A processing method for passive source surface wave dispersion spectrum analysis based on wavelet transform

Abstract

The application provides a processing method for passive source surface wave dispersion spectrum analysis based on wavelet transform, specifically comprising the following steps: S1, adaptive time segmentation cutting: calculating frequency points according to a detection depth range, resolution and estimated surface wave velocity, and calculating corresponding segmentation time according to each frequency point, and cutting original data d into segmented data d m ; S2, data segment frequency processing: performing wavelet convolution frequency division on the segmented data d m ; S3, single-frequency data analysis: evaluating and screening the signal-to-noise ratio of the data after frequency division, and calculating effective phase information P of the data; S4, dispersion spectrum imaging process: calculating phase difference ΔP between each channel according to the effective phase information P of each channel; combining distance difference Δx between two channels, and sequentially calculating phase velocity synthesis f-V two-dimensional dispersion energy spectrum under frequency f. The application effectively guarantees the integrity of low-frequency signals and the stability of high-frequency signals of passive source surface wave data through adaptive time segmentation cutting.

Description

Technical Field

[0001] This invention relates to the field of seismic exploration data processing technology, specifically to a processing method based on wavelet transform for passive source surface wave dispersion spectrum analysis. Background Technology

[0002] In traditional passive source surface wave dispersion spectrum analysis, the data time is often as long as several minutes or even several hours. In order to speed up the data processing, long data is often divided into segments of a fixed length, such as 5 seconds, and then the calculation tasks of the segments are evenly distributed on multiple threads of the computer for parallel computing.

[0003] For surface wave data, low-frequency signals contain wave velocity information from deep underground, but their period is relatively long. Therefore, the vibration signal needs to be segmented into longer time intervals; otherwise, the integrity of the low-frequency portion of the dispersion spectrum will be reduced. High-frequency signals contain wave velocity information from shallow underground layers, but they are more susceptible to environmental noise interference and have higher wavefield randomness, requiring more stacking operations to improve data stability. Therefore, the vibration signal needs to be segmented into shorter time intervals; otherwise, the accuracy of the high-frequency portion of the dispersion spectrum and the limiting resolution of the high-frequency aliasing portion will be reduced.

[0004] In traditional passive source surface wave dispersion spectrum analysis, the time segment is generally of fixed length, with the corresponding sampling point length being a power of 2. This facilitates computer segmentation and parallel computation. However, fixed-length time segments cannot adequately meet the different length requirements for low-frequency and high-frequency data analysis. Specifically:

[0005] Low-frequency signal integrity is low: Fixed-length time segmentation causes severe amplitude and phase distortion in low-frequency data with signal periods close to or longer than the segment length, resulting in a significant reduction in low-frequency data integrity. Low-frequency signals are used in the dispersion spectrum to calculate deep underground wave velocities. When low-frequency signal integrity is compromised, the low-frequency energy clusters in the dispersion spectrum are unclear, making it difficult to extract the dispersion curve, thus reducing the effective detection depth.

[0006] High-frequency signals exhibit low stability: Fixed-length time segments are excessively long for high-frequency signals with periods much shorter than the segment length. This increases the probability of wavefield reversal within long segments, making it difficult to extract stable and effective phase information. Furthermore, with the total data duration remaining constant, excessively long time segments reduce the total number of segments, decreasing the number of superpositions and resulting in a lower signal-to-noise ratio after superposition. This makes it easier for obvious anomalous energy spectra to appear in the dispersion spectrum. High-frequency signals are used in the dispersion spectrum to calculate shallow subsurface wave velocities. When high-frequency signals are unstable, high-frequency energy clusters in the dispersion spectrum are unclear and easily confused with high-frequency aliasing regions, making it difficult to identify surface wave modes and extract dispersion curves, thus increasing the shallow subsurface blind zone.

[0007] Typical detection projects have strict requirements on detection depth and shallow blind zone range, necessitating the use of longer time segments and higher stacking times for data acquisition. This significantly prolongs data acquisition time and reduces detection efficiency. Furthermore, traditional passive source surface wave dispersion spectrum analysis methods calculate all time segments without selection, making it difficult to remove time segments with obviously contaminated data, which also reduces data quality and detection efficiency.

