Microseismic multi-order spac coefficient and video dispersion curve joint imaging method, device and medium under relief topography

Abstract

This invention discloses a method, device, and medium for joint imaging of micro-motion multi-order SPAC coefficients and video dispersion curves under undulating terrain. The method includes: Step 1, dividing the undulating terrain into multiple local units, calculating the overall surface wave phase velocity, and then obtaining the SPAC coefficients to obtain the theoretical video dispersion curve; Step 2, obtaining the surface normal component from the three-component micro-motion data obtained from micro-motion observations, and extracting the SPAC coefficients and the optimal video dispersion curve under strong constraints of the SPAC coefficient zero point; Step 3, inputting the shear wave velocity of the strata, the elevation of the underlying stratum interface, and the elevation of the undulating terrain surface into a pre-trained LSTM network, and performing joint inversion based on the SPAC coefficients and video dispersion curves to obtain subsurface space imaging. This invention is applicable to undulating terrain conditions and can fully utilize multi-order surface wave dispersion information, exhibiting higher imaging accuracy and stability, thereby achieving high-resolution micro-motion imaging of subsurface space.

Description

Technical Field

[0001] This invention relates to underground space micro-motion imaging technology in the field of underground space development and utilization, and particularly to a method, device and medium for joint imaging of micro-motion multi-order SPAC coefficients and video dispersion curves under undulating terrain. Background Technology

[0002] The development and utilization of underground space is an important direction for scientific and technological development. Among them, micro-motion detection technology, which extracts surface wave dispersion information from micro-motion signals and performs inversion to obtain the underground shear wave velocity structure, is an important method for fine imaging of underground space. Methods for extracting surface wave dispersion curves from micro-motion data mainly include spatial autocorrelation methods (SPAC, ESPAC, etc.) (Aki, 1957), frequency-wavenumber domain (FK) methods (Capon, 1969), and frequency-Bessel (FJ) methods (Wang et al., 2019; You et al., 2022). Studies have shown that when extracting surface wave phase velocities from micro-motion data observed by arrays, SPAC methods are more accurate than FK methods (Ohori et al., 2002), while FJ methods have advantages in extracting higher-order surface wave dispersion (Zhao et al., 2023).

[0003] The SPAC method was first proposed by Aki in 1957. The data acquisition and observation system consists of a circularly distributed array and a central station (Aki, 1957). Ling and Okada proposed the extended spatial autocorrelation method, which allows for the design of arrays of arbitrary shapes (Ling, Okada, 1993). Betting et al. modified the SPAC method, using data from stations within a certain radius to extract surface wave dispersion curves (Betting et al., 2001), improving the quality of dispersion curve extraction through spatial data superposition. In 2001, Louie developed the refraction microseismic array method for processing one-dimensional linearly arranged microseismic signals (Louie, 2001), achieving good application results. Cho et al. proposed a centerless circular microseismic exploration method and used spectral ratio function analysis to extract surface wave dispersion curves (Cho et al., 2006), improving the accuracy of surface wave phase velocity calculation. Studies have shown that when random noise sources are spatially uniformly distributed, averaging the correlation coefficients of station pairs over different observation periods in the time domain is equivalent to averaging the spatial orientation of a circular station (Okada et al., 2006; Lu Laiyu, 2021). Shabani et al. extended the MSPAC method, extending the SPAC formula for discrete and approximately continuous circular arrays to handle arrays with regular and irregular azimuth spacing (Shabani et al., 2010), and used the horizontal polarization noise spread method of circular arrays to extract Rayleigh wave information more effectively (Shabani et al., 2011). Hayashi et al. introduced the concept of common center point imaging in active source surface wave exploration (Hayashi, 2002, 2004) into micro-motion exploration, establishing the CMP-SPAC method (Hayashi et al., 2015; Li Hongxing et al., 2021). To improve exploration efficiency, the CMP-SPAC method has evolved from two-dimensional imaging to three-dimensional imaging (Li Hongxing et al., 2019). To improve the resolution of micromotion spatial autocorrelation imaging methods, Xu Peifen et al. proposed a two-dimensional microarray micromotion detection method (Xu Peifen et al., 2021). To improve the resolution of linearly arranged micromotion exploration in urban environments, Li Hongxing et al. (2022) proposed a multi-scale linear micromotion imaging method, which improved the lateral resolution of the shear wave velocity profile. Cho developed an improved scheme for reading the zero-crossing point of the SPAC coefficient curve, using a very simple microseismograph array to identify the Rayleigh surface wave phase velocity (Cho et al., 2021, 2023). In order to effectively utilize the multi-order Rayleigh surface wave information of micromotion, inversion methods based on the SPAC coefficients of multi-order Rayleigh waves of micromotion have been developed (Wathelet et al., 2005; Xu Peifen et al., 2020), thus partially overcoming the problem of the difficulty in completely extracting the dispersion curves of multi-order surface waves from micromotion data.Furthermore, the imaging approach of Active Source Surface Wave Multichannel Analysis (MASW) has been introduced into micromotion detection (Park et al., 2004, 2007; Cheng et al., 2016; Cheng et al., 2023), mainly for processing linear array data. Forward modeling of surface wave dispersion curves is mainly based on horizontal stratigraphic models. Effectively utilizing multi-order dispersion information to invert the shear wave velocity structure of the subsurface medium is one of the key points and challenges of micromotion imaging methods (Xia et al., 2015; Ai et al., 2024).

