Method and device for acquiring wellbore acoustic wave velocity based on frequency domain analysis

Abstract

This application discloses a method and device for acquiring borehole acoustic velocity based on frequency domain analysis, relating to the field of geophysical acoustic logging. The method comprises: acquiring an array waveform signal obtained by full-wavelength acoustic logging of a target layer; performing Fourier transform on the array waveform signal to obtain a waveform spectrum signal; extracting a noisy dispersion signal from the waveform spectrum signal, performing noise reduction processing on the noisy dispersion signal to obtain an effective dispersion signal; and performing signal analysis processing on the effective dispersion signal to obtain the borehole acoustic velocity. This application enables real-time acquisition of formation acoustic velocity while simultaneously collecting acoustic logging data, improving the accuracy and stability of acoustic velocity acquisition. It addresses the problems of traditional frequency domain processing methods, such as the need for manual interaction and slow processing efficiency, as well as the inaccurate and unstable acoustic velocity processing results. It also addresses the shortcomings of real-time time domain acoustic velocity processing methods.

Description

Technical Field

[0001] The present application relates to the field of geophysical acoustic logging, and in particular to a method and device for acquiring borehole acoustic wave velocity based on frequency domain analysis. Background Art

[0002] Acquiring elastic wave velocity profiles along the wellbore is the core task of acoustic logging, providing the necessary data foundation for in-depth formation analysis. Automated acoustic velocity processing is a pressing need for field logging operations and a key technology for automated formation analysis and real-time evaluation.

[0003] Time-domain processing methods are the mainstream technical means for obtaining formation sound velocity. The threshold method is a widely used real-time sound velocity processing method. This method uses the waveform amplitude that first reaches the threshold as the arrival time of the formation wave signal and combines it with the instrument's receiver array spacing to calculate the sound wave velocity. However, the threshold method cannot distinguish between noise and true formation wave signals, and its processing results are easily affected by noise. The slowness-time coherence method (STC) has good anti-interference ability and high processing efficiency and is the most widely used sound velocity processing method. In actual data processing, the STC method usually requires fine-tuning of multiple parameters such as time window, filtering, and correlation search based on expert experience to obtain formation sound velocity. Researchers have attempted to utilize methods such as peak searching, Kalman filtering, and artificial intelligence (Sun Zhifeng et al., "A Method and Apparatus for Real-Time Calculation of Formation P- and S-Wave Slownesses"; Sun et al., "A Stabilized Real-Time Slowness Estimation Method for Compressional Waves by Using Kalman Filtering"; and Fan Chuan et al., "A Method for Calculating Acoustic Time Difference") to automatically determine STC processing parameters and achieve automatic extraction of formation acoustic velocity. However, in complex geological environments, these methods struggle to remove noise interference and accurately identify correlation changes in array waveforms, resulting in large errors and unstable fluctuations in the acoustic velocity processing results. Therefore, a solution that can accurately extract borehole acoustic velocity in real time is urgently needed. Summary of the Invention

[0004] In view of the above problems, the present application is proposed to provide a method, apparatus, computing device, computer storage medium and computer program product for obtaining borehole acoustic wave velocity based on frequency domain analysis that overcomes the above problems or at least partially solves the above problems.

[0005] According to one aspect of an embodiment of the present application, a method for acquiring borehole acoustic wave velocity based on frequency domain analysis is provided, the method comprising:

[0006] Acquire array waveform signals obtained by full-wavelength acoustic logging of the target layer;

[0007] Performing Fourier transform on the array waveform signal to obtain a waveform spectrum signal;

[0008] Extracting a noisy dispersion signal from a waveform spectrum signal, performing noise reduction processing on the noisy dispersion signal to obtain an effective dispersion signal;

[0009] The effective dispersion signal is analyzed and processed to obtain the borehole acoustic wave velocity.

