A multi-source data separation method and device based on wideband radon transform

Abstract

The application discloses a multi-source data separation method and device based on wideband Radon transform, relates to the technical field of oil and gas resource exploration, and comprises the following steps: determining pseudo-separated data, performing frequency domain conversion on multi-source aliasing data and the pseudo-separated data to obtain a frequency-wavenumber spectrum, performing Radon transform calculation by using each theoretical bandwidth value in the frequency-wavenumber spectrum to obtain each Radon transform operator, and calculating corresponding each Radon parameter based on the Radon transform operator; constructing a self-adaptive filter by using each theoretical bandwidth value in the frequency-wavenumber spectrum, performing multi-source noise suppression based on each Radon parameter and the self-adaptive filter, and obtaining a coherent signal; performing calculation on the coherent signal by using an aliasing operator to obtain multi-source noise; and performing data separation on the multi-source aliasing data and the pseudo-separated data based on the multi-source noise. The application can avoid the situation that the amplitude is not preserved and noise is left in multi-source data separation, realizes multi-source noise suppression, and better preserves the horizontal propagation characteristics of seismic data.

Description

Technical Field

[0001] This invention relates to the field of oil and gas resource exploration technology, and in particular to a method and apparatus for separating multi-source data based on broadband Radon transform. Background Technology

[0002] Currently, researchers have proposed seismic data acquisition techniques using multiple sources simultaneously. However, multi-source acquisition techniques simultaneously excite multiple sources located in different spatial locations using a certain encoding method. The data acquired by the detectors is aliased data with interference from multiple wavefields. Simultaneous excitation of multiple sources leads to severe information aliasing, strong crosstalk noise between adjacent sources, and traditional seismic data processing workflows cannot directly analyze and process multi-source aliased data. Currently, multi-source data processing methods can be roughly divided into two categories: filtering and denoising, and inversion and separation. When multiple sources located in different spatial locations are simultaneously excited using a certain encoding method, the reflection signals of the main and auxiliary sources will exhibit different characteristics in different domains: in the common shot domain, both the main shot signal and the auxiliary shot signal are continuous and coherent, and multiple wavefields are aliased and intertwined; while in non-common shot domains, such as COG (Common Offset Gather) and CRG (Common Receiver Gather), only the main shot signal is continuous and coherent, while the auxiliary shot signal is discontinuous, incoherent, and discrete noise. Taking advantage of this characteristic, median filtering and its optimization methods are the main methods for separating and filtering multi-source data. Although easy to implement and computationally fast, these filtering methods may cause significant loss of effective signal when the effective signal is weak or the terrain data is complex. Separation methods based on sparse inversion treat the collected multi-source data as observation data and the data separation results as model variables. Therefore, the separation of aliased data can be viewed as an inverse problem solution process, which can be solved using a sparsity-driven iterative inversion method. Commonly used transform domains include Radon transform, Curvelet transform, and Seislet transform. Radon transform, as an important method for sparse characterization of seismic data, is currently the main method for achieving wavefield separation through high-resolution inversion. However, according to the principle of Radon transform, high-resolution Radon transform loses the lateral propagation characteristics of seismic waves and important amplitude variation information. Especially under multi-source noise interference, high-resolution Radon transform cannot effectively suppress strong multi-source noise.

[0003] As can be seen from the above, how to avoid the loss of amplitude and noise residue in the separation of multi-source data, and how to achieve multi-source noise suppression while better preserving the lateral propagation characteristics of seismic data, is a problem that needs to be solved in this field. Summary of the Invention

[0004] In view of this, the purpose of this invention is to provide a multi-source data separation method and apparatus based on broadband Radon transform, which can avoid the problems of amplitude loss and noise residue in multi-source data separation, and achieve multi-source noise suppression while better preserving the lateral propagation characteristics of seismic data. The specific solution is as follows:

[0005] In a first aspect, this application discloses a multi-source data separation method based on broadband Radon transform, including:

[0006] Based on the multi-source aliased data, pseudo-separated data is determined. Frequency domain transformation is performed on the multi-source aliased data and the pseudo-separated data to obtain the frequency-wavenumber spectrum. The theoretical bandwidth values ​​in the frequency-wavenumber spectrum are used to perform Radon transform calculation to obtain each Radon transform operator. Based on each Radon transform operator, the corresponding Radon parameters are calculated.

[0007] Each adaptive filter is constructed using the theoretical bandwidth values ​​in the frequency-wavenumber spectrum, and multi-source noise suppression is performed based on each Radon parameter and each adaptive filter to obtain a coherent signal;

[0008] The coherent signal is calculated using a pre-acquired aliasing operator to obtain multi-source noise;

[0009] Based on the multi-source noise, the multi-source aliased data and the pseudo-separated data are separated respectively.

[0010] Optionally, the step of performing frequency domain transformation on the multi-source aliased data and the pseudo-separated data to obtain a frequency-wavenumber spectrum includes:

[0011] The multi-source aliased data and the pseudo-separated data are transformed from the time-space domain to the frequency-space domain using a Fourier transform model to obtain the frequency-wavenumber spectrum.