[0004] In summary, although micromotion detection imaging technology has made significant progress, the following problems still exist: The theoretical forward and inverse methods for dispersion curves are based on the assumption of horizontal strata, making them suitable only for horizontal terrain conditions and severely limiting their applicability. When extracting dispersion curves from measured micromotion data, problems arise such as difficulty in revealing high-order dispersion spectra, difficulty in determining the order of multi-order dispersion spectra, and the extreme complexity of the dispersion spectrum due to the different and overlapping energy distributions of multi-order surface waves. These challenges make it difficult to accurately identify and extract multi-order dispersion curves, severely impacting the accuracy and stability of micromotion detection imaging. Therefore, how to make micromotion detection methods applicable to undulating terrain conditions and more effectively utilize the multi-order surface wave dispersion information from micromotion data for inverse imaging, thereby expanding its application range and improving imaging accuracy and stability, is an urgent scientific problem to be solved. Summary of the Invention

[0005] To address the limitations of current micro-motion detection imaging technologies, such as limited applicability, insufficient accuracy, and instability, this invention provides a method for joint imaging of micro-motion multi-order SPAC coefficients and video dispersion curves under undulating terrain conditions. This method is applicable to undulating terrain conditions, makes fuller use of multi-order surface wave dispersion information, and achieves higher imaging accuracy and stability.

[0006] To achieve the above-mentioned technical objectives, the technical solution of the present invention includes:

[0007] In a first aspect, the present invention provides a method for joint imaging of micro-motion multi-order SPAC coefficients and video dispersion curves under undulating terrain, comprising:

[0008] Step 1: Forward modeling of theoretical multi-order SPAC coefficients and video dispersion curves under undulating terrain:

[0009] The undulating terrain covered by the micro-motion detection array is divided into multiple local units. After treating the local units as horizontal terrain, the multi-order surface wave phase velocities of the local units are obtained based on the forward modeling of the theoretical dispersion curve. Then, based on weighted clustering analysis, the overall surface wave phase velocity is calculated from the multi-order surface wave phase velocities of the local units.

[0010] Based on the overall multi-order surface wave phase velocity and the corresponding energy ratio, the theoretical multi-order micro-motion SPAC coefficient and video dispersion curve considering the array distribution are calculated.

[0011] Step 2, Extraction of SPAC coefficient and video dispersion curve from measured micro-motion data under undulating terrain:

[0012] The component micro-motion data obtained from the micro-motion detection array is rotated and decomposed to obtain the surface normal component for calculating the measured SPAC coefficient. Based on the actual arc distance between the observation stations performing the micro-motion detection, the stations are grouped into a ring-shaped space of finite thickness. Within this ring-shaped space, the spatial superposition average of the SPAC coefficients is calculated as the measured SPAC coefficient with a radius equal to the center radius of the ring-shaped space. Then, under strong zero-point constraints on the measured SPAC coefficients, ... Minimize the optimal video dispersion curve;

[0013] Step 3, Joint inversion of micro-motion SPAC coefficients and video dispersion curves under undulating terrain:

[0014] The shear wave velocity of the strata and the elevation of the underlying strata interface are used as inversion parameters, and the elevation of the undulating terrain is used as a known parameter. These parameters are input into a trained LSTM network, and joint inversion is performed based on the SPAC coefficients and video dispersion curves to obtain subsurface space imaging.

