Passive source surface wave data screening and imaging method based on noise wave field feature analysis

Abstract

The invention provides a passive source surface wave data screening and imaging method based on noise wave field feature analysis, and the method comprises the steps: carrying out the preprocessing of an original passive source noise record, and carrying out the segmentation of the processed original data; performing statistics on the RMS of the segmented original record according to rows, and calculating a variable coefficient of each data segment about an RMS curve; setting a threshold value according to an RMS curve variable coefficient, and screening data segments; setting a threshold value to distinguish a data segment beneficial to imaging from a data segment interfering imaging by using the difference of the variable coefficients; and performing interference superposition processing on the screened data segments to obtain a high-quality surface wave signal. According to the method, effective surface wave signals can be extracted from short noise records through analysis and statistics of wave field features, the noise data duration requirement is reduced, the acquisition cost is reduced, and the noise data processing efficiency is improved. According to the method, noise data types can be effectively screened, surface wave signals with higher quality are extracted, and the signal-to-noise ratio of an imaging result is improved.

Description

Technical Field

[0001] The invention provides a passive surface wave data screening and imaging method based on noise wave field characteristic analysis, belonging to the technical field of geological exploration. Background Art

[0002] Surface wave exploration is a shallow surface exploration method that has developed rapidly in recent years. It can be divided into active surface wave exploration and passive surface wave exploration. Passive surface wave exploration can extract reliable surface wave signals from long-term background noise records to invert the shallow shear wave velocity structure, thereby realizing accurate analysis and evaluation of shallow geological structures and physical properties. Due to the advantages of passive surface wave exploration such as convenient operation, low interference, high detection accuracy and low economic cost, its application is becoming more and more widespread and has become an indispensable mainstream method in shallow geological surveys such as cities. In urban environments, traffic noise is an important type of passive surface wave data. It usually contains strong surface wave information and is easy to obtain. It has attracted extensive attention and research.

[0003] The key to passive surface wave imaging lies in data processing. First, the long-term data is segmented and interfered, and then the interference results are superimposed to obtain effective surface wave information. The traditional method is only to perform simple segmentation, interference and superposition processing on the data. This process requires a long period of noise recording (several hours or days) to obtain effective surface wave information, and the extracted surface wave signal has a relatively low signal-to-noise ratio. However, since the signal-to-noise ratio of different data segments often varies greatly, the imaging effect is also very different. If only the data segments with good signal-to-noise ratio are interfered and superimposed, not only will the imaging quality be greatly improved, but the data acquisition time and cost can also be reduced. If the data segments with poor signal-to-noise ratio are included in the superposition calculation, not only will the imaging effect be reduced, but the noise in them may also submerge the weak effective signal. Therefore, in the process of passive surface wave imaging, effective screening of the segmented data segments is a key point and a difficulty.

[0004] In recent years, scholars and experts have conducted many effective studies on the problem of passive surface wave data screening. Most of the research ideas are to convert noise records from the time-space domain to other domains. For example, Fourier transform is used to convert it to the frequency-wavenumber domain (fk), or Tau-p transform is used to convert it to the Tau-p domain, and then the difference between the effective signal and noise in other domains is used to distinguish and separate them accordingly. Relying on this method, noise and effective signals can be identified to a certain extent, and noise data segments with better signal-to-noise ratio can be retained, and data with poor signal-to-noise ratio can be eliminated.

[0005] Using methods such as f-k transformation or Tau-p transformation to transform the original data into other domains for signal identification can, to a certain extent, screen data segments, but it ignores some characteristics of the noise data itself. Especially for traffic noise, since the acquisition array is usually linearly arranged along the road. There is a certain positional relationship between the geophones and the passive sources (vehicles), so the wave field characteristic transformation is relatively obvious and there are certain rules to follow. However, the above methods do not fully identify and analyze this regular characteristic, and the intermediate calculation is relatively cumbersome when screening and identifying after data conversion, and the computational complexity of this type of method is relatively large.