[0008] Therefore, it is evident that improving the quality of passive source surface wave dispersion spectrum data analysis while taking into account the different time length requirements of low-frequency and high-frequency signals is an urgent problem to be solved. Summary of the Invention

[0009] To address the aforementioned technical problems, this invention provides a method for passive source surface wave dispersion spectrum analysis based on wavelet transform, specifically including the following steps:

[0010] S1. Adaptive Time Segmentation: Calculate the frequency points based on the detection depth range, resolution, and estimated surface wave velocity, and calculate the corresponding segment duration based on each frequency point. Then, segment the original data d into segmented data ds. m ;

[0011] S2, Data Segment Frequency Division Processing: The segmented data d... m Perform wavelet convolution frequency integration;

[0012] S3. Single-frequency data analysis: Evaluate and filter the signal-to-noise ratio of the frequency-divided data, and calculate the effective phase information P of the data;

[0013] S4. Dispersion Spectrum Imaging Process: Based on the effective phase information P of each channel, calculate the phase difference ΔP between each channel; combining this with the distance difference Δx between two channels, calculate the phase velocity at frequency f sequentially. Synthesize the fV two-dimensional dispersion energy spectrum.

[0014] Furthermore, S1 also includes:

[0015] S11. Based on the shallowest detection depth d1, the deepest detection depth d2, and the estimated surface wave velocity V... r The lowest frequency f1 and the highest frequency f2 are calculated using the following formula:

[0016]

[0017]

[0018] S12. Based on the number of frequency divisions per octave, divide the lowest frequency f1 and the highest frequency f2 into N equal parts. fThere are several frequencies, and then based on the number of frequencies N... f Calculate each frequency division f k The number of frequencies N f Satisfying the formula:

[0019]

[0020] Where vpo is the number of frequency divisions per octave;

[0021] The frequency division f k Satisfying the formula:

[0022]

[0023] Where f1 is the lowest frequency; vpo is the number of frequency divisions per octave.

[0024] Furthermore, S1 also includes:

[0025] S13, For each frequency division f k The time segment length is set to 5 times the period length, and the number of sampling points L is calculated. k And it satisfies the formula:

[0026]

[0027] Among them, F s The sampling rate of the data;

[0028] S14. Divide the original data d into multiple data of length L. k Segmented data d m The data is output in parallel to each frequency division processing system for processing, and the data output by the frequency division processing system is integrated into effective phase information P, which satisfies the relationship: P = P(f k ).

[0029] Furthermore, S2 also includes:

[0030] S21, divide the segmented data d m Parallel output to each frequency divider f k The corresponding single-frequency data analysis system;

[0031] S22. Integrate the segmented data output by the single-frequency data analysis system into segmented effective phase information P. m The segmented effective phase information P m Satisfying relation: P m =P m (f k ).

[0032] Furthermore, S3 also includes:

[0033] S31. Calculate each of the frequency divisions f. k The corresponding Morse wavelet basis b k ;

[0034] S32, the b k sequentially with the segmented data d m Perform convolution to obtain convolutional data Z mk ;

[0035] S33, convert the convolutional data Z... mk Decomposed into amplitude data A mk and phase data P s Calculate the amplitude data A mk The global root mean square value A k And filter out amplitude data greater than A k Time t mk The time t mk Corresponding phase data P s As valid phase information P mk Output the results.

[0036] Furthermore, if the amplitude data in S33 is less than or equal to A k Then return to step S31.

[0037] Compared with the prior art, the present invention has the following beneficial effects:

[0038] This invention effectively ensures the integrity of low-frequency signals and the stability of high-frequency signals in passive surface wave data through adaptive time segmentation. By decomposing the passive surface wave data using wavelet transform, the original signal can be decomposed into a complex signal, whose amplitude and phase information precisely describes the motion and changes of the wave field, significantly improving the analytical capability of passive surface wave signals. Data processing using a frequency division method facilitates the control of variables to evaluate the global data quality at a specific frequency, thereby simplifying data filtering and improving acquisition efficiency. Parallel processing of segmented and frequency-divided data maximizes the utilization of computational resources, thus increasing computational speed. Attached Figure Description

[0039] Figure 1 This is a flowchart of the present invention;

[0040] Figure 2 This is a flowchart of the adaptive time segmentation process in this invention;

[0041] Figure 3 This is a flowchart of the data segment frequency division processing in this invention;

[0042] Figure 4This is a flowchart of the single-frequency data analysis in this invention;

[0043] Figure 5 This is a flowchart of the dispersive spectrum imaging process in this invention. Detailed Implementation

[0044] To make the technical solutions and effects of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments.

