CN-121995494-A - Data driving determination method for short-period surface wave Gaussian filter parameters

CN121995494ACN 121995494 ACN121995494 ACN 121995494ACN-121995494-A

Abstract

The invention discloses a data driving determination method of short period surface wave Gaussian filter parameters, which comprises the steps of obtaining background noise data recorded by a plurality of seismic stations distributed in a target area and preprocessing, forming a station pair by every two seismic stations, obtaining an empirical green function of each station pair based on the preprocessed background noise data, selecting a plurality of periodic points in a preset periodic range, scanning each periodic point in a preset Gaussian filter window parameter numerical range, filtering each scanned parameter, calculating a corresponding signal-to-noise ratio of the empirical green function, generating a response curve of the signal-to-noise ratio changing along with Gaussian filter window parameters for each station pair based on the combination of the signal-to-noise ratio and each periodic point, identifying a global maximum point of the signal-to-noise ratio in each response curve, and recording the Gaussian filter window parameter corresponding to the global maximum point as an optimal parameter. The method can quickly and effectively determine the optimal filtering parameters.

Inventors

PANG GUANGHUA
JIANG HAIYU
XUE PENG
ZHANG NIAONA

Assignees

长春工业大学

Dates

Publication Date: 20260508
Application Date: 20260205

Claims (10)

1. A data-driven determination method for short-period surface wave gaussian filtering parameters, comprising: Acquiring background noise data recorded by a plurality of seismic stations arranged in a target area, and preprocessing the background noise data, wherein each two seismic stations form a station pair; Based on the preprocessed background noise data, acquiring an empirical green function of each station pair; Selecting a plurality of periodic points in a preset periodic range according to an empirical green function of each station pair; for each periodic point, scanning in a preset Gaussian filter window parameter value range with a preset step length, filtering the empirical green function by adopting a multiple filtering method for each scanned Gaussian filter window parameter, and calculating the signal-to-noise ratio of the filtered signal; Generating a response curve of the signal-to-noise ratio along with the Gaussian filter window parameter change for the combination of each station pair and each period point based on the signal-to-noise ratio; And identifying global maximum points of the signal to noise ratio in each response curve, and recording Gaussian filter window parameters corresponding to each global maximum point as optimal parameters of the current station pair at the current period point.
2. The data-driven determination method of short-period surface wave gaussian filter parameters according to claim 1, wherein said preprocessing comprises de-instrumental response, de-averaging, de-trending and band-pass filtering.
3. The method for determining data driving parameters of short periodic surface wave gaussian filtering according to claim 1, wherein said obtaining an empirical green function for each station pair based on the preprocessed background noise data specifically comprises: (1) Seismic station based on pretreatment With seismic stations Recorded background noise data, a single-segment cross-correlation function is calculated: Wherein, the Representing a single-segment cross-correlation function; A duration representing a background noise data segment; Representing a segment start time; representing seismic stations At the time of Background noise data recorded at that time; representing seismic stations At the time of Background noise data recorded at the time, and ; (2) Superposing a plurality of single-section cross-correlation functions to obtain a cross-correlation function: (3) Obtaining a time-reversed version of the cross-correlation function, recording it as a negative cross-correlation function, and superposing the cross-correlation function and the negative cross-correlation function to obtain a seismic station With seismic stations Empirical green's function of the composed station pairs.
4. The data-driven determination method of short-period surface wave gaussian filter parameters according to claim 1, wherein the preset period range is 3-14 seconds.
5. The method for determining the data driving of the short period surface wave gaussian filter parameter according to claim 1, wherein the preset step size is 0.001, and the preset gaussian filter window parameter value range is [0.001,20].
6. The method for determining the data driving of the short periodic surface wave Gaussian filter parameters according to claim 1, further comprising performing parameter quality control on all optimal parameters, specifically comprising eliminating Gaussian filter window parameters smaller than a first threshold value in all optimal parameters, and/or eliminating outlier data in all optimal parameters based on statistical distribution.
7. The data-driven determination method of short-period surface wave gaussian filter parameters according to claim 1 or 6, further comprising: And integrating all the optimal parameters to construct a parameter application structure for determining Gaussian filter window parameters, wherein the parameter application structure is used for outputting the corresponding Gaussian filter window parameters according to the designated station pair and the periodic point when extracting the short-period-surface wave dispersion curve aiming at the target area.
8. The method of claim 7, wherein the parameter application structure is a lookup table, and the lookup table stores the corresponding optimal parameters by using the station pair identifier and the periodic point as a joint query key.
9. The method of claim 8, wherein the parameter application structure further comprises a region fitting model, and the region fitting model is a quantitative relation model between the average value and the period point established by mathematical fitting based on the average value of the optimal parameters of all station pairs in each period in the lookup table.
10. The data-driven determination method of short-period-surface-wave gaussian filter parameters according to claim 9, wherein when a gaussian filter window parameter is determined by applying said parameter application structure, the following hierarchical rule is adopted: the first rule is that the accurate matching item of the current station pair and the target periodic point is inquired from the lookup table, and the corresponding optimal parameter is directly read; If the rule cannot determine the parameters, carrying out interpolation calculation based on Gaussian filter window parameters of adjacent periods in the lookup table to determine optimal parameters; And thirdly, if the first rule and the second rule can not determine the parameters, calculating by adopting a region fitting model to obtain an optimal parameter estimation value.