[0006] In addition, for the processing of passive source surface wave data, an important step is to segment the noise data, then perform interference calculation on it, superimpose the interference results, and finally obtain surface wave information. To improve the signal-to-noise ratio of the final superimposed result, many scholars have proposed screening and superimposing the interference results. Different from the aforementioned methods, this type of method does not screen the original noise data segments, but only screens the interference results of each data segment, and then retains the results with better signal-to-noise ratio. This method can, to a certain extent, improve the effect of passive source surface wave imaging.

[0007] The above method screens and processes the interference results, retains the results with higher signal-to-noise ratio for superimposition, and can obtain effective surface wave information. However, this method requires interference processing of all the original data segments, which is relatively time-consuming, and this method requires noise records with long-term recordings to extract higher-quality surface wave signals, which is not only time-consuming in processing, but also increases the acquisition cost. Summary of the Invention

[0008] The passive source surface wave imaging technology not only has high imaging accuracy but also can achieve non-destructive exploration, which is undoubtedly the preferred choice for exploring urban underground structures. Many scholars have conducted corresponding research on how to improve the passive source surface wave imaging accuracy, and one of the more important aspects is how to extract higher-quality surface wave signals from noise records. Taking actual traffic noise as an example, the present invention establishes a set of passive source surface wave data screening and imaging methods based on noise wave field characteristic analysis. By analyzing the wave field characteristics of noise records, using statistical methods to analyze the wave field changes of noise data segments, preferentially selecting data segments that contribute well to surface wave imaging, then performing interference calculation on them, and finally superimposing to obtain higher-quality surface wave signals. This method can not only extract effective surface wave signals from shorter noise records, thereby shortening the acquisition cost, but also effectively improve the surface wave imaging accuracy and extract higher-quality surface wave signals.

[0009] The complete technical solution provided by the present invention:

[0010] A passive source surface wave data screening and imaging method based on noise wave field characteristic analysis, comprising the following steps:

[0011] Step 1: Preprocess the original passive source noise record, and then cut and segment the processed original data.

[0012] Step 2: Statistically calculate the root mean square value RMS of the segmented original record row by row, and calculate the coefficient of variation of each data segment;

[0013] Using the method of sampling statistics, the RMS value distribution of the original data in the time direction is statistically obtained. After smoothing the discrete points, the RMS curve is obtained, and then the coefficient of variation of the curve is calculated. The coefficient of variation obtained by statistics is used to reflect the discrete situation of the RMS curve, so as to judge the change of the corresponding original data wave field characteristics.

[0014] The implementation process is as follows:

[0015] S2.1. Let each segment of the original data be data m (i, j), where m represents dividing the original data into m segments, and i and j represent that each segment of data is a two-dimensional array of i rows * j columns. Calculate the root mean square value RMS of each row of the original record horizontally, and its calculation formula is as follows:

[0016]

[0017] Among them: N is the number of data points in each row, x i,j is the data at the i-th row and j-th column, and RMS i represents the root mean square value at the i-th moment.

[0018] S2.2. Calculate the root mean square value of each row. Each segment of data obtains i root mean square values. Then obtain the change trend of the statistically calculated RMS value. Remove some noises or short-term fluctuations through smoothing to obtain the change of the RMS value. The specific calculation formula is as follows:

[0019]

[0020] Among them: RMS smoothed,k represents the smoothed RMS value, k represents the current sliding window position, w represents the smoothing window size, represents the left and right half widths of the window, rounded down;

[0021] S2.3. Calculate and smooth the RMS of the original noise data segment to obtain the corresponding curve. The characteristics of the smoothed RMS curve correspond to the wave field characteristics of the original data segment. Use the coefficient of variation of the RMS curve to describe the change of the wave field characteristics of the original data.