[0045] like Figure 1 As shown, this invention aims to provide a processing method for passive source surface wave dispersion spectrum analysis based on wavelet transform. This method, for signals of different frequencies, performs variable-length decomposition on the original data, and distributes the decomposed time segments to multiple threads for parallel computation and filtering. This approach can accommodate the different time length requirements of low-frequency and high-frequency data, improve data analysis quality, and reduce the total data acquisition time and total data computation time. The method includes an adaptive time segmentation system, one or more data segment frequency division processing systems, one or more single-frequency data analysis systems, and a dispersion spectrum imaging system, wherein:

[0046] The adaptive time segmentation system is responsible for calculating the frequency points based on the detection depth range, resolution, and estimated surface wave velocity, and for calculating the appropriate segment duration for each frequency point. It also divides the original data d into segmented data ds. m .

[0047] The data segmentation frequency division processing system is responsible for processing the segmented data d m Perform wavelet convolution frequency integration.

[0048] The single-frequency data analysis system is responsible for evaluating and filtering the signal-to-noise ratio of the frequency-divided data, and calculating the effective phase information of the data.

[0049] The dispersive spectrum imaging system is responsible for calculating a two-dimensional image of the dispersive spectrum based on the effective phase information of each channel.

[0050] Specifically, the following steps are included:

[0051] Step 1: Adaptive time segmentation, such as... Figure 2 As shown:

[0052] 1. Based on the shallowest detection depth d1, the deepest detection depth d2, and the estimated surface wave velocity V r The lowest frequency f1 and the highest frequency f2 are calculated using the following formula:

[0053]

[0054]

[0055] 2. Based on the number of divisions per octave, divide the lowest frequency f1 and the highest frequency f2 into N equal parts after taking the logarithm. f There are several frequencies, and then based on the number of frequencies N f Calculate each frequency division f k Where, the number of frequencies N f Satisfying the formula:

[0056]

[0057] Where vpo is the number of frequency divisions per octave;

[0058] Frequency division f k Satisfying the formula:

[0059]

[0060] Where f1 is the lowest frequency; vpo is the number of frequency divisions per octave.

[0061] 3. For each frequency division f k The time segment length is set to 5 times the period length, and the number of sampling points L is calculated. k L k Satisfying the formula:

[0062]

[0063] Among them, F s The sampling rate of the data;

[0064] 4. Divide the original data d into multiple data of length L. k Segmented data d m The data is output in parallel to each frequency division processing system for processing, and the data output from the frequency division processing system is integrated into effective phase information P. The effective phase information P satisfies the relationship: P = P(f k ).

[0065] Step 2: Data segment frequency division processing, such as... Figure 3 As shown:

[0066] 1. Divide the segmented data d m Parallel output to each frequency divider f k The corresponding single-frequency data analysis system;

[0067] 2. Integrate the segmented data output from the single-frequency data analysis system into segmented effective phase information P. m Segmented effective phase information P m Satisfying relation: P m =P m (f k ).

[0068] Step 3: Single-frequency data analysis, such as... Figure 4 As shown:

[0069] 1. Calculate each frequency division f k The corresponding Morse wavelet basis b k ;

[0070] 2. b k sequentially with segmented data d m Perform convolution to obtain convolutional data Z mk ;

[0071] 3. Convert the convolutional data Z... mk Decomposed into amplitude data A mk and phase data P s Calculate amplitude data A mk The global root mean square value A k And filter out amplitude data greater than A k Time t mk , time t mk Corresponding phase data P s As valid phase information P mk Output the data. If the amplitude data is less than or equal to the threshold A. k Then return to calculating each frequency division f k The corresponding Morse wavelet basis b k .