[0015] Preferably, in step 1, when dividing the undulating terrain area into multiple local units, the undulating terrain area is divided into grids according to a preset grid size, and each grid covers a columnar body from the ground surface to a preset depth underground as a local unit.

[0016] Preferably, in step 1, when calculating the overall surface wave phase velocity, a weighted clustering analysis method is used. The relationship between the surface wave phase velocity of a local element and the overall surface wave phase velocity of the micro-motion detection array coverage area is obtained based on the amplitude-normalized clustering analysis output power spectrum. That is:

[0017]

[0018] in This represents the output power spectrum of the amplitude-normalized cluster analysis. Indicates angular frequency; Indicates the overall surface wave phase velocity over the coverage area of ​​the array; Indicates the total number of nodes in the mesh; This represents the number of local cells traversed by the connection between the m-th and n-th grid nodes. Indicates the first The forward phase velocity of a local element Indicates the first The length of the terrain arc in a local unit. This represents the distance between the terrain arcs of the m-th and n-th grid nodes;

[0019] Then, when the amplitude-normalized cluster analysis output power spectrum in the above formula reaches its maximum value, the corresponding global surface wave phase velocity is obtained.

[0020] Preferably, in step 1, the expression for calculating the theoretical multi-order micro-motion SPAC coefficient is:

[0021]

[0022] in Indicates frequency as The distance between stations is The theoretical SPAC curve; Indicates the first Step wave energy; This represents the sum of the energies of multiple-order surface waves; Represents the zeroth-order Bessel function; Represented as the first Overall phase velocity of the step wave; This represents the medium response function.

[0023] Preferably, in step 1, the expression for solving the video dispersion curve is:

[0024]

[0025] in, Describe the objective function. This represents the theoretical video dispersion curve. Indicates the spacing between the j-th stations;

[0026] By solving the above equation, we obtain the objective function. Minimum theoretical video dispersion curve .

[0027] Preferably, in step 2, the component micro-motion data obtained by micro-motion detection is a three-component system of vertical Z, north-south N, and east-west E, or a single-component observation of the vertical Z component.

[0028] Preferably, in step 2, the strong constraint on the zero point of the SPAC coefficient is:

[0029]

[0030] And based on the strong constraint of the zero point of the SPAC coefficient, the following is achieved:

[0031]

[0032] To minimize the dispersion curve, thus extracting the optimal video dispersion curve. .

[0033] Preferably, in step 3, the training process of the LSTM network includes:

[0034] Step ①: Construct sample data based on the constrained Markov decision method, and then divide the sample data into training set and test set according to a preset ratio;

[0035] Step 2: Set the number of hidden neurons, learning rate, and number of training rounds of the LSTM network as optimization parameters, and initialize the population;

[0036] Step ③: The prediction error of the LSTM network optimized by the HGS algorithm is used as the fitness value of the particle. The fitness value changes with the number of iterations. Each particle updates its individual optimal position and global optimal position according to the fitness value, and then updates its own velocity and position.

[0037] Step 4: Iterate until the fitness value of the particles tends to stabilize, then stop the iteration and update, and determine the values ​​of the number of hidden layer units, learning rate, and number of training rounds of the LSTM.

[0038] Step 5: Input the parameters obtained in step 4 into the LSTM network model for training until training is complete.

[0039] Secondly, the present invention also provides an electronic device, comprising:

[0040] One or more processors;

[0041] Storage device for storing one or more programs;

[0042] When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described above.

[0043] Thirdly, the present invention also provides a computer-readable medium storing a computer program that, when executed by a processor, implements the method as described above.

[0044] The technical advantage of this invention lies in its ability to overcome the shortcomings of existing methods, such as their inability to be applied to undulating terrain, their inability to fully utilize multi-order surface wave information for inversion, and the low accuracy and stability of inversion results. This invention establishes a method for forward and inverse modeling of SPAC coefficients and video dispersion curves under undulating terrain conditions, making the method applicable to undulating terrain conditions and enabling more full utilization of multi-order surface wave dispersion information. This results in higher imaging accuracy and stability, providing technical support for high-resolution micro-motion imaging in underground spaces. Attached Figure Description

[0045] Figure 1 This is a flowchart of an embodiment of the present invention;

[0046] Figure 2 This is a flowchart of the forward modeling of theoretical multi-order SPAC coefficients and video dispersion curves under undulating terrain in step S1 of an embodiment of the present invention.