[0022] Calculate the standard deviation of the curve, which reflects the overall fluctuation size of the curve relative to the mean value. The specific calculation formula is as follows:

[0023]

[0024] Where: μ is the mean of the curve, and n is the number of samples.

[0025] S2.4. According to the standard deviation to reflect the fluctuation of the curve, the greater the fluctuation, the more obvious the change in the wave field of the original data, and the smaller the fluctuation, the smaller the change in the wave field of the original data, and the more stable the surface wave propagation. Calculate the coefficient of variation based on the standard deviation, that is, the ratio of the standard deviation to the mean. Use the magnitude of the coefficient of variation to directly reflect the fluctuation of the curve. The specific calculation formula is as follows:

[0026]

[0027] Where: CV is the coefficient of variation of each data segment with respect to the RMS curve, std is the standard deviation of the curve, and μ is the mean of the curve. If the CV value is larger, it indicates that the curve fluctuates more strongly, that is, the wave field of the original data changes violently. If the CV value is smaller, it indicates that the curve fluctuates less, that is, the corresponding wave field of the original data changes less, and the signal propagation is relatively stable.

[0028] Step 3: Set a threshold according to the coefficient of variation of the RMS curve to screen data segments. Using the difference in the coefficient of variation, setting a reasonable threshold will help distinguish the data segments for imaging from the interference data segments and improve the imaging quality. Specifically, it is expressed as follows:

[0029]

[0030] Where Data i is the data segment retained after screening, CV is the coefficient of variation of each data segment with respect to the RMS curve, and T is the threshold.

[0031] Step 4: Interference superposition calculation.

[0032] By performing interference processing on the noise records of two receivers to reconstruct the empirical Green's function between two points, an effective signal can be obtained.

[0033] In this invention, the cross-correlation interference method is mainly adopted, and at the same time, the amplitudes of each frequency are normalized to broaden the effective frequency band. The specific expression formula is as follows:

[0034]

[0035] Where: u(A, s, ω) represents the frequency-domain signal received by the receiver at A when the source is excited at S.

[0036] u(B, s, ω) represents the frequency-domain signal received by the receiver at B when the source is excited at S.

[0037] Among them, H AB (ω) represents the surface wave signal propagated from point A to point B extracted through mutual interference.

[0038] The passive source surface wave data screening and imaging method based on noise wave field feature analysis proposed by the present invention can achieve the following benefits:

[0039] 1. By analyzing and statistically analyzing the wave field characteristics, effective surface wave signals can be extracted from short noise records, further reducing the required duration of noise data, thereby reducing the acquisition cost and improving the processing efficiency of noise data.

[0040] 2. Using the method proposed by the present invention can effectively screen the types of noise data, extract higher-quality surface wave signals, and improve the signal-to-noise ratio of the imaging results. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 It is a record diagram of the background noise generated by a high-speed train in the original one minute of the embodiment;

[0042] Figure 2 It is a schematic diagram of the segmentation of Section 1 of the embodiment;

[0043] Figure 3 It is a schematic diagram of the segmentation of Section 2 of the embodiment;

[0044] Figure 4 It is a schematic diagram of the segmentation of Section 3 of the embodiment;

[0045] Figure 5 It is a schematic diagram of the segmentation of Section 4 of the embodiment;

[0046] Figure 6 It is a schematic diagram of the segmentation of Section 5 of the embodiment;

[0047] Figure 7 It is a schematic diagram of the segmentation of Section 6 of the embodiment;

[0048] Figure 8 It is an RMS statistical chart of Section 1 of the original noise data of the embodiment;

[0049] Figure 9 It is an RMS statistical chart of Section 2 of the original noise data of the embodiment;

[0050] Figure 10 It is an RMS statistical chart of Section 3 of the original noise data of the embodiment;

[0051] Figure 11 It is an RMS statistical chart of Section 4 of the original noise data of the embodiment;

[0052] Figure 12 It is an RMS statistical chart of Section 5 of the original noise data of the embodiment;