[0010] Furthermore, performing signal analysis and processing on the effective dispersion signal to obtain the borehole acoustic wave velocity further includes:

[0011] If the effective dispersion signal is a Stoneley wave dispersion signal and / or a longitudinal wave dispersion signal, frequency distribution statistics are performed on the Stoneley wave dispersion signal and / or the longitudinal wave dispersion signal, and a velocity corresponding to a maximum frequency is determined as the Stoneley wave sound velocity and / or the longitudinal wave sound velocity;

[0012] If the effective dispersion signal is a bending wave dispersion signal and / or a spiral wave dispersion signal, a fitting process is performed on the bending wave dispersion signal and / or the spiral wave dispersion signal to obtain the shear wave speed.

[0013] Furthermore, fitting the bending wave dispersion signal and / or the spiral wave dispersion signal to obtain the shear wave velocity further includes:

[0014] Input the wellbore acoustic parameters into a pre-trained fast dispersion simulation network to obtain the corresponding theoretical simulation dispersion signal, where the wellbore acoustic parameters include: wellbore mud velocity, wellbore diameter, and formation shear wave velocity;

[0015] The shear wave velocity of the formation at which the difference between the bending wave dispersion signal and the theoretical simulation dispersion signal is minimum is determined as the shear wave velocity, and / or the shear wave velocity of the formation at which the difference between the spiral wave dispersion signal and the theoretical simulation dispersion signal is minimum is determined as the shear wave velocity.

[0016] Furthermore, the training process of the fast dispersion simulation network includes:

[0017] generating a preset number of theoretical simulation dispersion signals based on the dispersion equation of the wellbore acoustic field;

[0018] The neural network is trained using the borehole acoustic sample parameters, including wellbore mud velocity, wellbore diameter, and formation shear wave velocity, as training dataset features and theoretically simulated dispersion signals as training dataset labels to obtain a fast dispersion simulation network.

[0019] Furthermore, performing noise reduction processing on the noisy dispersion signal to obtain an effective dispersion signal further includes:

[0020] The pre-trained dispersion denoising model is used to denoise the noisy dispersion signal to obtain an effective dispersion signal.

[0021] The training process of the dispersion denoising model includes:

[0022] Obtaining a noisy sample dispersion signal and a valid sample dispersion signal;

[0023] Use the noisy sample dispersion signal to train the deep learning model and obtain the model output signal;

[0024] Calculate the loss function between the model output signal and the effective sample dispersion signal;

[0025] Adjust the model parameters of the deep learning model based on the loss function;

[0026] The model training steps are iterated until the preset training termination conditions are met.

[0027] Furthermore, extracting the noisy dispersion signal from the waveform spectrum signal further includes:

[0028] The matrix bundle method is used to extract the noisy dispersion signal from the waveform spectrum signal.

[0029] According to another aspect of an embodiment of the present application, a device for acquiring borehole acoustic wave velocity based on frequency domain analysis is provided, the device comprising:

[0030] An acquisition module, adapted to acquire array waveform signals obtained by performing full-wavelength acoustic logging on a target layer;

[0031] A Fourier transform module is suitable for performing Fourier transform on the array waveform signal to obtain a waveform spectrum signal;

[0032] An extraction module, suitable for extracting a noisy dispersion signal from a waveform spectrum signal;

[0033] A noise reduction processing module is suitable for performing noise reduction processing on noisy dispersion signals to obtain effective dispersion signals;

[0034] The signal analysis and processing module is suitable for performing signal analysis and processing on the effective dispersion signal to obtain the borehole acoustic wave velocity.

[0035] According to another aspect of an embodiment of the present application, a computing device is provided, comprising: a processor, a memory, a communication interface, and a communication bus, wherein the processor, the memory, and the communication interface communicate with each other via the communication bus;

[0036] The memory is used to store at least one executable instruction, and the executable instruction enables the processor to execute operations corresponding to the above-mentioned method for acquiring borehole acoustic wave velocity based on frequency domain analysis.