[0012] Optionally, the step of constructing each adaptive filter using the theoretical bandwidth values ​​in the frequency-wavenumber spectrum, and performing multi-source noise suppression based on each Radon parameter and each adaptive filter to obtain a coherent signal includes:

[0013] Each adaptive filter is constructed using the theoretical bandwidth values ​​in the frequency-wavenumber spectrum; wherein, the adaptive filter is a Butterworth filter;

[0014] A frequency-varying Radon transform model is constructed based on the Radon parameters. Multi-source noise suppression is performed using the Butterworth filter and the frequency-varying Radon transform model to obtain a coherent signal.

[0015] Optionally, the multi-source data separation method based on broadband Radon transform further includes:

[0016] During the multi-source noise suppression process, it is detected whether the multi-source noise suppression corresponding to each frequency in the frequency-wavenumber spectrum has been completed. If the multi-source noise suppression corresponding to each frequency has not been completed, the current frequency is determined, the next frequency is determined based on the current frequency, and then the process jumps to the step of calculating the Radon transform of each theoretical bandwidth value in the frequency-wavenumber spectrum to perform the multi-source noise suppression process corresponding to the next frequency, until the multi-source noise suppression corresponding to all frequencies is completed.

[0017] Optionally, the step of calculating the coherent signal using a pre-acquired aliasing operator to obtain multi-source noise includes:

[0018] The coherent signal is calculated using the broadband Radon transform method and a pre-acquired aliasing operator to obtain multi-source noise; wherein the broadband Radon transform method is a Radon transform method in which the bandwidth value varies with the frequency.

[0019] Optionally, the step of separating the multi-source aliased data and the pseudo-separated data based on the multi-source noise includes:

[0020] Based on the multi-source noise, the multi-source aliased data and the pseudo-separated data are separated to obtain the current multi-source noise suppression data after separation. The Radon transform calculation process is iterated in a loop to obtain the multi-source noise suppression data after separation in the next loop.

[0021] Optionally, after obtaining the multi-source noise suppression data after the next cycle of separation, the method further includes:

[0022] Determine whether the difference between the multi-source noise suppression data after the next cycle separation and the current multi-source noise suppression data is less than a preset threshold;

[0023] If the difference between the multi-source noise suppression data after the next cycle separation and the current multi-source noise suppression data is less than a preset threshold, then the loop iteration ends and the multi-source noise suppression data after the next cycle separation is output.

[0024] Secondly, this application discloses a multi-source data separation device based on broadband Radon transform, comprising:

[0025] The Radon transform calculation module is used to determine pseudo-separated data based on multi-source aliased data, perform frequency domain transformation on the multi-source aliased data and the pseudo-separated data to obtain a frequency-wavenumber spectrum, perform Radon transform calculation using the theoretical bandwidth values ​​in the frequency-wavenumber spectrum to obtain each Radon transform operator, and calculate the corresponding Radon parameters based on each Radon transform operator.

[0026] The multi-source noise suppression module is used to construct each adaptive filter using each theoretical bandwidth value in the frequency-wavenumber spectrum, and to perform multi-source noise suppression based on each Radon parameter and each adaptive filter to obtain a coherent signal;

[0027] A multi-source noise determination module is used to calculate the coherent signal using a pre-acquired aliasing operator to obtain multi-source noise;

[0028] The data separation module is used to separate the multi-source aliased data and the pseudo-separated data based on the multi-source noise.

[0029] As can be seen, this application provides a multi-source data separation method based on broadband Radon transform, including determining pseudo-separated data based on multi-source aliased data, performing frequency domain transformation on the multi-source aliased data and the pseudo-separated data to obtain a frequency-wavenumber spectrum, performing Radon transform calculation using the theoretical bandwidth values ​​in the frequency-wavenumber spectrum to obtain Radon transform operators, calculating corresponding Radon parameters based on the Radon transform operators, constructing adaptive filters using the theoretical bandwidth values ​​in the frequency-wavenumber spectrum, and performing multi-source noise suppression based on the Radon parameters and the adaptive filters to obtain a coherent signal; calculating the coherent signal using pre-acquired aliasing operators to obtain multi-source noise; and performing data separation on the multi-source aliased data and the pseudo-separated data based on the multi-source noise. This application proposes a broadband curvature parameter Radon transform method, which is a method where the curvature parameter varies with frequency. Data separation is achieved through this broadband curvature parameter Radon transform, and frequency-wavenumber spectra are obtained by frequency-domain transformation of multi-source aliased data. This method effectively preserves the theoretical basis of amplitude and performs Radon transform inversion in the frequency domain, eliminating the need for high-resolution inversion and improving computational efficiency. It addresses the issues of amplitude preservation and residual multi-source noise in existing Radon transform methods for multi-source data separation. By employing an adaptive filter to suppress multi-source noise, the lateral propagation characteristics of the effective waves can be better preserved, avoiding the loss of seismic wave lateral propagation characteristics and important amplitude variation information, as well as the reduced imaging quality, that occurs with high-resolution Radon transform. Therefore, it can achieve multi-source noise suppression while effectively preserving the lateral propagation characteristics of seismic data. Attached Figure Description