[0053] Figure 13 RMS statistical chart of 6 segments of the original noise data of the embodiment;

[0054] Figure 14 Interference result diagram of 1 segment of the noise data of the embodiment;

[0055] Figure 15 Interference result diagram of 2 segments of the noise data of the embodiment;

[0056] Figure 16 Interference result diagram of 3 segments of the noise data of the embodiment;

[0057] Figure 17 Interference result diagram of 4 segments of the noise data of the embodiment;

[0058] Figure 18 Interference result diagram of 5 segments of the noise data of the embodiment;

[0059] Figure 19 Interference result diagram of 6 segments of the noise data of the embodiment;

[0060] Figure 20 Processing result diagram of the noise example by the traditional method;

[0061] Figure 21 Processing effect diagram of the actual example of the method of the present invention;

[0062] Figure 22 Processing flow chart of the present invention. Detailed implementation manners

[0063] The specific technical solutions of the present invention are described in conjunction with the embodiments.

[0064] As Figure 22 shown, the passive source surface wave data screening and imaging method based on the analysis of the noise wave field characteristics mainly includes the following steps:

[0065] Step 1: Preprocess the original passive source noise record (conventional processing: filtering, removing instrument response, removing bad channels, etc.), and then cut and segment the processed original data. Generally, the original noise record of dozens of minutes or several hours is segmented into data segments of several seconds.

[0066] As Figures 1 to 7 shown, Figure 1 is the background noise record generated by the high-speed train in the original one minute. It can be seen from Figure 1 that there are two very strong energy axes (white dotted lines), which indicates that there are two high-speed train passing events in the record within this one minute. The original high-speed train record is split into multiple data segments of 4 seconds each (a total of 15 segments). The schematic diagrams of the first 6 segments are Figure 2 , Figure 3 , Figure 4 ,Figure 5 , Figure 6 and Figure 7 as shown. From the first three data segments ( Figure 2 , Figure 3 and Figure 4 ), it can be seen that the original record contains obvious surface wave signals propagating steadily (within the range of the red dashed line), which indicates that the high-speed train is gradually approaching the geophone array from the far side. This type of data segment belongs to the data segments that contribute more to passive source surface wave imaging. For the latter three data segments ( Figure 5 , Figure 6 and Figure 7 ), it can be clearly seen that the wave field is relatively chaotic and there are signals in two directions. This indicates that the train has entered the range of the geophone array and the train source is close to the array. At this time, the surface waves have not fully developed, and this type of data segment belongs to the data segments that contribute less to passive source surface wave imaging, but it contains stronger other energies. If such data segments are involved in the calculation during the interference superposition process, it will bring more noise interference to the imaging result, thus reducing the quality of the extracted surface wave signals.

[0067] Step 2: Statistically analyze the RMS (root mean square value) of the segmented original record, and calculate the coefficient of variation of each data segment (used to reflect the characteristics of the original data wave field). The method of this invention samples the statistical method to statistically analyze the RMS value of the original data to obtain the RMS value distribution in the time direction, smooth the discrete points to obtain the RMS curve, and then calculate the coefficient of variation of the curve. Use the statistically obtained coefficient of variation (degree of dispersion) to reflect the discrete situation of the RMS curve, so as to judge the change situation of the corresponding original data wave field characteristics. The implementation process is as follows:

[0068] S2.1. Assume that each segment of the original data is data m (i,j), m represents dividing the original data into m segments, and i, j represent that each segment of data is a two-dimensional array of i rows * j columns. Calculate the root mean square value (RMS) of each row of the original record row by row (horizontally), and its calculation formula is as follows:

[0069]

[0070] where: N is the number of data points in each row, x i,j is the data of the i-th row and j-th column, and RMS i represents the root mean square value at the i-th moment.

