Earthquake service system for micro-motion exploration and data processing method

Abstract

The invention discloses a seismic service system for micro-motion exploration and a data processing method, and relates to the technical field of geophysical exploration, and the system comprises the steps: deploying a distributed sensor array, and achieving the high-precision time synchronization through a self-organizing wireless network; continuously collecting environment vibration data and preprocessing the environment vibration data; calculating a cross-correlation function based on a phase weighted stacking algorithm to extract an empirical Green function; acquiring a high-precision Rayleigh wave phase velocity frequency dispersion curve by adopting a time-frequency domain phase matching pursuit technology; constructing a three-dimensional shear wave velocity model by combining a Markov chain Monte Carlo inversion algorithm of Bayesian reasoning, and introducing transverse constraint and prior geological information; and establishing a quality feedback control mechanism to optimize acquisition and processing parameters. According to the method, the stability of micro-motion signal processing and the resolution of underground structure imaging in a low signal-to-noise ratio environment are remarkably improved, full-process automation is realized by relying on a cloud service platform, and a single exploration processing period is shortened to be within 48 hours.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of geophysical exploration, and particularly relates to a microtremor exploration seismic service system and a data processing method. BACKGROUND

[0002] As a core means of geophysical exploration, seismic exploration is widely used in oil and gas resource exploration, geological disaster assessment and underground structure imaging fields. Among them, microtremor exploration as a passive source seismic method, through the collection of Rayleigh wave or Love wave signals in the natural environmental noise, the shear wave velocity structure of the underground medium is inverted, which has the advantages of no artificial source, environmental friendly, suitable for densely populated urban areas and other significant advantages. In recent years, with the development of high-sensitivity seismic sensors and long-term continuous observation technology, microtremor exploration has shown higher and higher application value in shallow geological survey and engineering geological evaluation. However, this technology puts forward very high requirements for data quality, processing algorithm and system integration capability, especially in complex urban environment, how to effectively extract stable and reliable frequency dispersion characteristics from low signal-to-noise ratio continuous background vibration has become the key bottleneck restricting its precision and efficiency.

[0003] Among them, the data processing flow of microtremor exploration usually includes signal preprocessing, cross-correlation function calculation, frequency dispersion curve extraction and inversion modeling. The existing system mostly adopts discrete architecture, and the front-end acquisition device is only responsible for the original waveform record, and all data processing tasks are completed in the back-end workstation. This separation design causes that the data quality cannot be evaluated in real time on site, it is difficult to dynamically adjust the observation parameters, and the transmission and storage cost of massive raw data is high. At the same time, the traditional frequency dispersion curve extraction method relies on the Fourier transform or wavelet analysis with fixed window length, when facing non-stationary and non-Gaussian distributed urban environmental noise, it is easy to be affected by transient interference and coherent noise, causing the estimation deviation of phase velocity, and then reducing the spatial resolution and reliability of the inversion result.

[0004] There are three major defects in the prior art in the microtremor exploration system: first, there is lack of embedded processing capability for edge computing, which cannot complete key steps such as noise suppression, cross-correlation and time-frequency analysis locally at the acquisition node, resulting in system response lag and dependence on high-bandwidth communication link; second, the frequency dispersion feature extraction algorithm does not fully integrate the multi-channel spatial coherence and time stability constraints, making it difficult to achieve robust phase velocity estimation under low signal-to-noise ratio conditions; third, there is lack of closed-loop feedback mechanism between data processing flow and inversion model, which cannot automatically optimize subsequent acquisition strategy or reprocessing parameters according to the preliminary inversion result. The above problems are particularly prominent in the fine exploration of urban underground space, safety monitoring of major infrastructure and other application scenarios with strict requirements on timeliness and accuracy, which seriously restricts the large-scale deployment and intelligent upgrading of microtremor exploration technology. SUMMARY

[0005] The application aims to provide a micro exploration seismic service system and data processing method to solve the problems of low data acquisition quality, unstable dispersion feature extraction and insufficient inversion imaging resolution in the prior art.