Description

Data driving determination method for short-period surface wave Gaussian filter parameters Technical Field The invention relates to the technical field of geophysical exploration and seismic signal processing, in particular to a data driving determination method for Gaussian filtering parameters of short-period surface waves. Background In background noise plane wave imaging, multiple Filtering (MFT) is a key technique for extracting dispersion curves. The method is characterized in that a Gaussian filter is utilized to perform time-frequency analysis, the value of a Gaussian filter window parameter a directly determines the time-frequency resolution, and finally the extraction quality of a dispersion curve is affected. Currently, the setting of the parameter a generally depends on an empirical formula such as a fixed value (e.g., a=5) or a linear variation with the period T (e.g., a=0.2t+3). These methods imply an insufficiently verified preset that there is a simple, generic functional relationship between the parameter a and the period T, and that this relationship applies to all pairs of stations and all geological areas. However, the surface wave propagation characteristics are significantly affected by regional geologic structures and specific propagation paths, short-period surface waves being particularly sensitive to local inhomogeneities in shallow structures. Thus, fixed or linear empirical parameters cannot adapt to macroscopic regional differences nor respond to microscopic path-specific changes, resulting in parameter selection and physical discontinuities. In addition, the prior art ignores an optimization feature implicit in the physical properties of the signal itself, namely that the signal-to-noise ratio-parameter (SNR-a) response curve of the short period surface wave signal has a unimodal extremum. The applicant has found through a number of experiments that for short period (3-14 seconds) surface wave signals, the SNR-a curve exhibits identifiable peaks corresponding to the optimal energy aggregation state of the signal in the time-frequency domain. However, as the period increases (typically >14 seconds), this peak characteristic fades away. The prior art loses a definite physical target in the parameter optimization process due to the fact that the objective physical characteristics are not recognized and utilized, so that the prior art stays in an empirical and systematic suboptimal state for a long time, and the final signal-to-noise ratio potential of each propagation path cannot be self-adaptively mined. In view of the above limitations, the prior art is in a systematic suboptimal state for a long period of time when processing short-period surface wave signals. The highest signal-to-noise ratio potential which can be achieved by the measured data cannot be fully mined through parameter tuning, so that the reliability of high-resolution shallow structure imaging is restricted. Therefore, how to overcome the defects of poor adaptability and optimized physical characteristics of neglected signals of the existing empirical parameter setting method, develop a data driving scheme capable of adaptively determining optimal filtering parameters and systematically improving the extraction quality of the short-period-surface wave dispersion curve, and solve the problems of the technicians in the field. Disclosure of Invention In view of the above, the present invention proposes a data-driven determination method of short-period surface wave gaussian filter parameters in order to overcome or at least partially solve the above-mentioned problems. In order to achieve the above purpose, the present invention adopts the following technical scheme: In a first aspect, the present invention provides a data-driven determination method for a short-period surface wave gaussian filter parameter, including: Acquiring background noise data recorded by a plurality of seismic stations arranged in a target area, and preprocessing the background noise data, wherein each two seismic stations form a station pair; Based on the preprocessed background noise data, acquiring an empirical green function of each station pair; Selecting a plurality of periodic points in a preset periodic range according to an empirical green function of each station pair; for each periodic point, scanning in a preset Gaussian filter window parameter value range with a preset step length, filtering the empirical green function by adopting a multiple filtering method for each scanned Gaussian filter window parameter, and calculating the signal-to-noise ratio of the filtered signal; Generating a response curve of the signal-to-noise ratio along with the Gaussian filter window parameter change for the combination of each station pair and each period point based on the signal-to-noise ratio; And identifying global maximum points of the signal to noise ratio in each response curve, and recording Gaussian filter window parameters correspon