[0030] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0031] Figure 1 This is a flowchart of a multi-source data separation method based on broadband Radon transform disclosed in this application;

[0032] Figure 2 This is a flowchart of a multi-source data separation method based on broadband Radon transform disclosed in this application;

[0033] Figure 3 This is a flowchart of a multi-source data separation method based on broadband Radon transform disclosed in this application;

[0034] Figure 4 This is an example diagram of a broadband curvature parameter multi-source noise filtering method disclosed in this application;

[0035] Figure 5 This is an example diagram of a three-iteration estimation signal disclosed in this application;

[0036] Figure 6 This is an example diagram of a three-iteration noise estimation method disclosed in this application;

[0037] Figure 7 This is an example diagram of a three-stage noise suppression result disclosed in this application;

[0038] Figure 8 This is a schematic diagram of a multi-source data separation device based on broadband Radon transform disclosed in this application. Detailed Implementation

[0039] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0040] Currently, researchers have proposed seismic data acquisition techniques using multiple sources simultaneously. However, multi-source acquisition techniques simultaneously excite multiple sources located in different spatial locations using a certain encoding method. The data acquired by the detectors is aliased data with interference from multiple wavefields. Simultaneous excitation of multiple sources leads to severe information aliasing, strong crosstalk noise between adjacent sources, and traditional seismic data processing workflows cannot directly analyze and process multi-source aliased data. Currently, multi-source data processing methods can be roughly divided into two categories: filtering and denoising, and inversion and separation. When multiple sources located in different spatial locations are simultaneously excited using a certain encoding method, the reflection signals of the main and auxiliary sources will exhibit different characteristics in different domains: in the common shot domain, both the main shot signal and the auxiliary shot signal are continuous and coherent, and multiple wavefields are aliased and interleaved; while in non-common shot domains, such as COG and CRG, only the main shot signal is continuous and coherent, while the auxiliary shot signal is discontinuous, incoherent, and discrete noise. Taking advantage of this characteristic, median filtering and its optimization methods are the main methods for separating and filtering multi-source data. Although easy to implement and computationally fast, these filtering methods may cause significant loss of effective signal when the effective signal is weak or the terrain data is complex. Separation methods based on sparse inversion treat the acquired multi-source data as observation data and the data separation results as model variables. Therefore, the separation of aliased data can be viewed as an inverse problem solution process, which can be solved using a sparsity-promoted iterative inversion method. Commonly used transform domains include the Radon domain, Curvelet domain, and Seislet domain. The Radon transform, as an important method for sparse characterization of seismic data, is currently the main method for achieving wavefield separation through high-resolution inversion. However, according to the principle of the Radon transform, high-resolution Radon transform loses the lateral propagation characteristics of seismic waves and loses important amplitude variation information. Especially under multi-source noise interference, high-resolution Radon transform cannot effectively suppress strong multi-source noise. Therefore, how to avoid amplitude loss and noise residue in multi-source data separation, and how to achieve multi-source noise suppression while better preserving the lateral propagation characteristics of seismic data, are problems that need to be solved in this field.

[0041] See Figure 1 As shown, this embodiment of the invention discloses a multi-source data separation method based on broadband Radon transform, which specifically includes:

[0042] Step S11: Based on the multi-source aliased data, pseudo-separated data is determined. Frequency domain transformation is performed on the multi-source aliased data and the pseudo-separated data to obtain a frequency-wavenumber spectrum. Radon transform is performed using the theoretical bandwidth values ​​in the frequency-wavenumber spectrum to obtain Radon transform operators. Based on each Radon transform operator, the corresponding Radon parameters are calculated.

[0043] In this embodiment, pseudo-separated data is determined based on multi-source aliased data. Frequency domain transformation is performed on the multi-source aliased data and pseudo-separated data to obtain the FK (Frequency-wave number chart) spectrum. Based on the FK spectrum of the seismic data, the theoretical bandwidth value of the Radon transform curvature parameter q as a function of frequency is obtained. Radon transform calculation is performed using each theoretical bandwidth value in the frequency-wave number spectrum to obtain each Radon transform operator for the multi-source data at each frequency. Based on each Radon transform operator, the corresponding Radon parameters are calculated.

[0044] Step S12: Construct each adaptive filter using the theoretical bandwidth values ​​in the frequency-wavenumber spectrum, and perform multi-source noise suppression based on each Radon parameter and each adaptive filter to obtain a coherent signal.

[0045] Step S13: Calculate the coherent signal using the pre-acquired aliasing operator to obtain multi-source noise.

[0046] In this embodiment, a wideband Radon transform method is used, and a pre-acquired aliasing operator is used to calculate the coherent signal to obtain multi-source noise. The wideband Radon transform method is a Radon transform method where the bandwidth value varies with frequency. During the multi-source noise suppression process, it is detected whether the multi-source noise suppression corresponding to each frequency in the frequency-wavenumber spectrum is completed. If the multi-source noise suppression corresponding to each frequency is not completed, the current frequency is determined, and the next frequency is determined based on the current frequency. Then, the process jumps to the step of calculating the Radon transform of each theoretical bandwidth value in the frequency-wavenumber spectrum to execute the multi-source noise suppression process corresponding to the next frequency, until the multi-source noise suppression corresponding to all frequencies is completed, finally obtaining the multi-source noise.