[0071] S2.2. Calculate the root mean square value of each row through the above formula, so that each segment of data can obtain i root mean square values. Then obtain the changing trend of the statistically obtained RMS values. Remove some noise or periodic fluctuations through smoothing to obtain the change situation of the RMS values, and the specific calculation formula is as follows:

[0072]

[0073] Among them: RMS smoothed,k represents the smoothed RMS value, k represents the current sliding window position, w represents the smoothing window size, represents the left and right half widths (rounded down) of the window.

[0074] S2.3. After calculating and smoothing the RMS of the original noise data segment according to the above formula, the corresponding curve can be obtained. The characteristics of the smoothed RMS curve correspond to the wave field characteristics of the original data segment. In order to better reflect the wave field characteristics of different original data, the present invention uses the coefficient of variation of the RMS curve to describe the change of the wave field characteristics of the original data. First, calculate the standard deviation of the curve, which reflects the overall fluctuation size of the curve relative to the mean. The specific calculation formula is as follows:

[0075]

[0076] Among them: μ is the mean of the curve, and n is the number of samples.

[0077] S2.4. According to the standard deviation reflecting the fluctuation of the curve, the greater the fluctuation, the more obvious the wave field change of the original data, and the smaller the fluctuation, the smaller the wave field transformation of the original data and the more stable the surface wave propagation. Although the standard deviation can reflect the fluctuation size of the curve, it is impossible to further quantify the change of the original data. The present invention further calculates its coefficient of variation on the basis of the standard deviation, that is, the ratio of the standard deviation to the mean. Use the size of the coefficient of variation to directly reflect the fluctuation of the curve. The specific calculation formula is as follows:

[0078]

[0079] Among them: std is the standard deviation of the curve, and μ is the mean of the curve. If the CV value is larger, it means the curve fluctuation is stronger, that is, the wave field of the original data changes violently. If the CV value is smaller, it means the curve fluctuation is smaller, that is, the corresponding wave field of the original data changes less and the signal propagation is relatively stable.

[0080] Such as Figures 8 to 13 shown, the blue dots in the figure are the RMS of the original noise data segment calculated by formula 1, and the red curve is the RMS curve calculated and smoothed by formula 2. Figures 8 to 13 The relevant results of Figures 2 to 7 are calculated from the data segment of Figure 8 . Figure 9 and Figure 10 The RMS curves (red) in have relatively small fluctuations, and their coefficient of variation values are: 0.080; 0.086; 0.089. The RMS is mainly concentrated in (1 - 4) * 10 -3The corresponding original noise data segment Figure 2 , Figure 3 and Figure 4 It can also be clearly seen that the surface wave propagation of this type of data segment is stable and other coherent noises are relatively small. And Figure 11 , Figure 12 and Figure 13 have stronger fluctuations in the RMS curve, and the values of their coefficient of variation are: 0.289; 0.261; 0.124. The RMS is mainly concentrated in (6 - 10) * 10 -3 . In the corresponding original data segment, it can also be seen that the wave field in this type of data segment changes significantly and the amplitude energy changes violently. At this time, the surface wave has not fully developed, but it has very strong energy. If it is included in the interference superposition calculation, the signal-to-noise ratio of the superposition result will be seriously reduced.

[0081] Step 3: Set a threshold according to the coefficient of variation of the RMS curve to screen data segments. According to Figures 8 to 13 the calculation results, it can be seen that when the wave field of the original data propagates smoothly, the coefficient of variation of this type of data segment is relatively small. When the wave field changes greatly and the amplitude decays violently, the coefficient of variation of the RMS curve is relatively large at this time, and the RMS value is also large. Using the difference in the coefficient of variation, setting a reasonable threshold can distinguish the data segments beneficial to imaging from the interference data segments, so as to improve the imaging quality. Specifically expressed as follows:

[0082]

[0083] where Data i is the data segment screened and retained, CV is the coefficient of variation of each data segment with respect to the RMS curve. T is the threshold, and the specific threshold value is selected according to the signal-to-noise ratio of the actual data.

