3) and (4) to calculate Rayleigh velocity mode shapes and stress mode shapes according to corresponding Rayleigh displacement mode shapes, respectively; and using Equations (6) and (7) to calculate Love velocity mode shapes and stress mode shapes according to corresponding Love displacement mode shapes, respectively;
- 1.7. Extracting vector elements corresponding to the ground surface from the Rayleigh/Love velocity and stress mode shapes of each dispersive mode, and substituting the extracted vector elements into Equation (1) to calculate the average energy flux densities P_R,m¹and P_L,m¹for Rayleigh waves and Love waves, respectively:

$\begin{matrix} {\overline{P}}_{R, m}^{1} (ω) = \frac{\begin{matrix} Re [σ_{X, m}^{1} (x = 0, z = 0, ω)] Re [{\dot{U}}_{X, m}^{1} (x = 0, z = 0, ω)] + \\ IM [σ_{X, m}^{1} (x = 0, z = 0, ω)] IM [{\dot{U}}_{X, m}^{1} (x = 0, z = 0, ω)] \end{matrix}}{(2)} & (8) \end{matrix}$

$\begin{matrix} {\overline{P}}_{L, m}^{1} (ω) = \frac{\begin{matrix} Re [τ_{XY, m}^{1} (x = 0, z = 0, ω)] Re [{\dot{U}}_{Y, m}^{1} (x = 0, z = 0, ω)] + \\ IM [τ_{XY, m}^{1} (x = 0, z = 0, ω)] IM [{\dot{U}}_{Y, m}^{1} (x = 0, z = 0, ω)] \end{matrix}}{(2)} & (9) \end{matrix}$

- - where subscript m represents the number of dispersive modes of Rayleigh or Love waves; and Re and IM are symbols of the real part and imaginary part, respectively;
- 1.8. Adopting the average energy flux density of surface waves shown in step 1.7 as an index to characterize the modal participation quantity, and defining modal participation factors with the relative ratios between modal participation quantities:

$\begin{matrix} N_{R, m} (ω) = \frac{{\overline{P}}_{R, m}^{1} (ω)}{\sum_{m = 1}^{m = n} {\overline{P}}_{R, m}^{1} (ω)} (ω_{nn - 1} \leq ω \leq ω_{n n}) & (10) \end{matrix}$

$\begin{matrix} N_{L, m} (ω) = \frac{{\overline{P}}_{L, m}^{1} (ω)}{\sum_{m = 1}^{m = n} {\overline{P}}_{L, m}^{1} (ω)} (ω_{nn - 1} \leq ω \leq ω_{n n}) & (11) \end{matrix}$

- - where N_R,m(ω) and N_L,m(ω) are modal participation factors of any m^thmode of Rayleigh waves and Love waves corresponding to the frequency ω, respectively; n is the total number of participation modes; and (ω_nn-1, ω_nn) represents the frequency range composed of initial frequencies of two adjacent dispersive modes nn−1 and nn in the dispersion curves;
- Step 2: assigning ground Rayleigh components to each dispersive mode based on modal participation factors of Rayleigh waves: implementing an inversion algorithm to each single-mode Rayleigh component to obtain respective single-mode free fields; and superimposing all single-mode free fields to form the multi-mode Rayleigh-wave free fields based on the mode superposition principle, with a main process as follows:
- 2.1. transferring ground Rayleigh components that possess dispersion properties from the time domain to the frequency domain, and decomposing the total Rayleigh components into individual harmonic components using the FFT technology:
- 2.2. Multiplying frequency-domain horizontal displacement of each ground Rayleigh harmonic component U_X,R(ω) by a corresponding modal participation factor N_R,m(ω) to calculate the ground Rayleigh horizontal displacement assigned to each dispersive mode: U_X,m¹(ω)=U_X,R(ω)N_R,m(ω);
- 2.3. Calculating a ratio K of the first vector element in the Rayleigh horizontal displacement mode shapes and vertical displacement mode shapes for each dispersive mode, and multiplying the ratio K by U_X,m¹(ω) to calculate the ground Rayleigh vertical displacement assigned to each dispersive mode: U_Z,m¹(ω)=−iKU_X,m¹(ω);
- 2.4. Normalizing the displacement mode shapes of Rayleigh waves, and inverting Rayleigh displacements of each dispersive mode at each layer interface according to corresponding ground Rayleigh displacements:

$\begin{matrix} U_{X, m}^{j} (ω) = \frac{ϕ_{X, m}^{j} (ω)}{ϕ_{X, m}^{1} (ω)} U_{X, m}^{1} (ω) & (12) \end{matrix}$

$\begin{matrix} U_{Z, m}^{j} (ω) = \frac{ϕ_{Z, m}^{j} (ω)}{ϕ_{Z, m}^{1} (ω)} U_{Z, m}^{1} (ω) & (13) \end{matrix}$

- - where U_X,m^jis a Rayleigh horizontal displacement of the m^thmode at the j^thlayer interface; ϕ_X,m¹and ϕ_X,m^jrepresent vector element values corresponding to the m^thmode at the ground surface and the j^thlayer interface within the horizontal displacement mode shapes, respectively; U_Z,m^jis a Rayleigh vertical displacement of the m^thmode at the j^thlayer interface; and ϕ_Z,m¹and ϕ_Z,m^jrepresent vector element values corresponding to the m^thmode at the ground surface and the j^thlayer interface within the vertical displacement mode shapes, respectively;
- 2.5. Substituting obtained Rayleigh displacements of each dispersive mode at each layer interface into the local dynamic stiffness matrix of Rayleigh waves to calculate A_P, B_P, A_SV, and B_SVwithin each layer, and carrying out back substitution to Equations (2)-(4) to calculate Rayleigh-wave induced free-field displacements, free-field velocities, and free-field stresses at any coordinate; thereby constructing single-mode Rayleigh-wave free fields;
- 2.6. Superimposing all single-mode Rayleigh-wave free fields to form the multi-mode Rayleigh-wave free fields based on the mode superposition principle.
- Step 3: assigning ground Love components to each dispersive mode based on modal participation factors of Love waves; implementing an inversion algorithm to each single-mode Love component to obtain respective single-mode free fields; and superimposing all single-mode free fields to form the multi-mode Love-wave free fields based on the mode superposition principle, with a main process as follows:
- 3.1. Transferring ground Love components that possess dispersion properties from the time domain to the frequency domain, and decomposing the total Love components into individual harmonic components using FFT technology;
- 3.2. Multiplying frequency-domain displacement of each ground Love harmonic component U_Y,L(ω) by a corresponding modal participation factor N_L,m(ω) to calculate the ground Love displacement assigned to each dispersive mode: U_Y,m¹(ω)=U_Y,L(ω)N_L,m(ω);
- 3.3. Normalizing the displacement mode shapes of Love waves, and inverting Love displacement of each dispersive mode at each layer interface according to the corresponding ground Love displacement:

$\begin{matrix} U_{Y, m}^{j} (ω) = \frac{ϕ_{Y, m}^{j} (ω)}{ϕ_{Y, m}^{1} (ω)} U_{Y, m}^{1} (ω) & (14) \end{matrix}$

- - where U_Y,m^jis a Love displacement of the m^thmode at the j^thlayer interface; and ϕY,m¹and ϕ_Y,m^jrepresent vector element values corresponding to the m^thmode at the ground surface and the j^thlayer interface within the displacement mode shapes, respectively:
- 3.4. Substituting obtained Love displacement of each dispersive mode at each layer interface into the local dynamic stiffness matrix of Love waves to calculate A_SHand B_SHwithin each layer, and carrying out back substitution to Equations (5)-(7) to calculate Love-wave induced free-field displacements, free-field velocities, and free-field stress at any coordinate: thereby constructing single-mode Love-wave free fields.
- 3.5. Superimposing all single-mode Love-wave free fields to form the multi-mode Love-wave free fields based on the mode superposition principle.

The present invention has the following beneficial effects:

- (1) The present invention provides a multi-mode surface-wave free-field inversion method based on dispersion properties of layered media, which fully considers the multi-mode dispersion properties of surface waves. It makes up for the deficiency of considering only fundamental mode in the existing inversion methods, and greatly improves the inversion accuracy. The comparative analysis results show that inverted multi-mode free fields in layered media of stiffness-increment type, hard-interbed type, and weak-interbed type follow the real wavefield characteristics. Based on this, the present invention can provide more accurate seismic excitation for seismic analysis of SSI systems, and has important engineering applications for revealing the real failure mechanism, and failure behavior of structures.
- (2) The method of the present invention introduces an index of average energy flux density to characterize the modal participation quantity of each dispersive mode of surface waves, and transforms the propagation kinematics patterns of surface waves into dynamics responses of soil particles. Furthermore, the modal participation factors are defined as the relative ratios between modal participation quantities. The present invention effectively solves the difficulty in quantitatively identifying each dispersive mode of surface waves.