[0037] According to another aspect of the embodiments of the present application, a computer storage medium is provided, which stores at least one executable instruction, and the executable instruction enables a processor to perform operations corresponding to the above-mentioned method for acquiring wellbore acoustic wave velocity based on frequency domain analysis.

[0038] According to another aspect of the present application, a computer program product is provided, comprising at least one executable instruction, wherein the executable instruction enables a processor to execute operations corresponding to the above-mentioned method for acquiring borehole acoustic wave velocity based on frequency domain analysis.

[0039] This application transforms array waveforms from the time domain to the frequency domain and calculates the noisy borehole mode wave dispersion signal from the waveform spectrum. Secondly, a targeted, trained and optimized deep learning model is used to efficiently clean the dispersion signal without human interaction. Finally, through a dispersion correction technique driven by both data and models, P-wave, S-wave, and Stoneley wave velocities are automatically acquired in both cable and LWD environments.

[0040] By converting the array waveform signals obtained from full-wavelength acoustic logging into frequency-domain processing to obtain the velocities of various mode waves, this method overcomes the problems of traditional frequency-domain processing methods, such as the need for manual interaction and slow processing efficiency, enabling the simultaneous processing of acoustic logging data and the real-time acquisition of formation sound velocity. This application can be applied to the real-time acquisition of various acoustic wave velocities, such as longitudinal waves, shear waves, and Stoneley waves, in wireline logging and logging-while-drilling scenarios. It can become a powerful tool for field data processing and real-time analysis and decision-making, while overcoming the inaccurate and unstable sound velocity processing results found in existing technologies and addressing the shortcomings of time-domain real-time sound velocity processing methods.

[0041] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Various other advantages and benefits will become apparent to those skilled in the art upon reading the detailed description of the preferred embodiment below. The accompanying drawings are for illustration purposes only and are not to be considered as limiting the present application. The same reference symbols are used throughout the drawings to represent the same components. In the drawings:

[0043] Figure 1 A schematic diagram of a flow chart of a method for acquiring borehole acoustic wave velocity based on frequency domain analysis according to one embodiment of the present application is shown;

[0044] Figure 2 A schematic diagram of an array acoustic wave waveform signal is shown;

[0045] Figure 3 A schematic diagram showing a waveform spectrum signal obtained by Fourier transforming an array acoustic waveform signal is shown;

[0046] Figure 4 A schematic diagram showing a noisy dispersion signal extracted from a waveform spectrum signal is shown;

[0047] Figure 5 shows a structural diagram of the dispersion noise reduction model;

[0048] Figure 6 A schematic diagram showing the effect of the dispersion denoising model on the denoising of a noisy dispersion signal is shown;

[0049] Figure 7 A schematic diagram showing the effect of the dispersion noise reduction model on noise-containing dispersion signals in the well section;

[0050] Figure 8 A schematic diagram showing the longitudinal wave velocity obtained by processing a monopole acoustic logging signal while drilling in a soft formation using the method of the present application;

[0051] Figure 9 A schematic diagram showing the Stoneley wave velocity obtained by processing a monopole acoustic logging while drilling signal in a soft formation using the method of the present application;

[0052] Figure 10 A schematic diagram showing the formation shear wave velocity obtained by processing the wireline dipole acoustic logging signals in hard and soft formations using the method of the present application;

[0053] Figure 11 A schematic flow chart of a method for acquiring borehole acoustic wave velocity based on frequency domain analysis according to another embodiment of the present application is shown;

[0054] Figure 12 The following is a structural block diagram of a device for acquiring borehole acoustic wave velocity based on frequency domain analysis according to one embodiment of the present application;

[0055] Figure 13 A schematic structural diagram of a computing device according to an embodiment of the present application is shown. DETAILED DESCRIPTION

[0056] Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0057] Figure 1 FIG. 1 is a flow chart of a method for acquiring borehole acoustic wave velocity based on frequency domain analysis according to an embodiment of the present application. Figure 1 As shown, the method includes the following steps:

[0058] Step S101: Acquire an array waveform signal obtained by performing full-wavelength acoustic logging on a target layer.