[0006] To solve the above technical problems, the application provides the following technical solutions. A micro exploration seismic service system and data processing method, comprising the following specific steps: Step S1: deploying a distributed seismic sensor array, at least 36 three-component broadband seismometers are arranged according to a preset geometric pattern in a target exploration area, each seismometer is internally provided with a 24-bit analog-to-digital converter and a global positioning system synchronization module, the sampling rate is set to 100 Hz, the dynamic range is greater than 130 decibels, and the time synchronization accuracy between nodes is better than 1 millisecond through a self-organizing wireless sensor network; Step S2: collecting continuous micro motion signals, continuously recording environmental vibration data for at least 30 days through the distributed seismic sensor array, adopting a circular buffer mechanism for data storage, generating an independent data file every 24 hours, and the file format is an international standard mini-seed format, while real-time monitoring of the battery state and storage space usage of each node is performed; Step S3: preprocessing the original waveform data, sequentially performing mean removal, trend removal and instrument response correction processing on the collected continuous micro motion signals, adopting a 4th order Butterworth band-pass filter to retain an effective frequency band of 0.1 Hz to 20 Hz, and automatically detecting and removing abnormal data segments containing obvious human interference or instrument failure; Step S4: calculating a cross-correlation function, cutting a continuous data segment with a length of 1 hour from the preprocessed continuous data, calculating a cross-correlation function for the vertical component records of any two sensor nodes, adopting a time domain normalization processing method to suppress the influence of non-stationary noise, and obtaining an empirical Green function estimate between all sensor pairs; Step S5: extracting dispersion features, performing time-frequency analysis on the cross-correlation function, adopting a multiple window spectral estimation method to calculate a Rayleigh wave phase velocity dispersion curve, the frequency resolution is 0.01 Hz, the base mode dispersion energy is automatically identified in a frequency range of 0.5 Hz to 10 Hz, and the signal-to-noise ratio of the dispersion curve is enhanced through group velocity filtering; Step S6: constructing a three-dimensional velocity model, based on the extracted dispersion curve, adopting a Markov Chain Monte Carlo inversion algorithm based on Bayesian inference to reconstruct the underground shear wave velocity structure, the inversion parameters include layer thickness, layer velocity and layer density, and the prior constraints come from regional geological data and drilling data; Step S7: realizing quality feedback control, real-time evaluation of dispersion curve extraction quality and inversion result uncertainty, automatic adjustment of data acquisition parameters or triggering of supplementary observation of specific nodes when the quality index is lower than a preset threshold, and forming a closed-loop optimized workflow.

[0007] Preferably, in step S1, the distributed seismic sensor array adopts a hexagonal grid layout pattern, and the node spacing is dynamically adjusted according to the detection depth requirements. The shallow detection depth is set to 10 to 50 meters, and the medium-deep detection depth is set to 100 to 500 meters. Each node is equipped with a solar power supply system and a 4G or 5G wireless communication module to realize remote status monitoring and data transmission.

[0008] Preferably, the preprocessing process in step S3 further includes using an independent component analysis algorithm to separate noise components from different sources, establishing a feature template library for major interference sources such as traffic vibration, industrial machinery vibration and wind-induced ground vibration, and achieving targeted noise suppression through template matching, with a signal-to-noise ratio improvement of more than 10 dB.

[0009] Preferably, the cross-correlation function calculation in step S4 adopts an improved algorithm based on phase weighted superposition. By performing phase consistency weighting on the cross-correlation results of multiple 1-hour data segments, weak signal components are enhanced. The weighting factor is proportional to the signal coherence, and the effective signal-to-noise ratio threshold is reduced to negative 15 dB.

[0010] Preferably, in step S5, the dispersion feature extraction adopts time-frequency domain phase matching tracking technology. By constructing a redundant Gabor dictionary to sparsely represent the signal, the optimal time-frequency atomic matching Rayleigh wave dispersion mode is adaptively selected, and the frequency estimation accuracy is better than 0.005, and the phase velocity estimation standard deviation is less than 0.02.