[0047] Figure 3 This is a flowchart of step S2 of the present invention for extracting the SPAC coefficient and video dispersion curve from measured micro-motion data under undulating terrain;

[0048] Figure 4 This is a flowchart of the joint inversion of micro-motion SPAC coefficients and video dispersion curves under undulating terrain in step 3 of this embodiment of the invention;

[0049] Figure 5 This is a schematic diagram of micro-motion observation segment data in an embodiment of the present invention;

[0050] Figure 6 This is a schematic diagram illustrating the process and results of extracting the SPAC coefficients and video dispersion curves from micro-motion data in an embodiment of the present invention.

[0051] Figure 7 This is a schematic diagram illustrating the theoretical multi-order SPAC coefficients and the forward modeling process and results of the video dispersion curve in an embodiment of the present invention;

[0052] Figure 8 This is a schematic diagram of the joint inversion strategy of micro-motion SPAC coefficients and video dispersion curves in an embodiment of the present invention;

[0053] Figure 9 The following are comparison diagrams of the inversion results of underground shear wave velocity in embodiments of the present invention, wherein (a) is the inversion result of the basic dispersion curve, (b) is the inversion result of the multi-order dispersion curve, (c) is the inversion result of the multi-order SPAC coefficient, (d) is the joint inversion result of the multi-order SPAC coefficient and dispersion curve, and (e) is the joint inversion result of the multi-order SPAC coefficient and dispersion curve. Detailed Implementation

[0054] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings. It should be noted that the following illustrative embodiments and descriptions of the present invention are only for explaining the present invention and are not intended to limit the present invention.

[0055] like Figure 1 As shown, this embodiment provides a method for joint imaging of micro-motion multi-order SPAC coefficients and video dispersion curves under undulating terrain. The method includes:

[0056] Step S1: Forward modeling of theoretical multi-order SPAC coefficients and video dispersion curves under undulating terrain.

[0057] See Figure 2 , Figure 2This is a flowchart of the forward modeling process for theoretical multi-order SPAC coefficients and video dispersion curves under undulating terrain in step S1. Figure 7 The diagram illustrates the corresponding process and results. First, the coverage area of ​​the micro-motion observation array is vertically meshed according to a preset grid size, thus dividing the entire coverage area into several local units. Then, each local unit is considered as horizontal terrain. This decomposes the relatively complex undulating terrain into multiple relatively simple local horizontal processing units, allowing the multi-order surface wave phase velocities of the local units to be obtained using the traditional theoretical dispersion curve forward modeling method.

[0058] Secondly, using the weighted clustering analysis method, a formula is obtained to calculate the overall surface wave phase velocity of the coverage area of ​​the undulating terrain observation array from the surface wave phase velocity of local units.

[0059]

[0060] in This represents the output power spectrum of the amplitude-normalized cluster analysis. Indicates angular frequency; Indicates the total number of nodes in the mesh; This represents the number of local cells traversed by the connection between the m-th and n-th grid nodes; Indicates the first Forward modeling of multi-order surface wave phase velocities of local elements Indicates the first The length of the terrain arc of each local unit, therefore Used to indicate the first The propagation time of surface waves in actual undulating terrain within a local unit; and This represents the overall surface wave phase velocity over the coverage area of ​​the array. This represents the distance between the terrain arcs of the m-th and n-th grid nodes, therefore This is used to represent the theoretical propagation time of a wave between the m and n grid nodes. From the entire formula, when the two terms within the parentheses, i.e. and The closer the value is to , the closer the value of cosine is to the maximum value of 1, and thus the corresponding The closer to the peak value, the higher the angular frequency. In this embodiment, the golden section extreme value search algorithm is used to make... satisfy The peak demand, that is, the result at this angular frequency. Below, the overall surface wave phase velocity at the location covered by the array. .