[0047] Step S14: Based on the multi-source noise, separate the multi-source aliased data and the pseudo-separated data respectively.

[0048] In this embodiment, the multi-source aliased data and the pseudo-separated data are separated based on the multi-source noise to obtain the current multi-source noise suppression data. The Radon transform calculation process is iterated to obtain the multi-source noise suppression data after the next iteration. It is determined whether the difference between the multi-source noise suppression data after the next iteration and the current multi-source noise suppression data is less than a preset threshold. If the difference between the multi-source noise suppression data after the next iteration and the current multi-source noise suppression data is less than the preset threshold, the iteration ends and the multi-source noise suppression data after the next iteration is output.

[0049] In this embodiment, an adaptive filter with bandwidth varying with frequency is constructed using the theoretical bandwidth values ​​in the frequency-wavenumber spectrum. Multi-source noise suppression is performed based on the Radon parameters and the adaptive filters to obtain a coherent signal. It is then determined whether all adaptive filters have been constructed. If all adaptive filters have been constructed, the coherent signal is calculated using a pre-acquired aliasing operator to obtain multi-source noise. Multi-source noise is subtracted from the multi-source aliasing data to achieve data separation. Finally, it is determined whether the suppression effect has been achieved, i.e., whether the difference between the multi-source noise suppression data after the next cycle and the current multi-source noise suppression data is less than a preset threshold. If it is greater than the preset threshold, it indicates that the suppression effect has not been achieved. The above steps are repeated until the effect requirement or the number of cycles is met, and then the loop iteration ends, outputting multi-source noise suppression data that meets the effect requirement.

[0050] The specific process of multi-source data separation based on broadband Radon transform in this application is as follows: Figure 2 As shown, (1) pseudo-separated data is determined based on multi-source aliasing data, and frequency domain transformation is performed on the multi-source aliasing data and the pseudo-separated data to obtain a frequency-wavenumber spectrum; (2) the theoretical bandwidth values ​​of the Radon transform curvature parameter q as a function of frequency are obtained according to the frequency-wavenumber spectrum, and Radon transform calculation is performed using the theoretical bandwidth values ​​in the frequency-wavenumber spectrum to obtain Radon transform operators, and corresponding Radon parameters are calculated based on the Radon transform operators; (3) an adaptive filter with bandwidth as a function of frequency is constructed using the different curvature parameters corresponding to different frequencies, and multi-source noise suppression is performed based on the Radon parameters and the adaptive filter; (4) during the multi-source noise suppression process, it is detected whether the multi-source noise suppression corresponding to each frequency in the frequency-wavenumber spectrum has been completed. (5) If the multi-source noise suppression corresponding to each frequency is not completed, the broadband Radon transform method is used and the pre-acquired aliasing operator is used to calculate the coherent signal to obtain the multi-source noise; (6) Based on the multi-source noise, the multi-source aliasing data and the pseudo-separation data are separated to obtain the current multi-source noise suppression data after separation. The Radon transform calculation process is iterated in a loop to obtain the multi-source noise suppression data after separation in the next loop; (7) It is determined whether the difference between the multi-source noise suppression data after separation in the next loop and the current multi-source noise suppression data is less than a preset threshold. If the difference between the multi-source noise suppression data after separation in the next loop and the current multi-source noise suppression data is less than the preset threshold, the loop iteration ends and the multi-source noise suppression data after separation in the next loop is output.

[0051] In this embodiment, pseudo-separated data is determined based on multi-source aliased data. Frequency domain transformation is performed on the multi-source aliased data and the pseudo-separated data to obtain a frequency-wavenumber spectrum. Radon transform calculation is performed using the theoretical bandwidth values ​​in the frequency-wavenumber spectrum to obtain Radon transform operators. Corresponding Radon parameters are calculated based on each Radon transform operator. Adaptive filters are constructed using the theoretical bandwidth values ​​in the frequency-wavenumber spectrum, and multi-source noise suppression is performed based on the Radon parameters and the adaptive filters to obtain a coherent signal. The coherent signal is calculated using pre-acquired aliasing operators to obtain multi-source noise. Data separation is performed on the multi-source aliased data and the pseudo-separated data based on the multi-source noise. This application proposes a broadband curvature parameter Radon transform method, which is a method where the curvature parameter varies with frequency. Data separation is achieved through this broadband curvature parameter Radon transform, and frequency-wavenumber spectra are obtained by frequency-domain transformation of multi-source aliased data. This method effectively preserves the theoretical basis of amplitude and performs Radon transform inversion in the frequency domain, eliminating the need for high-resolution inversion and improving computational efficiency. It addresses the issues of amplitude preservation and residual multi-source noise in existing Radon transform methods for multi-source data separation. By employing an adaptive filter to suppress multi-source noise, the lateral propagation characteristics of the effective waves can be better preserved, avoiding the loss of seismic wave lateral propagation characteristics and important amplitude variation information, as well as the reduced imaging quality, that occurs with high-resolution Radon transform. Therefore, it can achieve multi-source noise suppression while effectively preserving the lateral propagation characteristics of seismic data.