[0011] Preferably, the three-dimensional velocity model construction in step S6 adopts the layered medium assumption, the initial model layer number is set to 8 to 15 layers, the thickness of each layer increases with the depth, an adaptive sampling strategy is adopted in the inversion process, the Markov chain length is 10,000 iterations, the combustion period is 2,000 iterations, and the convergence criterion is that the Gelman-Rubin statistic is less than 1.1.

[0012] Preferably, in step S7, the quality feedback control system is based on a comprehensive evaluation of multiple quality indicators, including the continuity index of the dispersion curve, the distribution of the inversion residual, and the width of the posterior probability interval of the model parameters. When any quality indicator exceeds a preset threshold, the system automatically generates optimization suggestions, such as adjusting the observation duration, modifying filter parameters, or adding sensor pairs in specific directions.

[0013] Preferably, the method further includes inserting a spatial coherence analysis step between step S4 and step S5, calculating the spatial coherence coefficients of all sensor pairs in multiple frequency bands, constructing a frequency-dependent coherence matrix, identifying the dominant propagation direction through eigenvalue decomposition, and optimizing the weighting strategy of the cross-correlation function based on this.

[0014] Preferably, in step S6, the inversion process introduces lateral constraints, and the prior information of adjacent measuring points is incorporated into the inversion objective function through the Kriging interpolation method, which smooths the lateral velocity variation and reduces the ambiguity of the inversion results, making it particularly suitable for fine imaging of complex geological structures.

[0015] Preferably, it also includes establishing a micro-motion exploration database to store all collected raw data, process intermediate results and final inversion models. The database adopts a time series database architecture, supports multi-dimensional data retrieval and visualization, and provides data services to third-party analysis tools through application programming interfaces.

[0016] Preferably, the method is integrated into a cloud-based earthquake service platform, providing fully automated services from data acquisition and processing to result display. Users can configure exploration parameters, monitor processing progress, and download analysis reports through a web interface, reducing the processing time for a single exploration data session from several weeks using traditional methods to less than 48 hours.

[0017] Compared with the prior art, the beneficial technical effects of the present invention are as follows: This invention combines a distributed seismic sensor array with embedded processing capabilities to achieve high-quality acquisition and preliminary on-site processing of micro-motion signals, significantly reducing data transmission and storage requirements. The improved cross-correlation function calculation and dispersion feature extraction algorithms effectively enhance the accuracy and stability of Rayleigh wave phase velocity estimation in low signal-to-noise ratio environments.

[0018] This invention, based on the inversion method of Bayesian inference and the quality feedback control mechanism, forms a closed-loop optimized workflow, which enhances the system's adaptability to complex urban environments and improves the resolution and reliability of underground velocity structure imaging. The cloud-based integrated service platform provides fully automated services, which greatly improves the work efficiency and user experience of micro-motion exploration, and provides strong technical support for fine exploration of urban underground space and infrastructure safety monitoring. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the overall technical architecture of a seismic service system and data processing method for micro-motion exploration proposed in this invention; Figure 2 This is a schematic diagram of the core principle framework of Markov chain Monte Carlo inversion and quality feedback control closed-loop optimization based on Bayesian inference in this invention. Figure 3 This is a schematic diagram illustrating the specific steps of a seismic service system and data processing method for micro-motion exploration proposed in this invention. Detailed Implementation

[0020] The features and exemplary embodiments of various aspects of the present invention will now be described in detail. To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are merely intended to explain the present invention and not to limit the present invention. For those skilled in the art, the present invention can be practiced without some of these specific details. The following description of the embodiments is merely to provide a better understanding of the present invention by illustrating examples of the invention.

[0021] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus that includes said element.

[0022] In the embodiments of the present invention, the same reference numerals denote the same components, and for the sake of brevity, detailed descriptions of the same components are omitted in different embodiments. It should be understood that the thickness, length, width, and other dimensions of various components in the embodiments of the present invention shown in the accompanying drawings, as well as the overall thickness, length, width, and other dimensions of the integrated device, are merely illustrative and should not constitute any limitation on the present invention; the term "multiple" in the present invention refers to two or more.