[0059] Specifically, full-wave acoustic logging is performed in the target interval. The entire acoustic array is collected by N receivers. The array waveform signal obtained by the nth receiver can be expressed as x n (t), thereby obtaining the array waveform signal. The array waveform signal collected here is a time domain signal.

[0060] Figure 2 A schematic diagram of an array acoustic wave waveform signal is shown, as Figure 2 As shown, it schematically shows the array waveform signals collected by 8 receivers.

[0061] To make the signal spectrum after Fourier transform dense enough, the array waveform signal in the time domain can be zero-padded before the discrete Fourier transform. This allows the total number of data points in the array waveform signal after zero-padded processing to reach a preset threshold. For example, the array waveform signal in the time domain can be zero-padded until the total number of data points in the entire array waveform signal reaches 2048. This zero-padding operation can improve the spectral resolution during the subsequent Fourier transform.

[0062] Step S102: Perform Fourier transform on the array waveform signal to obtain a waveform spectrum signal.

[0063] Specifically, the array waveform signal can be processed using discrete Fourier transform to transform the array waveform signal from the time domain to the frequency domain. After the transformation is completed, the waveform spectrum signal X corresponding to the array waveform signal can be obtained. n By transforming the array waveform signal from the time domain to the frequency domain, it is convenient to use the different frequency characteristics of different signals to perform signal denoising, separate the noise signal from the effective signal in the frequency domain, and thus provide a basis for the subsequent accurate acquisition of the sound speed.

[0064] For example, the discrete Fourier transform formula (1) can be used to transform the array waveform signal to obtain the waveform spectrum signal:

[0065] (1)

[0066] Among them, X n (ω) is the waveform spectrum signal, x n(t) is the array waveform signal, and N is the total number of receivers.

[0067] In actual production, all array waveform signals are discrete time-domain signals. Therefore, discrete Fourier transform is required to transform them into discrete waveform spectrum signals.

[0068] Figure 3 Shows the Figure 2 Schematic diagram of the waveform spectrum signal obtained after Fourier transform of the array waveform signal in . After Fourier transform processing, the array waveform signal collected by the 8 receivers is transformed from the time domain to the frequency domain.

[0069] Step S103 : extracting a noisy dispersion signal from the waveform spectrum signal, and performing noise reduction processing on the noisy dispersion signal to obtain an effective dispersion signal.

[0070] After obtaining the waveform spectrum signal X n (ω) After that, filtering methods, wavelet transform, machine learning and other methods can be used to obtain the waveform spectrum signal X n By extracting the dispersion signal from the waveform spectrum signal, it is easy to reveal the differences in the propagation characteristics of different frequency components in the signal, and then obtain the sound speed of various mode waves and facilitate noise reduction processing.

[0071] Signal noise is present not only in the frequency-domain amplitude spectrum but also in the frequency-domain phase spectrum of the array waveform. Denoising only the full-waveform spectrum signal is difficult to achieve effective noise reduction. Compared to full-band spectrum signals, dispersion signals can combine the signal's frequency-domain amplitude spectrum and phase spectrum, directly mapping the spectrum to the frequency-velocity domain. Compared to other methods, they can more directly characterize and correct for velocity dispersion effects, thereby obtaining more accurate sound velocity information.

[0072] In an optional embodiment, a matrix bundle method can be used to obtain a noisy dispersion signal from a waveform spectrum signal. The array waveform signal contains waveform signals of various modes such as Stoneley waves, longitudinal waves, flexural waves, and spiral waves. Therefore, the waveform spectrum signal can be expressed as a linear superposition of M mode wave shapes. The specific expression can be seen in formula (2):

[0073]

[0074] Among them, X n (ω) is the spectrum of the waveform collected by the nth receiver at the angular frequency ω, M is the number of mode waves, A m (ω), and v m (ω) are the amplitude, phase and noisy dispersion signal of the mth mode wave, d is the receiver spacing, and N is the total number of receivers.