[0052] See Figure 3 As shown, this embodiment of the invention discloses a multi-source data separation method based on broadband Radon transform, which specifically includes:

[0053] Step S21: Based on the multi-source aliased data, pseudo-separated data is determined. The multi-source aliased data and pseudo-separated data are transformed from the time-space domain to the frequency-space domain using a Fourier transform model to obtain the frequency-wavenumber spectrum. The theoretical bandwidth values ​​in the frequency-wavenumber spectrum are used to perform Radon transform calculations to obtain Radon transform operators. The corresponding Radon parameters are calculated based on the Radon transform operators.

[0054] In this embodiment, the multi-source superimposed data is represented as: d bl =d1+Γd2; where d bl This is multi-source aliased data, where d1 and d2 are single-shot data to be recovered, and Γ is the aliasing operator determined by the known observation system. Based on the multi-source aliased data, pseudo-separation data is determined, represented as: Γ H d bl =Γ H d1+d2.

[0055] In this embodiment, the Fourier transform model is used to transform the multi-source aliased data and pseudo-separated data in the time-space domain to the frequency-space domain to obtain the FK spectrum. Based on the FK spectrum, the theoretical bandwidth value of the Radon transform curvature parameter q as a function of frequency is determined. Then, based on the theoretical bandwidth value, the Radon transform operator is constructed, and the Radon parameter distribution of the multi-source data at each frequency is calculated to obtain the frequency-varying Radon transform model.

[0056] Seismic data in the spatiotemporal domain under a certain curvature can be represented as:

[0057] d(t,x)=a(t=τ+qx 2 ,x);

[0058] Where t, x, and τ represent time, spatial domain, and intercept variables, respectively, and a() represents the seismic wavelet, then its FK spectrum is:

[0059]

[0060] Where A(ω,x) is the spectrum of a(t,x), and as shown in the above equation, the spectrum of the parabolic Radon transform is a linearly modulated spectrum, whose spectral width is related to three parameters: ω, q, and x. Since parameter x is related to the observation system, we will focus on the influence of parameter x on the spectrum, and then deduce the selection principle of the curvature parameter q. Let q∈(q min ,q max ), and q min <0, q max When k > 0, x ∈(ωq min q max ,ωq max x max );

[0061] This shows that, under a given observation system, the wavenumber bandwidth at a certain frequency is determined by the range of the theoretical bandwidth value q. When the wavenumber k of the seismic data... x When the bandwidth is wide, such as when the lateral amplitude changes significantly, only by expanding the range of the curvature parameter q can the lateral propagation characteristics of seismic data be well fitted.

[0062] According to the sampling theorem, combined with formula k x ∈(ωq min q max ,ωq max x max The spatial sampling step size must meet the following requirements:

[0063]

[0064] Thus, the critical value of the curvature parameter range is obtained:

[0065]

[0066] Let the Radon domain seismic data be represented as:

[0067] m(τ,q)=b(τ=t-qx 2 ,q);

[0068] b() represents the Radon domain wavelet, then its FK spectrum is:

[0069]

[0070] Where B(ω, q) is the spectrum of m(τ, q), from which the spectral width k of the curvature parameter q can be obtained. q satisfy:

[0071] k q ∈(ω(x 2 ) min ,ω(x 2 ) max );

[0072] According to the sampling theorem, the critical value of the curvature parameter sampling step size Δq is:

[0073]

[0074] As can be seen from the above formula, the step size of the curvature parameter is also related to the frequency.

[0075] In summary, the range and step size of the curvature parameter are inversely proportional to the frequency, making it a frequency-dependent parameter. Moreover, its range is often greater than the range of curvature parameters determined by the highest conventional frequency. Therefore, the method proposed in this case, which allows the curvature parameter to vary with frequency, is called the wideband curvature parameter Radon transform.

[0076] Step S22: Construct each adaptive filter using the theoretical bandwidth values ​​in the frequency-wavenumber spectrum, construct a frequency-varying Radon transform model based on each Radon parameter, and use the Butterworth filter and the frequency-varying Radon transform model to suppress multi-source noise and obtain a coherent signal; wherein, the adaptive filter is a Butterworth filter.

[0077] In this embodiment, an adaptive filter with bandwidth varying with frequency is constructed by utilizing the different curvature parameters corresponding to different frequencies. This adaptive filter includes, but is not limited to, a Butterworth filter, and operates on the Radon parameter in the τ-q domain to obtain a coherent signal. The Butterworth filter is designed to suppress multi-source noise.

[0078]

[0079] Here, μ represents the mean position of the signal in the q-parameter. σ(ω) reflects the bandwidth of the filter and is a frequency-dependent parameter. This parameter is designed to adapt to frequency and vary with the critical width of the curvature parameter:

[0080]

[0081] This filter can adaptively adjust the filter coefficients according to the degree of aliasing. The adaptive filter is more sensitive to the waveform changes of the processed data. Applying this filter to wideband Radon domain data can achieve adaptive filtering based on signal distribution.