Geological studies have shown that the calculated errors are very significant if considering only the fundamental mode of surface waves. Therefore, a joint influence of multiple modes should be highly considered in the inversion of surface waves in layered media, and quantitative identification of each mode is the key to achieving the multi-mode surface-wave free-field inversion. It can be seen from the dispersion curves that the fixed harmonic component of surface waves at higher frequencies is jointly composed of different dispersive modes. Among these, the fundamental mode starts at f=0 Hz, whereas higher modes start from a specific cutting frequency f_c. At a given frequency, the propagation velocities of surface waves increase with the number of modes. However, all dispersive modes are mixed no matter in the frequency domain or time domain, and it is difficult to identify and separate each mode quantitatively. This further results in an obstruction to accurately inverting multi-mode surface-wave free fields. The method of the present invention derives the modal participation factors of Rayleigh and Love waves by introducing the average energy flux density, thereby quantitatively determining the surface-wave components of each mode. The results of equation derivation are simple, and the calculation process is simple and efficient.

- (3) The method of the present invention utilizes the mode superposition theory from structural dynamics to invert single-mode surface-wave free fields firstly in the frequency domain, and then superimpose all single-mode results to construct the multi-mode surface-wave free fields. It effectively solves the problem of multi-mode surface-wave free-field joint inversion based on dispersion properties.

Eigenvector results of the dynamic stiffness matrix of surface waves show that the eigenvectors of different modes at the same frequency are orthogonal to each other. Based on the characteristics, the method of the present invention uses the mode superposition theory to define the multi-mode free fields as a linear superposition sum of all single-mode free fields in the frequency domain, which decomposes a complex multi-mode problem into various simple single-mode problems. The process has clear physical meaning, sufficient theoretical basis, and accurate and objective results.

DESCRIPTION OF DRAWINGS

The FIGURE is the flow chart of the method of the present invention.

DETAILED DESCRIPTION

The present invention will be further described below in combination with the drawings in the embodiment of the present invention.

The first step in the FIGURE is to calculate the eigenvalues and eigenvectors of the global dynamic stiffness matrix of surface waves according to site parameters of layered media and determine the dispersion curves and the dispersive mode shapes. Then, the average energy flux density of surface waves is defined based on the elements characterizing the ground surface within velocity and stress mode shape vectors to calculate the participation quantity of each dispersive mode. Accordingly, the participation factor of each mode is determined, which lays a foundation for subsequent inversion of single-mode surface-wave free fields.

Surface waves are essentially the waveform products of inhomogeneous body waves near the ground surface. Among others, Rayleigh waves are a type of two-dimensional surface waves generated by the superposition of inhomogeneous P-waves and inhomogeneous SV-waves at the layer interface and polarized in the P-SV plane. Love waves are a type of one-dimensional surface waves generated by the superposition of inhomogeneous SH waves at the layer interface and polarized perpendicular to the P-SV plane. The polarization directions of the two types of surface waves are perpendicular to each other and independent of each other. When establishing the coordinates for layer media, it is generally assumed that the X-Z plane is the P-SV plane where Rayleigh waves are polarized, while the Y-direction is the polarization direction of Love waves. The force-displacement relationship of soil particles caused by the Rayleigh waves can be established by the dynamic stiffness matrix, as shown in Equation (15):

$\begin{matrix} {\begin{matrix} F_{X}^{j} \\ {iF}_{Z}^{j} \\ F_{X}^{j + 1} \\ i F_{Z}^{j + 1} \end{matrix}} = {[K^{j}]}_{R} {\begin{matrix} U_{X}^{j} \\ i U_{Z}^{j} \\ U_{X}^{J + 1} \\ i U_{Z}^{j + 1} \end{matrix}} & (15) \end{matrix}$

where F_X^jand F_Z^j(and F_X^jand F_Z^j) are external load amplitudes of soil particles in the j^thlayer along X and Z directions at upper (and bottom) interfaces, respectively. U_X^jand U_Z^j(and U_X^j+1and U_Z^j+1) are corresponding displacement amplitudes of soil particles in the j^thlayer along X and Z directions at upper (and bottom) interfaces, respectively. [K^j/]_Ris the Rayleigh local dynamic stiffness matrix of the j^thlayer, and is related to soil layer parameters and the propagation velocity of Rayleigh waves V_R. In order to maintain the symmetry of the [K^j]_Rmatrix, the load and displacement amplitudes along the Z-direction are multiplied by the imaginary number i, respectively. For the underlying half-space of the layered media, the displacement and force at infinity of the bottom are zero, so that only the force-displacement dynamic relationship at the upper interface is considered, as shown in Equation (16):