[0023] Example 1 To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0024] Currently, seismic exploration, as a core means of geophysical exploration, is widely used in fields such as oil and gas resource exploration, geological hazard assessment, and subsurface structure imaging. Among these, micromotion exploration, as a passive source seismic method, inverts the shear wave velocity structure of the subsurface medium by acquiring Rayleigh or Love wave signals from natural environmental noise. It boasts significant advantages such as not requiring artificial seismic sources, being environmentally friendly, and suitable for densely populated urban areas. In recent years, with the development of high-sensitivity seismic sensors and long-term continuous observation technology, micromotion exploration has demonstrated increasingly high application value in shallow geological surveys and engineering geological evaluations. However, this technology places extremely high demands on data quality, processing algorithms, and system integration capabilities. Especially in complex urban environments, effectively extracting stable and reliable dispersion features from continuous background vibrations with low signal-to-noise ratios has become a key bottleneck restricting its accuracy and efficiency. To address these technical problems, this invention proposes a combination of distributed seismic sensor arrays and embedded processing capabilities to achieve high-quality acquisition and preliminary on-site processing of micromotion signals, significantly reducing data transmission and storage requirements. Improved cross-correlation function calculation and dispersion feature extraction algorithms effectively enhance the accuracy and stability of Rayleigh wave phase velocity estimation in low signal-to-noise ratio environments. The inversion method based on Bayesian inference and the quality feedback control mechanism form a closed-loop optimization workflow, enhancing the system's adaptability to complex urban environments and improving the resolution and reliability of underground velocity structure imaging. The cloud-based integrated service platform provides fully automated services, significantly improving the efficiency and user experience of micro-motion exploration, providing strong technical support for fine-grained exploration of urban underground space and infrastructure safety monitoring, and is applied to a seismic service system and data processing method for micro-motion exploration.

[0025] refer to Figure 1 The schematic diagram of the overall technical architecture of the seismic service system and data processing method for micro-motion exploration proposed in this invention clearly illustrates the complete data flow and control logic from the front-end distributed sensor network to the back-end cloud service platform. The entire system consists of a distributed seismic sensor array deployed in the target area, edge computing nodes, a wireless communication network, a cloud-based seismic service platform, and a user interaction terminal.

[0026] Step S1: Deploy a distributed seismic sensor array. At least 36 three-component broadband seismometers are deployed in the target exploration area according to a preset geometric pattern. Each seismometer has a built-in 24-bit analog-to-digital converter and a global positioning system synchronization module. The sampling rate is set to 100 Hz, the dynamic range is greater than 130 dB, and the time synchronization accuracy between nodes is better than 1 millisecond through a self-organizing wireless sensor network.

[0027] Specifically, in step S1, the distributed seismic sensor array adopts a hexagonal grid layout. This geometric configuration maximizes spatial coverage efficiency and ensures uniform wavefield sampling in all directions. The node spacing is dynamically adjusted according to the required detection depth. For shallow detection tasks, the node spacing is set to 10 to 50 meters; for medium-deep detection tasks, the node spacing is extended to 100 to 500 meters. Each sensor node is equipped with a solar power system to support long-term unattended operation and integrates a 4G or 5G wireless communication module for real-time transmission of key status information (such as battery voltage, storage capacity, and temperature) and processed compressed data, while also receiving remote commands from the cloud platform. The sensor nodes integrate embedded processors, enabling them to perform local data preprocessing and preliminary analysis, thereby reducing the amount of raw data by more than 90% and greatly alleviating communication bandwidth pressure.

[0028] Step S2: Collect continuous micro-motion signals and continuously record environmental vibration data for at least 30 days through a distributed seismic sensor array. Data storage adopts a circular buffer mechanism, generating an independent data file every 24 hours in the international standard mini-seed format. At the same time, monitor the battery status and storage space utilization of each node in real time.

[0029] Specifically, in step S2, the continuous data acquisition process is autonomously managed by the embedded system of each node. A circular buffer mechanism ensures that, given limited storage space, the system prioritizes retaining the latest valid data segments. At the end of a single 24-hour cycle, the system automatically packages the data in the buffer into a mini-seed format file conforming to international standards. This format contains complete metadata information, such as sampling rate, channel gain, geographical location, and timestamp. Simultaneously, the nodes continuously monitor their own health status. Once the battery level drops below 20% or storage space occupancy exceeds 90%, an alarm is immediately sent to the cloud platform via wireless link, allowing maintenance personnel to intervene promptly.