[0061] Then, since multiple order surface waves overlap simultaneously in the surface waves, and the energy contributions of surface waves of different orders are different, this embodiment continues to calculate the energy proportion of multiple order surface waves for the surface waves, and then calculates the theoretical multi-order micro-motion SPAC coefficient considering the array distribution by combining the overall multi-order surface wave phase velocity and its energy proportion:

[0062]

[0063] in Indicates frequency as The distance between stations is The theoretical SPAC curve; Indicates the first Step wave energy; This represents the sum of the energies of multiple-order surface waves; Represents the zeroth-order Bessel function; Indicates the first Overall phase velocity of the step wave; Indicates the first The first-order medium response function, which also represents the second-order medium response function. The amplitude of a surface wave; that is, the energy of each surface wave is related to its phase velocity and amplitude.

[0064] Finally, the video dispersion curve is solved. The video dispersion curve is a comprehensive representation of the phase velocity of multi-order surface wave dispersion curves with different energy proportions. By solving the following equation, the objective function can be obtained. Minimum theoretical video dispersion curve :

[0065]

[0066] in This represents the spacing between the j-th stations.

[0067] Specifically, in the formula above It means through The theoretical SPAC coefficient is calculated to obtain the following: The SPAC coefficients, which are closer to reality, are calculated based on the phase velocities and energy proportions of the surface waves obtained earlier. The objective function of this embodiment... This actually represents the error between the two, so it can be adjusted... This is done to reduce the error. When the error is adjusted to its minimum, the theoretical video dispersion curve is obtained.

[0068] Step S2: Extracting SPAC coefficients and video dispersion curves from measured micro-motion data under undulating terrain.

[0069] Figure 3This document presents a flowchart for extracting SPAC coefficients and video dispersion curves from measured micromotion data under undulating terrain. When the terrain is undulating, calculating the SPAC coefficients using the surface normal component yields higher quality results. After data acquisition, the observed Z, N, and E (vertical, north-south, east-west) three-component or single-component Z (vertical) component micromotion data is first rotated and decomposed to obtain the surface normal component Sz used for SPAC coefficient calculation. When calculating the SPAC coefficients, the spacing between micromotion observation stations is based on the actual arc distance considering the terrain. Under undulating terrain conditions, it is difficult to guarantee that the arc distance between stations conforms to a strictly regular geometric shape, and the arrangement of observation arrays is often irregular. Therefore, under undulating terrain conditions, it is impractical to perform azimuth superposition and averaging of SPAC coefficients at specific station spacings. Therefore, in this embodiment, the observation stations are grouped according to their arc spacing within a finite-thickness annular space. Within this annular area, the spatial superposition average of the SPAC coefficients is taken as the SPAC coefficient with a radius equal to the center radius of the annular space. .

[0070] The zeros of SPAC coefficients are unaffected by incoherent noise, and calculating dispersion curves using the zeros of SPAC coefficients offers higher accuracy, stability, and anti-interference capabilities. A method for calculating video dispersion curves using SPAC coefficient zeros with strong constraints is described below: Under constraints, make To minimize the interference, extract the optimal video dispersion curve. .

[0071] Step S3: Joint inversion of micro-motion SPAC coefficients and video dispersion curves under undulating terrain.

[0072] See Figure 4In this step, the underlying medium of the observation array is treated as a whole for parameter inversion. Under undulating terrain conditions, the thickness of the underlying strata is not uniform at each station location and cannot be used as an inversion parameter. Therefore, the shear wave velocity of the strata and the elevation of the underlying stratum interface are used as the main inversion parameters. The elevation of the undulating terrain can be obtained through measurement and is used as a known parameter input. Both the micro-motion SPAC coefficients and video dispersion curves contain multi-order surface wave dispersion information. Multi-order micro-motion SPAC coefficients and video dispersion curves are used to carry out joint inversion of SPAC coefficients and video dispersion curves based on a hybrid algorithm of deep learning and intelligent optimization. In this embodiment, an LSTM network is used for joint inversion of SPAC coefficients and video dispersion curves. In the LSTM network parameter selection stage, the Hunger Game search algorithm is first used to optimize the network parameters to overcome the difficulties in parameter selection and low network prediction accuracy when manually tuning parameters. The overall workflow of the HGS-LSTM algorithm is as follows: ① Construct sample data based on the constrained Markov decision method and divide it into training and test sets; ② Set the number of hidden neurons, learning rate, and number of training rounds of the LSTM network as optimization parameters and initialize the population; ③ Use the prediction error of the LSTM network optimized by the HGS algorithm as the fitness value of the particles. The fitness value changes with the number of iterations. Each particle updates its individual optimal position and global optimal position based on the fitness value, and then updates its own velocity and position; ④ Stop iterative updates when the fitness value of the particles tends to stabilize, and determine the values ​​of the number of hidden layer units, learning rate, and number of training rounds of the LSTM; ⑤ Input the optimal parameters into the LSTM network model for training and prediction.