$\begin{matrix} {\begin{matrix} F_{X}^{0} \\ i F_{Z}^{0} \end{matrix}} = {[K^{0}]}_{R} {\begin{matrix} U_{X}^{0} \\ i U_{Z}^{0} \end{matrix}} & (16) \end{matrix}$

where F_X⁰and F_Z⁰are external load amplitudes of soil particles at the upper interface of the underlying half-space along X and Z directions, respectively: U_X⁰and U_Z⁰are corresponding displacement amplitudes of soil particles at the upper interface of the underlying half-space along X and Z directions, respectively; and [K⁰]_Ris the Rayleigh local dynamic stiffness matrix of the underlying half-space. Similarly, the force-displacement relationships of soil particles in the layered media and the underlying half-space caused by Love waves are shown in Equations (17) and (18), respectively:

$\begin{matrix} {\begin{matrix} F_{Y}^{j} \\ F_{Y}^{j + 1} \end{matrix}} = {[K^{j}]}_{L} {\begin{matrix} U_{Y}^{j} \\ U_{Y}^{j + 1} \end{matrix}} & (17) \end{matrix}$

$\begin{matrix} F_{Y}^{0} = k_{L}^{0} U_{Y}^{0} & (18) \end{matrix}$

where F_Y^jand F_Y^j+1are external load amplitudes of soil particles in the j^thlayer along the Y-direction at the upper and bottom interfaces, respectively; U_Y^jand U_Y^j+1are corresponding displacement amplitudes of soil particles in the j^thlayer along the Y-direction at the upper and bottom interfaces, respectively; [K^j]_Lis the Love local dynamic stiffness matrix of the j^thlayer, and is related to soil layer parameters and the propagation velocity of Love waves V_L; F_Y⁰is the external load amplitude of soil particles along the Y-direction at the upper interface of the underlying half-space; U_Y⁰is the displacement amplitude of soil particles along the Y-direction at the upper interface of the underlying half-space; and k_L⁰is the Love local dynamic stiffness of the underlying half-space.

According to the continuity principle for interlayer loads and displacements, the local dynamic stiffness matrices of each layer and the underlying half-space are assembled to form the global dynamic stiffness matrix, and the global force-displacement equations of the layered media are shown in Equations (19) and (20), respectively:

[Image Omitted]

where the external force vectors on the left side of the equations are zero. To ensure that non-zero solutions exist for the displacements in the equations' right side, the determinant of the global dynamic stiffness matrix must be equal to zero. All discrete points on the dispersion curves of the Rayleigh and Love waves are composed of eigenvalue pairs (ω, V_R) and (ω, V_L) of the global dynamic stiffness matrices. For every frequency ω, there exist one or multiple solutions within V_R, which form multi-mode solution sets for dispersion curves. Additionally, the eigenvectors describing the vibration patterns of different displacement modes are mutually orthogonal. The key step for inverting multi-mode Rayleigh-wave free fields and multi-mode Love-wave free fields is to determine mode participation factors of multiple dispersive modes, thus assigning components of ground surface waves to each mode. The difficulty lies in selecting a reasonable index to characterize the participation quantity of each dispersive mode of surface waves. Such an index is supposed to be associated with two inversion conditions: the surface-wave components of the surface and the site parameters. According to the law of energy conservation, the carried energy during the propagation of surface waves is converted into mechanical energy of soil particles, which causes motions of soil particles: specifically, rocking motions associated with Rayleigh waves and torsion motions associated with Love waves. Surface waves of different dispersive modes carry varying amounts of energy, leading to distinct rocking and torsion motions. Hence, energy serves as a valuable index for characterizing the modal participation quantity within dispersive surface waves.