[0030] Step S3: Preprocess the raw waveform data. The acquired continuous micro-motion signals are subjected to mean removal, trend removal and instrument response correction in sequence. A fourth-order Butterworth bandpass filter is used to retain the effective frequency band from 0.1 Hz to 20 Hz, and abnormal data segments containing obvious human interference or instrument malfunctions are automatically detected and removed.

[0031] Specifically, in step S3, the preprocessing process first removes the DC offset of the signal (mean removal), and then uses linear or polynomial fitting to eliminate long-term drift trends (desalination). Subsequently, the data is deconvolved using a pre-calibrated instrument response function to convert the recorded voltage signal into actual ground motion velocity or acceleration. A fourth-order Butterworth filter is used in the bandpass filtering stage; its steep roll-off effectively suppresses out-of-band high-frequency electronic noise and low-frequency drift interference. Furthermore, the preprocessing also includes using independent component analysis (ICA) to separate noise components from different sources, establishing a feature template library for major interference sources such as traffic vibration, industrial machinery vibration, and wind-induced ground vibration, and achieving targeted noise suppression through template matching, resulting in a signal-to-noise ratio improvement of more than 10 dB. Automatic detection of abnormal data segments is based on multi-dimensional threshold judgment, including indicators such as sudden changes in signal amplitude, abnormal spectral energy distribution, and a sharp drop in correlation with other neighboring nodes. Data segments marked as abnormal are either zeroed out or discarded directly and do not participate in subsequent processing.

[0032] Step S4: Calculate the cross-correlation function. Extract a continuous data segment of length 1 hour from the preprocessed continuous data. Calculate the cross-correlation function for the vertical component records of any two sensor nodes. Use time-domain normalization to suppress the influence of non-stationary noise and obtain the empirical Green's function estimate between all sensor pairs.

[0033] Specifically, in step S4, the system slides and extracts multiple 1-hour data windows from the continuous records. For any two nodes i and j in the array, the time series of their vertical components is extracted. and Calculate its cross-correlation function as follows: To suppress the influence of non-stationary noise, time-domain normalization is employed. This involves first standardizing each 1-hour data segment using a sliding window, followed by cross-correlation calculations. In step S4, the cross-correlation function calculation uses an improved algorithm based on phase-weighted superposition. By applying phase-consistency weighting to the cross-correlation results of multiple 1-hour data segments, weak signal components are enhanced. The weighting factor is proportional to the signal coherence, effectively reducing the signal-to-noise ratio threshold to -15 dB. This weighting strategy significantly improves the ability to extract weak Green's function signals against strong noise backgrounds.

[0034] The aforementioned seismic service system and data processing method for micro-motion exploration further includes a spatial coherence analysis step inserted between steps S4 and S5. This step calculates the spatial coherence coefficients of all sensor pairs in multiple frequency bands, constructs a frequency-dependent coherence matrix, identifies the dominant propagation direction through eigenvalue decomposition, and optimizes the weighting strategy of the cross-correlation function based on this.

[0035] Specifically, this spatial coherence analysis step divides the effective frequency band, ranging from 0.1 Hz to 20 Hz, into several sub-bands, for example, with a step size of 1 Hz. Within each sub-band, the coherence coefficient γ(f) between all sensor pairs is calculated, and a... The coherence matrix C(f) is given, where N is the total number of sensors. Eigenvalue decomposition is performed on this matrix; the eigenvector corresponding to the largest eigenvalue indicates the dominant propagation direction of the noise field at that frequency. Based on this, the system dynamically adjusts the weighting strategy of the cross-correlation function, assigning higher weights to signals propagating along the dominant direction, thereby further improving the signal-to-noise ratio and directional selectivity of dispersion feature extraction.