[0075] Phase The matrix beam method solves the waveform phase difference between adjacent receivers by To obtain dispersion.

[0076] First, construct two (NM)×M matrices, Y1 and Y2. The specific expressions of Y1 and Y2 can be found in formula (3) and formula (4):

[0077] (3)

[0078] (4)

[0079] Among them, X n It's X n The abbreviation of (ω) (n=1,2,3,4,…,M,M+1,…NM, N-M+1,…,N-1,N) represents the waveform spectrum signal of the nth receiver at angular frequency ω. Then, the generalized eigenvalues of Y1 and Y2 are solved, and the specific expression for the solution is:

[0080] (5)

[0081] in, is the Moore-Penrose inverse of Y1, λ is the generalized eigenvalue, and I is the identity matrix of order M.

[0082] According to formula (5), we can find M generalized eigenvalues λ, and we can prove that λ is related to are equal and one-to-one corresponding, that is,

[0083] (6)

[0084] According to formula (6), the phase difference can be obtained as:

[0085] (7)

[0086] Among them, Im(λ m ) and Re(λ m ) are λ m The imaginary and real parts of , phase difference It should always be less than or equal to 0. The range of formula (7) is (-π,π]. Therefore, the range of formula (7) can be mapped from (-π,π) to [-2π,-π] by subtracting 2π from the positive result, that is, the [0,π] part of the range.

[0087] According to formula (6) and formula (7), the noisy dispersion signal corresponding to the mth mode wave can be obtained as:

[0088] (8)

[0089] Figure 4 Shown from Figure 3 Schematic diagram of extracting noisy dispersion signal from waveform spectrum signal shown, where channel 1 is the noisy dispersion signal corresponding to the first mode wave, and channel 2 is the noisy dispersion signal corresponding to the second mode wave.

[0090] The noisy dispersion signal v(ω) is subjected to noise reduction processing to obtain the effective dispersion signal v without noise T (ω), for example, a wavelet transform method, an adaptive threshold method, or an EMD method may be used for noise reduction processing.

[0091] In an optional embodiment, a dispersion denoising model is pre-trained. Therefore, the pre-trained dispersion denoising model can be used to perform denoising on the noisy dispersion signal to obtain an effective dispersion signal. By performing denoising on the noisy dispersion signal, noise interference can be suppressed, effective signal characteristics can be enhanced, and the accuracy, reliability and effectiveness of subsequent analysis can be improved.

[0092] Among them, the training process of the dispersion denoising model includes: obtaining the noisy sample dispersion signal and the effective sample dispersion signal; using the noisy sample dispersion signal to train the deep learning model to obtain the model output signal; calculating the loss function between the model output signal and the effective sample dispersion signal; based on the loss function, adjusting the model parameters of the deep learning model; iteratively executing the model training steps until the preset training termination conditions are met.

[0093] Specifically, after a large amount of processing in step S103, a data set can be constructed, wherein the data set includes a noisy sample dispersion signal and a valid sample dispersion signal, the noisy sample dispersion signal v(ω) is used as the data set feature, and the valid sample dispersion signal v T (ω) is used as the dataset label, and a deep learning model is trained based on the dataset to train a dispersion denoising model for dispersion intelligent denoising. The trained dispersion denoising model can be used to achieve automatic denoising of noisy dispersion signals.

[0094] Specifically, a noisy sample dispersion signal is input into a deep learning model. After computation and feature extraction at each layer within the model (referred to as forward propagation), the model output signal is obtained. To evaluate the difference between the model output signal and the true signal, a loss function is calculated between the model output signal and the valid sample dispersion signal. This function quantifies the magnitude of the error between the two. Based on the calculated loss function value, the model parameters of the deep learning model are adjusted using a backpropagation algorithm and an optimizer (such as stochastic gradient descent or Adam optimizer). By continuously updating the parameters, the loss function value is reduced, bringing the model output signal closer to the valid sample dispersion signal. The above model training steps are then iteratively executed, continuously inputting the noisy sample dispersion signal, calculating the loss function, and adjusting the model parameters until a preset training termination criterion is met. For example, the loss function value drops below a certain threshold, the maximum number of training rounds is reached, or performance on the validation set no longer improves. At this point, the deep learning model training is considered complete, resulting in a dispersion denoising model.