[0073] The following specific data illustrates the effects of the present invention. Figure 5 This is a fragment of micro-motion observation data. Figure 6 The diagram shows the process and results of extracting SPAC coefficients from micro-motion data and video dispersion curves. Figure 7 The figure shows the forward modeling process and results of the theoretical multi-order SPAC coefficients and the video dispersion curve. Figure 8 This is a schematic diagram of the joint inversion strategy of micro-motion SPAC coefficients and video dispersion curves. Figure 9 The image shows a comparison of the inversion results for subsurface shear wave velocity, including (a) the inversion result of the fundamental dispersion curve, (b) the inversion result of the multi-order dispersion curve, (c) the inversion result of the multi-order SPAC coefficients, (d) the joint inversion result of the multi-order SPAC coefficients and dispersion curve, and (e) the joint inversion result of the multi-order SPAC coefficients and dispersion curve. From the inversion results ( Figure 9 As can be seen, the joint inversion results of multi-order SPAC coefficients and video dispersion curves are the best in terms of both inversion accuracy and stability, which proves the effectiveness of the method of this invention.

[0074] According to embodiments of the present invention, the present invention also provides an electronic device and a computer-readable medium.

[0075] Electronic devices include:

[0076] One or more processors;

[0077] Storage device for storing one or more programs.

[0078] When the one or more programs are executed by the one or more processors, the one or more processors implement the aforementioned method for joint imaging of micro-motion multi-order SPAC coefficients and video dispersion curves under undulating terrain.

[0079] In practical use, users can interact with servers, which are also electronic devices, via a network to receive or send messages. Terminal devices are generally various electronic devices equipped with displays and user interfaces, including but not limited to smartphones, tablets, laptops, and desktop computers, especially resource-constrained edge devices. Various specific application software can be installed on these terminal devices as needed, including but not limited to web browsers, instant messaging software, social media platforms, and shopping apps.

[0080] Similarly, the computer-readable medium of the present invention stores a computer program that, when executed by a processor, implements the aforementioned method for joint imaging of micro-motion multi-order SPAC coefficients and video dispersion curves under undulating terrain.

[0081] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for joint imaging of micro-motion multi-order SPAC coefficients and video dispersion curves under undulating terrain, characterized in that, include: Step 1: Forward modeling of theoretical multi-order SPAC coefficients and video dispersion curves under undulating terrain: The undulating terrain covered by the micro-motion detection array is divided into multiple local units. After treating the local units as horizontal terrain, the multi-order surface wave phase velocities of the local units are obtained based on the forward modeling of the theoretical dispersion curve. Then, based on weighted clustering analysis, the overall surface wave phase velocity is calculated from the multi-order surface wave phase velocities of the local units. Based on the overall multi-order surface wave phase velocity and the corresponding energy ratio, the theoretical multi-order micro-motion SPAC coefficient and video dispersion curve considering the array distribution are calculated. Step 2, Extraction of SPAC coefficient and video dispersion curve from measured micro-motion data under undulating terrain: The component micro-motion data obtained by the micro-motion detection array is rotated and decomposed to obtain the surface normal component for calculating the measured SPAC coefficient. Based on the actual arc distance between the observation stations that performed the micro-motion detection, the observation stations are grouped into annular spaces of finite thickness. Within the annular space, the spatial superposition average value of the SPAC coefficient is calculated as the measured SPAC coefficient when the radius is the center radius of the annular space. Then, under the strong constraint of zero point of the measured SPAC coefficient, make Minimize the optimal video dispersion curve; Step 3, Joint inversion of micro-motion SPAC coefficients and video dispersion curves under undulating terrain: The shear wave velocity of the strata and the elevation of the underlying strata interface are used as inversion parameters, and the elevation of the undulating terrain is used as a known parameter. These parameters are input into a trained LSTM network, and joint inversion is performed based on the SPAC coefficients and video dispersion curves to obtain subsurface space imaging.