[0036] Step S5: Extract dispersion features, perform time-frequency analysis on the cross-correlation function, calculate the Rayleigh wave phase velocity dispersion curve using the multiple window spectrum estimation method with a frequency resolution of 0.01 Hz, automatically identify the dispersion energy of the fundamental mode in the frequency range of 0.5 Hz to 10 Hz, and enhance the signal-to-noise ratio of the dispersion curve by group velocity filtering.

[0037] Specifically, in step S5, a time-frequency transform is performed on the weighted and optimized cross-correlation function. The dispersion feature extraction in step S5 employs a time-frequency domain phase-matched tracking technique. By constructing a redundant Gabor dictionary to sparsely represent the signal, the optimal time-frequency atom is adaptively selected to match the Rayleigh wave dispersion mode, achieving a frequency estimation accuracy better than 0.005 and a phase velocity estimation standard deviation less than 0.02. This technique, by searching for the atom combination that best matches the theoretical Rayleigh wave dispersion mode on the time-frequency plane, can accurately track the instantaneous frequency of non-stationary signals, effectively overcoming the limitations of traditional Fourier transform in processing short-time, non-stationary signals. The automatic identification algorithm, based on energy concentration and phase continuity criteria, locks the energy ridge of the fundamental mode within the range of 0.5 Hz to 10 Hz. Subsequently, a group velocity filter is applied along this ridge to focus energy, suppressing sidelobe interference, and finally outputting a high signal-to-noise ratio phase velocity-frequency relationship curve.

[0038] Step S6: Construct a three-dimensional velocity model. Based on the extracted dispersion curves, use a Markov chain Monte Carlo inversion algorithm based on Bayesian inference to reconstruct the underground shear wave velocity structure. The inversion parameters include layer thickness, layer velocity, and layer density. The prior constraints are derived from regional geological data and borehole data.

[0039] Specifically, in step S6, the inversion process is conducted within a probabilistic framework. The system's likelihood function is defined as the sum of squared residuals between the observed dispersion curve and the dispersion curve calculated by the model forward modeling. The prior probability distribution integrates stratigraphic information provided by the regional geological map and lithology-velocity relationships revealed by nearby boreholes. The three-dimensional velocity model construction in step S6 adopts a layered medium assumption, with the initial model layer count set to 8 to 15 layers, and the thickness of each layer increasing with depth. An adaptive sampling strategy is employed during the inversion process, with a Markov chain length of 10,000 iterations, a burning period of 2,000 iterations, and a convergence criterion of a Gelman-Rubin statistic less than 1.1. Through MCMC sampling, the algorithm can explore the model parameter space and ultimately obtain the posterior probability distribution, providing not only the optimal model but also quantifying the uncertainty of the model parameters. Furthermore, the inversion process in step S6 introduces lateral constraints, incorporating prior information from adjacent measuring points into the inversion objective function through Kriging interpolation, smoothing lateral velocity variations, reducing the ambiguity of the inversion results, and making it particularly suitable for fine imaging of complex geological structures. (Reference) Figure 2 The figure illustrates the core principle framework of Markov chain Monte Carlo inversion and quality feedback control closed-loop optimization based on Bayesian inference, clearly demonstrating the interaction between prior information, observation data, posterior model, and uncertainty assessment.

[0040] Step S7 implements quality feedback control, which evaluates the quality of dispersion curve extraction and uncertainty of inversion results in real time. When the quality index is lower than the preset threshold, the data acquisition parameters are automatically adjusted or supplementary observations of specific nodes are triggered to form a closed-loop optimization workflow.

[0041] Specifically, in step S7, the quality feedback control system, based on a comprehensive evaluation of multiple quality indicators, including the continuity index of the dispersion curve, the distribution of inversion residuals, and the width of the posterior probability interval of model parameters, automatically generates optimization suggestions when any quality indicator exceeds a preset threshold. These suggestions may include adjusting the observation duration, modifying filter parameters, or adding sensor pairs in specific directions. For example, if the dispersion curve of a certain area shows a break in the high-frequency band, the system may instruct an extension of the data acquisition time for that area; if the inversion results show that the velocity uncertainty of a certain layer is too large, it may suggest densifying the deployment of temporary sensor nodes in that area. This closed-loop mechanism enables the entire exploration process to have adaptive and self-optimizing capabilities.