[0072] Step 4, the dispersive spectral imaging process, as follows: Figure 5 As shown:

[0073] 1. Calculate the phase difference ΔP between each channel based on the effective phase information P of each channel;

[0074] 2. Combining the distance difference Δx between the two paths, calculate the phase velocity at frequency f.

[0075] 3. Synthesize the fV two-dimensional dispersion energy spectrum.

[0076] This invention effectively ensures the integrity of low-frequency signals and the stability of high-frequency signals in passive surface wave data through adaptive time segmentation. By decomposing the passive surface wave data using wavelet transform, the original signal can be decomposed into a complex signal, whose amplitude and phase information precisely describes the motion and changes of the wave field, significantly improving the analytical capability of passive surface wave signals. Data processing using a frequency division method facilitates the control of variables to evaluate the global data quality at a specific frequency, thereby simplifying data filtering and improving acquisition efficiency. Parallel processing of segmented and frequency-divided data maximizes the utilization of computational resources, thus increasing computational speed.

[0077] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A processing method for passive source surface wave dispersion spectrum analysis based on wavelet transform, characterized in that, Specifically, the following steps are included: S1. Adaptive Time Segmentation: Calculate the frequency points based on the detection depth range, resolution, and estimated surface wave velocity, and calculate the corresponding segment duration based on each frequency point. Then, convert the original data... Segmented data ;in: S11, Based on the shallowest detection depth Deepest Detection Depth And the estimated surface wave velocity Calculate the lowest frequency and highest frequency The calculation formula is as follows: S12. Based on the number of frequency divisions per octave, divide the lowest frequency... and highest frequency After taking the logarithm, the average is divided into One frequency, and then according to the number of frequencies Calculate each frequency division The number of frequencies Satisfying the formula: in, The number of frequency divisions per octave; The frequency division Satisfying the formula: in, The lowest frequency; The number of frequency divisions per octave; S2, Data Segment Frequency Division Processing: This involves processing the segmented data... Perform wavelet convolution frequency integration; S3. Single-frequency data analysis: Evaluate and filter the signal-to-noise ratio of the frequency-divided data, and calculate the effective phase information of the data. ; S4. Dispersion Spectrum Imaging Process: Based on the effective phase information of each channel... Calculate the phase difference between each channel. Combine the distance difference between the two tracks Calculate the frequency in sequence Phase velocity below ;synthesis - Two-dimensional dispersion energy spectrum.

2. The processing method for passive source surface wave dispersion spectrum analysis based on wavelet transform according to claim 1, characterized in that, S1 also includes: S13, For each frequency divider The time segment length is set to 5 times the period length, and the number of sampling points is calculated. And it satisfies the formula: in, The sampling rate of the data; S14, Transfer the original data Cut into multiple lengths Segmented data The data is output in parallel to each frequency division processing system for processing, and the data output by the frequency division processing system is integrated into effective phase information. The effective phase information Satisfying Relationship: .

3. The processing method for passive source surface wave dispersion spectrum analysis based on wavelet transform according to claim 1, characterized in that, S2 also includes: S21. The segmented data Parallel output to each frequency divider The corresponding single-frequency data analysis system; S22. Integrate the segmented data output by the single-frequency data analysis system into segmented effective phase information. The segmented effective phase information Satisfying Relationship: .

4. The processing method for passive source surface wave dispersion spectrum analysis based on wavelet transform according to claim 3, characterized in that, S3 also includes: S31. Calculate each of the frequency divisions. Corresponding Morse wavelet basis ; S32, the aforementioned sequentially with the segmented data Convolution is performed to obtain convolutional data. ; S33, the convolutional data Decomposed into amplitude data and phase data Calculate the amplitude data global root mean square value And filter out amplitude data greater than time The time Corresponding phase data As valid phase information Output the results.

5. The processing method for passive source surface wave dispersion spectrum analysis based on wavelet transform according to claim 4, characterized in that, If the amplitude data in S33 is less than or equal to the Then return to step S31.

Images