[0082] Step S23: Calculate the coherent signal using the pre-acquired aliasing operator to obtain multi-source noise.

[0083] Step S24: Based on the multi-source noise, perform data separation on the multi-source aliased data and the pseudo-separation data respectively.

[0084] For example, the effect of broadband parameter filtering on a certain frequency of simulated seismic data is shown in the figure below. Figure 4 As shown, (a) and (b) are broadband Radon parameter distribution diagrams containing multi-source noise at 10Hz and 20Hz, respectively. This demonstrates that the curvature parameter range of the data varies at different frequencies, and the theoretical bandwidth value changes with frequency. Broadband Radon transform operators are constructed based on the theoretical bandwidth values ​​at 10Hz and 20Hz, and the Radon parameter distribution of the multi-source data is calculated. Butterworth filters with bandwidth varying with the corresponding frequency are designed based on the curvature parameters at different frequencies, as shown in (c) and (d). Effective signals can pass within the selected parameter range, while noise can be effectively suppressed. The designed filters can effectively filter out multi-source noise. Figure 5 It is based on Figure 4 The following is an example of the estimated signal generated by the three iterations of the multi-source noise suppression for all frequencies. (1), (2), and (3) are the noisy signals estimated by the first, second, and third iterations, respectively. It can be clearly seen that after three iterations, the residual multi-source noise in the signal is significantly reduced in turn. This iterative process is calculated based on the broadband Radon transform algorithm proposed above. Figure 6 Examples of iterative noise estimation for different iterations are shown in the figures. (1), (2), and (3) represent the noise estimated in the first, second, and third iterations, respectively. The noise estimates differ depending on the number of iterations. The multi-source noise is estimated by applying the Radon transform of the broadband curvature parameter to the result of each suppression of multi-source noise. As can be seen from the figures, the multi-source noise can be estimated very accurately, which facilitates subsequent iterative processing.

[0085] Figure 7The final result of the three-stage noise suppression is shown in (1), (2), and (3), which represent the signals separated in the first, second, and third iterations, respectively. The separation result is obtained by subtracting the estimated noise from the noisy signal estimated after wideband Radon transform and filtering. The noise distribution at different frequencies will vary, so the parameters in the iterative algorithm are determined based on the distribution of the signal and noise. It can be clearly seen from the figure that after three iterations, the residual multi-source noise in the signal is significantly reduced in sequence, resulting in successful suppression of multi-source noise.

[0086] The innovations of this application are: (1) a new method for selecting the q-parameter of Radon transform, namely the wideband curvature parameter Radon transform method, which is derived from the FK spectrum and has a good amplitude-preserving theoretical basis; (2) the least squares method is used in the frequency domain to realize Radon transform inversion, which does not require high-resolution inversion and has high computational efficiency; (3) the smoothing Butterworth filter is used to suppress multi-source noise, which can better preserve the transverse propagation characteristics of the effective wave.

[0087] In this embodiment, pseudo-separated data is determined based on multi-source aliased data. Frequency domain transformation is performed on the multi-source aliased data and the pseudo-separated data to obtain a frequency-wavenumber spectrum. Radon transform calculation is performed using the theoretical bandwidth values ​​in the frequency-wavenumber spectrum to obtain Radon transform operators. Corresponding Radon parameters are calculated based on each Radon transform operator. Adaptive filters are constructed using the theoretical bandwidth values ​​in the frequency-wavenumber spectrum, and multi-source noise suppression is performed based on the Radon parameters and the adaptive filters to obtain a coherent signal. The coherent signal is calculated using pre-acquired aliasing operators to obtain multi-source noise. Data separation is performed on the multi-source aliased data and the pseudo-separated data based on the multi-source noise. This application proposes a broadband curvature parameter Radon transform method, which is a method where the curvature parameter varies with frequency. Data separation is achieved through this broadband curvature parameter Radon transform, and frequency-wavenumber spectra are obtained by frequency-domain transformation of multi-source aliased data. This method effectively preserves the theoretical basis of amplitude and performs Radon transform inversion in the frequency domain, eliminating the need for high-resolution inversion and improving computational efficiency. It addresses the issues of amplitude preservation and residual multi-source noise in existing Radon transform methods for multi-source data separation. By employing an adaptive filter to suppress multi-source noise, the lateral propagation characteristics of the effective waves can be better preserved, avoiding the loss of seismic wave lateral propagation characteristics and important amplitude variation information, as well as the reduced imaging quality, that occurs with high-resolution Radon transform. Therefore, it can achieve multi-source noise suppression while effectively preserving the lateral propagation characteristics of seismic data.

[0088] See Figure 8As shown in the figure, this invention discloses a multi-source data separation device based on broadband Radon transform, which may specifically include:

[0089] The Radon transform calculation module 11 is used to determine pseudo-separated data based on multi-source aliasing data, perform frequency domain transformation on the multi-source aliasing data and the pseudo-separated data to obtain a frequency-wavenumber spectrum, perform Radon transform calculation using the theoretical bandwidth values ​​in the frequency-wavenumber spectrum to obtain each Radon transform operator, and calculate the corresponding Radon parameters based on each Radon transform operator.