[0042] To further illustrate the technical effects of this invention, a specific application example is presented. In a geological safety monitoring task along a subway tunnel in a certain city, 49 sensor nodes were deployed using a hexagonal grid with a node spacing of 30 meters, aiming to detect weak interlayers within a depth of 50 meters. The system ran continuously for 35 days, collecting massive amounts of environmental vibration data. Using the method of this invention, edge nodes completed preliminary preprocessing and noise suppression, compressing the original data volume by 92%. After receiving the processed data, the cloud platform performed cross-correlation calculations, spatial coherence analysis, and dispersion feature extraction. Thanks to phase-weighted superposition and time-frequency domain phase-matching tracking technology, even under extremely low traffic noise conditions at night, the system successfully extracted stable dispersion curves in the range of 0.8 Hz to 8 Hz.

[0043] Subsequently, based on the Bayesian MCMC inversion algorithm and incorporating prior information from three nearby boreholes, a high-resolution three-dimensional shear wave velocity model was constructed. The quality feedback system detected slightly higher uncertainty in the model in the region directly above the tunnel, triggering a three-day supplementary observation period, which ultimately reduced the velocity model uncertainty in that area by 40%. The entire data processing was completed within 40 hours, far faster than the three weeks required by traditional methods, providing timely and reliable data support for engineering decisions.

[0044] Furthermore, this invention includes establishing a micro-motion exploration database to store all acquired raw data, intermediate processing results, and the final inversion model. The database adopts a time-series database architecture, supports multi-dimensional data retrieval and visualization, and provides data services to third-party analysis tools through an application programming interface (API). The method is integrated into a cloud-based seismic service platform, providing fully automated services from data acquisition and processing to result display. Users can configure exploration parameters, monitor processing progress, and download analysis reports through a web interface. The processing time for a single exploration data session is reduced from several weeks using traditional methods to less than 48 hours. This combination of database and cloud platform not only ensures the security and traceability of data assets but also lays a solid foundation for subsequent long-term monitoring and big data analysis.

[0045] The above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Therefore, all equivalent changes made in accordance with the structure, shape, and principle of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A seismic service system and data processing method for micro-motion exploration, characterized in that, The method comprises the following specific steps: Step S1: deploying a distributed seismic sensor array, at least 36 three-component broadband seismometers are arranged according to a preset geometric pattern in a target exploration area, each seismometer is provided with a 24-bit analog-to-digital converter and a global positioning system synchronization module, and time synchronization precision between nodes is better than 1 millisecond through a self-organizing wireless sensor network; Step S2: collecting continuous micro-motion signals, continuously recording environmental vibration data for at least 30 days through the distributed seismic sensor array, adopting a circular buffer mechanism for data storage, generating an independent data file every 24 hours, and the file format is an international standard mini-seed format, while real-time monitoring of battery status and storage space usage of each node is performed; Step S3: preprocessing original waveform data, sequentially performing mean removal, trend removal and instrument response correction processing on the collected continuous micro-motion signals, adopting a 4th order Butterworth band-pass filter to retain an effective frequency band of 0.1 Hz to 20 Hz, and automatically detecting and removing abnormal data segments containing obvious human interference or instrument failure; Step S4: calculating a cross-correlation function, cutting a continuous data segment with a length of 1 hour from the preprocessed continuous data, calculating a cross-correlation function for the vertical component records of any two sensor nodes, adopting a time domain normalization processing method to suppress the influence of non-stationary noise, and obtaining an empirical Green function estimation between all sensor pairs; Step S5: extracting dispersion characteristics, performing time-frequency analysis on the cross-correlation function, adopting a multiple window spectral estimation method to calculate a Rayleigh wave phase velocity dispersion curve, a frequency resolution is 0.01 Hz, a base mode dispersion energy is automatically identified in a frequency range of 0.5 Hz to 10 Hz, and a group velocity filtering is used to enhance the signal-to-noise ratio of the dispersion curve; Step S6: constructing a three-dimensional velocity model, based on the extracted dispersion curve, adopting a Markov Chain Monte Carlo inversion algorithm based on Bayesian inference to reconstruct a shear wave velocity structure underground, inversion parameters include layer thickness, layer velocity and layer density, and prior constraints come from regional geological data and drilling data; Step S7: realizing quality feedback control, real-time evaluation of dispersion curve extraction quality and inversion result uncertainty, automatic adjustment of data acquisition parameters or triggering of supplementary observation of a specific node when a quality index is lower than a preset threshold, and forming a closed-loop optimized workflow.