[0095] Before model training, the deep learning model is initialized first. The deep learning model includes a one-dimensional convolution layer, a batch normalization layer, an activation function, a pooling layer, and a deconvolution layer. The structure of the deep learning model is as follows: Figure 5 As shown in the figure, the forward propagation process of this deep learning model is as follows: the encoder, consisting of one-dimensional convolutional layers, extracts local features and resizes the dimensionality of the input noisy sample dispersion signal. The pooling layer performs a local maximum operation on the features output by the encoder to reduce the spatial size of the feature map, thereby reducing computational complexity and extracting key features. Multiple encoders are combined to extract signal features at different resolutions, which are then processed by the deconvolution layer to restore the length of the noisy sequence. Finally, a sigmoid activation function clamps the output of the transposed convolution layer to the range [0, 1].

[0096] When using the dispersion denoising model for denoising, the processing process is similar to the forward propagation process in the training process and will not be repeated here.

[0097] The output of the dispersion denoising model has the same shape as the noisy dispersion signal v(ω), and the magnitude of its output value represents the validity probability of the data point in the noisy dispersion signal v(ω). The set of data points with a validity probability greater than 0.9 constitutes the effective dispersion signal v after denoising. T (ω). Figure 6 The dispersion noise reduction model is shown to be Figure 4 The schematic diagram of the effect of noise reduction processing on the noisy dispersion signal shown in the figure is as follows. Figure 7 A schematic diagram shows the effect of the dispersion noise reduction model on the noise dispersion signal of the well section. The unit of the noisy dispersion signal is converted from m / s in the velocity domain to μs / ft in the slowness domain before display. Figure 7The first track is the depth, the second track is the array waveform, the third track is the noisy dispersion signal extracted from the waveform, and the fourth track is the effective dispersion signal after denoising using the dispersion denoising model.

[0098] Step S104: performing signal analysis processing on the effective dispersion signal to obtain the borehole acoustic wave velocity.

[0099] The effective dispersion signal is the dispersion signal after noise reduction processing and is noise-free. Therefore, performing signal analysis on the effective dispersion signal accurately derives the borehole acoustic velocity. This signal analysis primarily involves frequency distribution statistics and fitting. Different processing methods are used for the effective dispersion signal of different wave modes.

[0100] For example, if the effective dispersion signal is a Stoneley wave dispersion signal and / or a longitudinal wave dispersion signal, the frequency distribution statistics of the Stoneley wave dispersion signal and / or the longitudinal wave dispersion signal are performed, and the velocity corresponding to the maximum frequency is determined as the Stoneley wave sound velocity and / or the longitudinal wave sound velocity. Figure 8 A schematic diagram showing the longitudinal wave velocity obtained by processing a monopole acoustic logging signal while drilling in a soft formation using the method of the present application; Figure 9 A schematic diagram showing the Stoneley wave velocity obtained by processing a while-drilling monopole acoustic logging signal in a soft formation using the method of the present application is shown.

[0101] If the effective dispersion signal is a bending wave dispersion signal and / or a spiral wave dispersion signal, the bending wave dispersion signal and / or the spiral wave dispersion signal is fitted to obtain the shear wave velocity. For example, the fitting process can be performed using a nonlinear least squares method or a neural network. Other suitable fitting methods are also within the scope of protection of this application.