Average energy flux density, as one of the typical indicators to evaluate the seismic wave energy, represents the energy per unit of time through the unit area perpendicular to the direction of wave propagation, and is expressed as:

$\begin{matrix} {\overline{P}}_{k} = \frac{ω}{2 π} \int_{0}^{2 π / ω} T_{k l} (t) {\dot{U}}_{l} (t) d t & (21) \end{matrix}$

where T_kl(t) and {dot over (U)}_l(t) are tensor time histories of soil particles' stresses and velocities subjected to surface waves, respectively. The propagation of Rayleigh waves would induce free fields of horizontal velocities {dot over (U)}_X^jand vertical velocities {dot over (U)}_Z^jas well as corresponding normal stresses σ_X^jand shear stresses τ_XZ^jof soil particles in each layer. Therefore, the average energy flux density of Rayleigh waves in the j^thlayer can be defined as:

$\begin{matrix} {\bar{P}}_{R}^{j} (x, z, ω) = \frac{ω}{2 π} \int_{0}^{2 π / ω} {Re [σ_{X}^{j} (x, z, ω) \exp (i ω t)] Re [{\dot{U}}_{X}^{j} (x, z, ω) \exp (i ω t)] + Re [τ_{XZ}^{j} (x, z, ω) \exp (i ω t)] Re [{\dot{U}}_{Z}^{j} (x, z, ω) \exp (i ω t)]} dt & (22) \end{matrix}$

By introducing Euler's formula exp(iωt)=cos(ωt)+i sin(ωt), Equation (22) can be simplified as:

$\begin{matrix} {\overline{P}}_{R}^{j} (x, z, ω) = \frac{IP (σ_{X}^{j}, {\dot{U}}_{X}^{j}) + IP (τ_{XZ}^{j}, {\dot{U}}_{Z}^{j})}{2} & (23) \end{matrix}$

where IP is the inner product symbol, satisfying:

$\begin{matrix} {\begin{matrix} \begin{matrix} IP (σ_{X}^{j}, {\dot{U}}_{X}^{j}) = Re [σ_{X}^{j} (x, z, ω)] Re [{\dot{U}}_{X}^{j} (x, z, ω)] + \\ IM [σ_{X}^{j} (x, z, ω)] IM [{\dot{U}}_{X}^{j} (x, z, ω)] \end{matrix} \\ \begin{matrix} IP (τ_{XZ}^{j}, {\dot{U}}_{Z}^{j}) = Re [τ_{XZ}^{j} (x, z, ω)] Re [{\dot{U}}_{Z}^{j} (x, z, ω)] + \\ IM [τ_{XZ}^{j} (x, z, ω)] IM [{\dot{U}}_{Z}^{j} (x, z, ω)] \end{matrix} \end{matrix} & (24) \end{matrix}$

For Love waves, their propagation would induce free fields of horizontal velocities {dot over (U)}_Y^jand corresponding shear stresses τ_XY^jof soil particles in each layer. Consequently, the average energy flux density of Love waves in the j^thlayer is expressed as:

$\begin{matrix} {\overline{P}}_{L}^{j} (x, z, ω) = \frac{ω}{2 π} \int_{0}^{2 π / ω} Re [τ_{XY}^{j} (x, z, ω) \exp (i ω t)] Re [{\dot{U}}_{Y}^{j} (x, z, ω) \exp (i ω t)] dt & (25) \end{matrix}$

By introducing Euler's formula, Equation (25) can be further simplified as:

$\begin{matrix} {\overline{P}}_{L}^{j} (x, z, ω) = \frac{\begin{matrix} Re [τ_{XY}^{j} (x, z, ω)] Re [{\dot{U}}_{Y}^{j} (x, z, ω)] + \\ IM [τ_{XY}^{j} (x, z, ω)] IM [{\dot{U}}_{Y}^{j} (x, z, ω)] \end{matrix}}{2} & (26) \end{matrix}$

In Equations (23) and (26), the key to calculating the average energy flux densities of Rayleigh and Love waves lies in determining the corresponding free-field velocities and stresses, which are derived from the free-field displacements. According to the dynamic-stiffness matrix method, the frequency-domain displacement solutions of soil particles in each layer induced by the Rayleigh and Love waves are illustrated in Equations (2) and (5), respectively. By applying partial differentiation with respect to time and position in Equation (2) respectively, the free-field velocities and stresses induced by Rayleigh waves can be calculated as:

{

U
.

X
j

(

x
,
z
,
ω

)

=

i
⁢
ω
⁢

U
X
j

(

x
,
z
,
ω

)