[0090] The multi-source noise suppression module 12 is used to construct each adaptive filter using each theoretical bandwidth value in the frequency-wavenumber spectrum, and to perform multi-source noise suppression based on each Radon parameter and each adaptive filter to obtain a coherent signal.

[0091] The multi-source noise determination module 13 is used to calculate the coherent signal using a pre-acquired aliasing operator to obtain multi-source noise;

[0092] The data separation module 14 is used to separate the multi-source aliased data and the pseudo-separated data based on the multi-source noise.

[0093] In this embodiment, pseudo-separated data is determined based on multi-source aliased data. Frequency domain transformation is performed on the multi-source aliased data and the pseudo-separated data to obtain a frequency-wavenumber spectrum. Radon transform calculation is performed using the theoretical bandwidth values ​​in the frequency-wavenumber spectrum to obtain Radon transform operators. Corresponding Radon parameters are calculated based on each Radon transform operator. Adaptive filters are constructed using the theoretical bandwidth values ​​in the frequency-wavenumber spectrum, and multi-source noise suppression is performed based on the Radon parameters and the adaptive filters to obtain a coherent signal. The coherent signal is calculated using pre-acquired aliasing operators to obtain multi-source noise. Data separation is performed on the multi-source aliased data and the pseudo-separated data based on the multi-source noise. This application proposes a broadband curvature parameter Radon transform method, which is a method where the curvature parameter varies with frequency. Data separation is achieved through this broadband curvature parameter Radon transform, and frequency-wavenumber spectra are obtained by frequency-domain transformation of multi-source aliased data. This method effectively preserves the theoretical basis of amplitude and performs Radon transform inversion in the frequency domain, eliminating the need for high-resolution inversion and improving computational efficiency. It addresses the issues of amplitude preservation and residual multi-source noise in existing Radon transform methods for multi-source data separation. By employing an adaptive filter to suppress multi-source noise, the lateral propagation characteristics of the effective waves can be better preserved, avoiding the loss of seismic wave lateral propagation characteristics and important amplitude variation information, as well as the reduced imaging quality, that occurs with high-resolution Radon transform. Therefore, it can achieve multi-source noise suppression while effectively preserving the lateral propagation characteristics of seismic data.

[0094] In some specific embodiments, the Radon transform calculation module 11 may specifically include:

[0095] The conversion module is used to convert the multi-source aliased data and the pseudo-separated data from the time-space domain to the frequency-space domain using a Fourier transform model to obtain the frequency-wavenumber spectrum.

[0096] In some specific embodiments, the multi-source noise suppression module 12 may specifically include:

[0097] An adaptive filter construction module is used to construct each adaptive filter using the theoretical bandwidth values ​​in the frequency-wavenumber spectrum; wherein, the adaptive filter is a Butterworth filter;

[0098] The multi-source noise suppression module is used to construct a frequency-varying Radon transform model based on the Radon parameters, and to suppress multi-source noise using the Butterworth filter and the frequency-varying Radon transform model to obtain a coherent signal.

[0099] In some specific embodiments, it may further include:

[0100] The frequency determination module is used to detect whether the multi-source noise suppression corresponding to each frequency in the frequency-wavenumber spectrum has been completed during the multi-source noise suppression process. If the multi-source noise suppression corresponding to each frequency has not been completed, the current frequency is determined, the next frequency is determined based on the current frequency, and then the process jumps to the step of calculating the Radon transform of each theoretical bandwidth value in the frequency-wavenumber spectrum to execute the multi-source noise suppression process corresponding to the next frequency, until the multi-source noise suppression corresponding to all frequencies is completed.

[0101] In some specific embodiments, the multi-source noise determination module 13 may specifically include:

[0102] The multi-source noise determination module is used to calculate the coherent signal using the broadband Radon transform method and a pre-acquired aliasing operator to obtain the multi-source noise; wherein the broadband Radon transform method is a Radon transform method in which the bandwidth value varies with the frequency.

[0103] In some specific embodiments, the data separation module 14 may specifically include:

[0104] The iterative loop module is used to separate the multi-source aliased data and the pseudo-separated data based on the multi-source noise to obtain the current multi-source noise suppression data after separation, and iteratively iterates the Radon transform calculation process to obtain the multi-source noise suppression data after separation in the next cycle.

[0105] In some specific embodiments, the data separation module 14 may specifically include:

[0106] The judgment module is used to determine whether the difference between the multi-source noise suppression data after the next cycle separation and the current multi-source noise suppression data is less than a preset threshold.

[0107] The loop iteration termination module is used to terminate the loop iteration and output the multi-source noise suppression data after the next loop separation if the difference between the multi-source noise suppression data after the next loop separation and the current multi-source noise suppression data is less than a preset threshold.