2. The microtremor exploration seismic service system and data processing method according to claim 1, characterized by comprising: The distributed seismic sensor array in step S1 adopts a hexagonal grid arrangement mode, the node spacing is dynamically adjusted according to the detection depth requirement, the shallow layer detection is set to 10 m to 50 m, the medium-deep layer detection is set to 100 m to 500 m, each node is provided with a solar power supply system and a 4G or 5G wireless communication module, and remote state monitoring and data backhaul are realized.

3. The microtremor exploration seismic service system and data processing method according to claim 1, characterized by: The preprocessing process in step S3 further comprises separating noise components of different sources by adopting an independent component analysis algorithm, establishing a feature template library for main interference sources such as traffic vibration, industrial machinery vibration and wind-induced ground vibration, realizing targeted noise suppression through template matching, and the signal-to-noise ratio is improved by more than 10 decibels.

4. The microtremor exploration seismic service system and data processing method according to claim 1, characterized by: The step S4 correlation function calculation uses an improved algorithm based on phase-weighted stacking, which enhances weak signal components by phase-consistency weighting of the correlation results of multiple 1-hour data segments. The weighting factor is proportional to signal coherence, and the effective extraction SNR threshold is reduced to -15 decibels.

5. The microtremor exploration seismic service system and data processing method according to claim 1, characterized by: A spatial coherence analysis step is inserted between steps S4 and S5, which calculates the spatial coherence coefficients of all sensor pairs in multiple frequency bands, constructs a frequency-dependent coherence matrix, identifies the dominant propagation direction through eigenvalue decomposition, and optimizes the weighting strategy of the correlation function based on this.

6. The microtremor exploration seismic service system and data processing method according to claim 1, wherein: The step S5 dispersion feature extraction uses a time-frequency domain phase matching tracking technique, which sparsely represents the signal by constructing a redundant Gabor dictionary, and adaptively selects the optimal time-frequency atom matching the Rayleigh wave dispersion pattern. The frequency estimation accuracy is better than 0.005, and the phase velocity estimation standard deviation is less than 0.

02.

7. The microtremor exploration seismic service system and data processing method according to claim 1, wherein: The step S6 three-dimensional velocity model construction uses a layered medium assumption, with the initial model layer number set to 8-15 layers, and each layer thickness increasing with depth. An adaptive sampling strategy is used in the inversion process, with a Markov chain length of 10,000 iterations and a burn-in period of 2,000 iterations. The convergence criterion is that the Gelman-Rubin statistic is less than 1.

1.

8. The microtremor exploration seismic service system and data processing method according to claim 1, wherein: The step S6 inversion process introduces lateral constraints, which use Kriging interpolation to incorporate prior information from adjacent measurement points into the inversion objective function, smooth lateral velocity variations, and reduce the multi-solution nature of the inversion results.

9. The microtremor exploration seismic service system and data processing method according to claim 1, wherein: The step S7 quality feedback control system is based on the comprehensive evaluation of multiple quality indicators, including dispersion curve continuity indicators, inversion residual distribution, and model parameter posterior probability interval width. When any quality indicator exceeds the preset threshold, the system automatically generates optimization suggestions.

10. The microtremor exploration seismic service system and data processing method according to claim 1, wherein: It also includes establishing a microseismic exploration database to store all collected raw data, processing intermediate results, and final inversion models. The database uses a time series database architecture, supports multi-dimensional data retrieval and visualization, and provides data services for third-party analysis tools through an application programming interface.

Images