[0084] Step 4: Interference superposition calculation. Seismic interferometry is a new method widely used in ambient noise imaging in recent years. This method reconstructs the empirical Green's function between two points by performing interference processing on the noise records of two receivers (seismic traces), so as to obtain an effective signal. There are three commonly used seismic interferometry methods: cross-correlation, deconvolution, and cross-coherence. In this invention, the cross-coherence interferometry method is mainly used this time, which can eliminate the influence of the source wavelet. At the same time, by normalizing the amplitudes of each frequency, the effective frequency band is broadened. The specific expression formula is as follows:

[0085]

[0086] where: u(A, s, ω) represents the frequency-domain signal excited by the source at S and received by the receiver at A.

[0087] u(B, s, ω) represents the frequency-domain signal excited by the source at S and received by the receiver at B.

[0088] Where H AB (ω) represents the surface wave signal propagated from point A to point B extracted by mutual interference.

[0089] Using the mutual interference formula to perform interference calculation on the original data segment ( Figures 2 to 7 ) to obtain the interference result ( Figures 14 to 19 ). As Figures 14 to 16 It can be seen that the surface wave signal has a good signal-to-noise ratio, which is also consistent with the previous analysis of the original noise data segment. In the original data segment ( Figures 2 to 4 ) obvious characteristics of a stable propagating wave field can be seen, and the coefficient of variation of the RMS curve ( Figures 8 to 10 ) calculated in step 2 is also small. From the original data to the interference result, it is also proved that this type of data segment is effective for surface wave imaging. By setting a threshold according to the coefficient of variation, it can be screened and retained. From Figures 17 to 19 of the interference result, it can be seen that in addition to a small amount of surface wave signals, there is more noise interference. The fundamental reason for this result is that the wave field characteristics in the original data segment ( Figures 5 to 7 ) are complex, the surface waves have not fully developed, but the noise energy is very strong at this time. From the perspective of interference superposition, if such signals are included in the superposition calculation, it will lead to a poor signal-to-noise ratio of the final result, which is not conducive to the calculation and inversion of the dispersion curve in the later stage. The coefficient of variation calculated according to the RMS curve of the original data also shows that the coefficient of variation of this type of data segment is large, indicating that the wave field of this type of data changes violently. Similarly, corresponding thresholds can be set according to step 3 to screen and eliminate such data segments.

[0090] Figure 20 For the processing result of the noise example by the traditional method, this method does not analyze the wave field characteristics of the noise data and directly performs segmented interference superposition on the data in a simple and crude manner. From the result, it can be seen that the signal-to-noise ratio of the extracted surface wave signal is poor, the noise interference is very strong, and the imaging result is also poor. According to the passive source surface wave data screening and imaging method based on noise wave field characteristic analysis provided by the present invention, the original data can be processed and analyzed first. Specifically, first segment the data, and then use mathematical statistics methods to calculate the RMS of the segmented data and statistically obtain the coefficient of variation of the RMS curve. According to the statistical result, set the screening threshold to 0.010. Using this method, data segments with obvious changes in wave field characteristics and large noise content in the original data segment can be eliminated. Finally, a surface wave signal with a good signal-to-noise ratio is obtained through mutual interference superposition. According to the processing result ( Figure 21 ), the signal-to-noise ratio of the surface wave signal has been greatly improved, and the noise is relatively small. Using the method provided by the present invention can significantly improve the signal-to-noise ratio and imaging quality of the processing result.