2. The method for joint imaging of micro-motion multi-order SPAC coefficients and video dispersion curves under undulating terrain according to claim 1, characterized in that, In step 1, when dividing the undulating terrain area into multiple local units, the undulating terrain area is divided into grids according to a preset grid size, and each grid covers a columnar body from the surface to a preset depth underground as a local unit.

3. The method for joint imaging of micro-motion multi-order SPAC coefficients and video dispersion curves under undulating terrain according to claim 2, characterized in that, In step 1, the calculation of the overall surface wave phase velocity is based on the weighted clustering analysis method. The relationship between the surface wave phase velocity of local cells and the overall surface wave phase velocity of the micro-motion detection array coverage area is obtained from the amplitude-normalized clustering analysis output power spectrum, namely: ； in This represents the output power spectrum of the amplitude-normalized cluster analysis. Indicates angular frequency; Indicates the overall surface wave phase velocity over the coverage area of ​​the array; Indicates the total number of nodes in the mesh; This represents the number of local cells traversed by the connection between the m-th and n-th grid nodes. Indicates the first The forward phase velocity of a local element Indicates the first The length of the terrain arc in a local unit. This represents the distance between the terrain arcs of the m-th and n-th grid nodes; Then, when the amplitude-normalized clustering analysis output power spectrum in the above formula reaches its maximum value, the corresponding global surface wave phase velocity is obtained.

4. The method for joint imaging of micro-motion multi-order SPAC coefficients and video dispersion curves under undulating terrain according to claim 3, characterized in that, In step 1, the expression for calculating the theoretical multi-order micro-motion SPAC coefficient is as follows: ； in Indicates frequency as The distance between stations is The theoretical SPAC curve; Indicates the first Step wave energy; This represents the sum of the energies of multiple-order surface waves; Represents the zeroth-order Bessel function; Represented as the first Overall phase velocity of the step wave; This represents the medium response function.

5. The method for joint imaging of micro-motion multi-order SPAC coefficients and video dispersion curves under undulating terrain according to claim 4, characterized in that, In step 1, the expression for solving the video dispersion curve is: ； in, Describe the objective function. This represents the theoretical video dispersion curve. Indicates the spacing between the j-th stations; By solving the above equation, we obtain the objective function. Minimum theoretical video dispersion curve .

6. The method for joint imaging of micro-motion multi-order SPAC coefficients and video dispersion curves under undulating terrain according to claim 5, characterized in that, In step 2, the component micro-motion data obtained by micro-motion detection are three components: vertical Z, north-south N, and east-west E, or the vertical Z component observed by a single component.

7. The method for joint imaging of micro-motion multi-order SPAC coefficients and video dispersion curves under undulating terrain according to claim 6, characterized in that, In step 2, the strong constraint on the zero point of the SPAC coefficient is: ； And based on the strong constraint of the zero point of the SPAC coefficient, the following is achieved: ； To minimize the dispersion curve, thus extracting the optimal video dispersion curve. .

8. The method for joint imaging of micro-motion multi-order SPAC coefficients and video dispersion curves under undulating terrain according to claim 1, characterized in that, In step 3, the training process of the LSTM network includes: Step ①: Construct sample data based on the constrained Markov decision method, and then divide the sample data into training set and test set according to a preset ratio; Step 2: Set the number of hidden neurons, learning rate, and number of training rounds of the LSTM network as optimization parameters, and initialize the population; Step ③: The prediction error of the LSTM network optimized by the HGS algorithm is used as the fitness value of the particle. The fitness value changes with the number of iterations. Each particle updates its individual optimal position and global optimal position according to the fitness value, and then updates its own velocity and position. Step 4: Iterate until the fitness value of the particles tends to stabilize, then stop the iteration and update, and determine the values ​​of the number of hidden layer units, learning rate, and number of training rounds of the LSTM. Step 5: Input the parameters obtained in step 4 into the LSTM network model for training until training is complete.

9. An electronic device, characterized in that, include: One or more processors; Storage device for storing one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any one of claims 1-8.

10. A computer-readable medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1-8.

Images