[0108] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only 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 one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0109] The above provides a detailed description of a multi-source data separation method and apparatus based on broadband Radon transform provided by the present invention. Specific examples have been used to illustrate the principle and implementation of the present invention. The description of the above embodiments is only for the purpose of helping to understand the method and core idea of ​​the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation and application scope based on the idea of ​​the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A multi-source data separation method based on broadband Radon transform, characterized in that, include: Based on multi-source aliased data, pseudo-separated data is identified. Frequency domain transformation is performed on the multi-source aliased data and the pseudo-separated data to obtain a frequency-wavenumber spectrum. Radon transform is calculated using the theoretical bandwidth values ​​in the frequency-wavenumber spectrum to obtain Radon transform operators. Based on the Radon transform operators, the corresponding Radon parameters are calculated. The theoretical bandwidth value is the bandwidth value of the Radon transform curvature parameter as a function of frequency. Each adaptive filter is constructed using the theoretical bandwidth values ​​in the frequency-wavenumber spectrum, and multi-source noise suppression is performed based on each Radon parameter and each adaptive filter to obtain a coherent signal; The coherent signal is calculated using a pre-acquired aliasing operator to obtain multi-source noise; Based on the multi-source noise, the multi-source aliased data and the pseudo-separated data are separated respectively.

2. The multi-source data separation method based on broadband Radon transform according to claim 1, characterized in that, The step of performing frequency domain transformation on the multi-source aliased data and the pseudo-separated data to obtain the frequency-wavenumber spectrum includes: The multi-source aliased data and the pseudo-separated data are transformed from the time-space domain to the frequency-space domain using a Fourier transform model to obtain the frequency-wavenumber spectrum.

3. The multi-source data separation method based on broadband Radon transform according to claim 1, characterized in that, The process of constructing adaptive filters using theoretical bandwidth values ​​from the frequency-wavenumber spectrum and performing multi-source noise suppression based on the Radon parameters and the adaptive filters to obtain coherent signals includes: Each adaptive filter is constructed using the theoretical bandwidth values ​​in the frequency-wavenumber spectrum; wherein, the adaptive filter is a Butterworth filter; A frequency-varying Radon transform model is constructed based on the Radon parameters. Multi-source noise suppression is performed using the Butterworth filter and the frequency-varying Radon transform model to obtain a coherent signal.

4. The multi-source data separation method based on broadband Radon transform according to claim 1, characterized in that, Also includes: During the multi-source noise suppression process, it is detected whether the multi-source noise suppression corresponding to each frequency in the frequency-wavenumber spectrum is completed. If the multi-source noise suppression corresponding to each frequency is not completed, the current frequency is determined, the next frequency is determined based on the current frequency, and then the process jumps to the step of performing Radon transform calculation using the theoretical bandwidth values ​​in the frequency-wavenumber spectrum to perform the multi-source noise suppression process corresponding to the next frequency, until the multi-source noise suppression corresponding to all frequencies is completed.

5. The multi-source data separation method based on broadband Radon transform according to any one of claims 1 to 4, characterized in that, The step of calculating the coherent signal using a pre-acquired aliasing operator to obtain multi-source noise includes: The coherent signal is calculated using the broadband Radon transform method and a pre-acquired aliasing operator to obtain multi-source noise; wherein the broadband Radon transform method is a Radon transform method in which the bandwidth value varies with the frequency.

6. The multi-source data separation method based on broadband Radon transform according to claim 1, characterized in that, The step of separating the multi-source aliased data and the pseudo-separated data based on the multi-source noise includes: Based on the multi-source noise, the multi-source aliased data and the pseudo-separated data are separated to obtain the current multi-source noise suppression data after separation. The Radon transform calculation process is iterated in a loop to obtain the multi-source noise suppression data after separation in the next loop.

7. The multi-source data separation method based on broadband Radon transform according to claim 6, characterized in that, After obtaining the multi-source noise suppression data after the next cycle of separation, the method further includes: Determine whether the difference between the multi-source noise suppression data after the next cycle separation and the current multi-source noise suppression data is less than a preset threshold; If the difference between the multi-source noise suppression data after the next cycle separation and the current multi-source noise suppression data is less than a preset threshold, then the loop iteration ends and the multi-source noise suppression data after the next cycle separation is output.

8. A multi-source data separation device based on broadband Radon transform, characterized in that, include: The Radon transform calculation module is used to determine pseudo-separated data based on multi-source aliased data, perform frequency domain transformation on the multi-source aliased data and the pseudo-separated data to obtain a frequency-wavenumber spectrum, perform Radon transform calculation using the theoretical bandwidth values ​​in the frequency-wavenumber spectrum to obtain various Radon transform operators, and calculate the corresponding Radon parameters based on the Radon transform operators; wherein, the theoretical bandwidth value is the bandwidth value of the Radon transform curvature parameter as a function of frequency; The multi-source noise suppression module is used to construct each adaptive filter using each theoretical bandwidth value in the frequency-wavenumber spectrum, and to perform multi-source noise suppression based on each Radon parameter and each adaptive filter to obtain a coherent signal; A multi-source noise determination module is used to calculate the coherent signal using a pre-acquired aliasing operator to obtain multi-source noise; The data separation module is used to separate the multi-source aliased data and the pseudo-separated data based on the multi-source noise.

Images