Claims

1. A passive surface wave data screening and imaging method based on noise wave field characteristic analysis, characterized in that: The following steps are involved: Step 1: Preprocess the original passive source noise record, and then cut and segment the processed raw data; Step 2: Count the root mean square value (RMS) of the segmented original records and calculate the coefficient of variation of each data segment; The sampling statistics method is used to obtain the RMS value distribution in the time direction by statistically analyzing the RMS value of the original data. The discrete points are smoothed to obtain the RMS curve, and then the coefficient of variation of the curve is calculated. The coefficient of variation obtained by statistics is used to reflect the discreteness of the RMS curve, so as to judge the change of the corresponding original data wave field characteristics. Step 3: Set a threshold value based on the coefficient of variation of the RMS curve to filter the data segment; By using the difference in coefficient of variation, a reasonable threshold is set to distinguish the data segments that are beneficial to imaging from the interfering data segments, thus improving the imaging quality. Step 4: Interference superposition calculation to obtain effective surface wave signals; The noise records of the two receivers are interfered with to reconstruct the empirical Green's function between the two points and obtain the effective signal.

2. The passive surface wave data screening and imaging method based on noise wave field characteristic analysis according to claim 1 is characterized in that: Step 2 specifically includes the following sub-steps: S2.

1. Let each piece of original data be data m (i,j), m means that the original data is divided into m segments, i, j means that each segment of data is a two-dimensional array of i rows and j columns; the root mean square value RMS of each row of the original record is obtained horizontally by row, and its calculation formula is as follows: Where: N is the number of data points in each row, x i,j is the data in row i and column j, RMS i represents the RMS value at the i-th moment; S2.

2. Calculate the root mean square value of each row, and obtain i root mean square values ​​for each segment of data; then calculate the trend of the statistical RMS value; remove some noise or short-term fluctuations by smoothing, and obtain the change of the RMS value. The specific calculation formula is as follows: Of which: RMS smoothed,k represents the smoothed RMS value, k represents the current sliding window position, and w represents the smoothing window size. Indicates the left and right half widths of the window, rounded down; S2.

3. The RMS of the original noise data segment is calculated and smoothed to obtain the corresponding curve. The characteristics of the smoothed RMS curve correspond to the wave field characteristics of the original data segment. The coefficient of variation of the RMS curve is used to describe the change of the wave field characteristics of the original data. Calculate the standard deviation of the curve to reflect the overall fluctuation of the curve relative to the mean; the specific calculation formula is as follows: Where: std is the standard deviation of the curve, μ is the mean of the curve, and n is the number of samples; S2.

4. The fluctuation of the curve is reflected by the standard deviation. The larger the fluctuation, the more obvious the wave field change of the original data. The smaller the fluctuation, the smaller the wave field change of the original data, and the more stable the surface wave propagation. The coefficient of variation is calculated on the basis of the standard deviation, that is, the ratio of the standard deviation to the mean. The size of the coefficient of variation is used to directly reflect the fluctuation of the curve. The specific calculation formula is as follows: Among them: CV is the coefficient of variation of each segment of data with respect to the RMS curve, std is the standard deviation of the curve, and μ is the mean of the curve; if the CV value is larger, it means that the curve fluctuation is stronger, that is, the original data wave field changes violently; if the CV value is smaller, it means that the curve fluctuation is smaller, that is, the corresponding original data wave field changes less, and the signal propagation is relatively stable.

3. The passive surface wave data screening and imaging method based on noise wave field characteristic analysis according to claim 1 is characterized in that: Step 3 is specifically expressed as follows: Data i is the data segment retained after screening, CV is the coefficient of variation of each data segment with respect to the RMS curve; T is the threshold value.

4. The passive surface wave data screening and imaging method based on noise wave field characteristic analysis according to claim 1 is characterized in that: Step 4 uses the mutual interference method and normalizes the amplitude of each frequency to broaden the effective frequency band. The specific expression formula is as follows: Where: u(A,s,ω) represents the frequency domain signal received by the receiver at A when the source is excited at S; u(B,s,ω) represents the frequency domain signal excited by the source at S and received by the receiver at B; Among them, H AB (ω) represents the extraction of the surface wave signal propagating from point A to point B through mutual interference.

Images