[0102] In order to improve the fitting efficiency, the pre-trained dispersion fast simulation network can be used to obtain the theoretical simulation dispersion signal. Specifically, during well logging, the borehole acoustic parameters are obtained, among which the borehole acoustic parameters mainly include: the high sensitivity parameter of dispersion, the borehole mud sound velocity v f , wellbore diameter R and formation shear wave velocity v s , input the borehole acoustic parameters into the pre-trained dispersion fast simulation network to obtain the corresponding theoretical simulation dispersion signal v sim (ω), the measured effective dispersion signal is compared with the theoretical simulation dispersion signal v sim The effective dispersion signal refers to the flexural wave dispersion signal for dipole logging data and the spiral wave dispersion signal for quadrupole logging data.

[0103] For example, the theoretical simulation dispersion signal v sim (ω) and the effective dispersion signal v TThe Euclidean norm of (ω) (flexural wave dispersion signal or spiral wave dispersion signal) is used as the objective function to represent the difference between the two. The grid search method is used to find the optimal formation shear wave velocity that can minimize the objective function. The search result is the shear wave velocity. Figure 10 The figure shows the schematic diagram of the formation shear wave velocity obtained by processing the cable dipole acoustic logging signal in hard and soft formations using the method of the present application. Figure 10 The part above 3681.5m is hard stratum, and the part below is soft stratum.

[0104] In an optional embodiment, the training process of the dispersion fast simulation network includes: generating a preset number of theoretical simulation dispersion signals based on the dispersion equation of the wellbore acoustic field; obtaining wellbore acoustic sample parameters, and training the neural network using the wellbore acoustic sample parameters as training data set features and the theoretical simulation dispersion signals as training data set labels to obtain the dispersion fast simulation network, wherein the wellbore acoustic sample parameters include: wellbore mud sound velocity, well diameter, and formation shear wave sound velocity.

[0105] When conducting network training, it is necessary to prepare the data set required for training. The data set required for training mainly includes: borehole acoustic sample parameters and theoretical simulation dispersion signals. The theoretical simulation dispersion signal can be generated based on the dispersion equation of the borehole acoustic field. The dispersion equation of the borehole acoustic field can be found in formula (9):

[0106] (9)

[0107] Where k is the wave number, ω is the angular frequency, B represents the waveguide in the wellbore, the influence of the wellbore mud and the logging instrument, and F represents the relevant parameters of the elastic isotropic formation, including the P-wave velocity and the S-wave velocity and the formation density, which vary with the radial distance r. By numerically solving the dispersion equation at each frequency ω, the wave number k of the flexural wave and / or spiral wave can be obtained. The velocity is inversely proportional to the wave number k, and the corresponding velocity can be obtained to obtain the theoretical simulation dispersion signal v of the flexural wave and / or spiral wave. sim (ω), these signals reflect the propagation characteristics of acoustic waves in the wellbore acoustic field at different frequencies.

[0108] Wellbore acoustic sample parameters are highly sensitive parameters of dispersion, mainly including: wellbore mud sound velocity v f , wellbore diameter R and formation shear wave velocity v sThese parameters significantly influence the dispersion characteristics of the borehole acoustic field. The neural network is trained using the borehole acoustic sample parameters as dataset features and the theoretically simulated dispersion signal as the dataset label (i.e., the target reference for model learning). Specifically, the training dataset is input into the neural network, and by adjusting the network's internal parameters, the model learns the mapping relationship between the borehole acoustic sample parameters and the theoretically simulated dispersion signal. After multiple rounds of iterative training, a rapid dispersion simulation network is ultimately obtained. This network can quickly and accurately simulate the theoretically simulated dispersion signal under different borehole conditions, providing strong support for subsequent borehole acoustic wave analysis and formation property assessment. The rapid dispersion simulation network is a fully connected neural network with a single hidden layer that is used to rapidly generate the theoretically simulated dispersion signal, improving fitting efficiency.

[0109] In summary, this application transforms array waveforms from the time domain to the frequency domain and calculates the noisy borehole mode wave dispersion signal from the waveform spectrum. Secondly, a targeted, trained and optimized deep learning model is used to achieve efficient data cleaning of the dispersion signal without human interaction. Finally, through a dispersion correction technology driven by both data and models, P-wave, S-wave, and Stoneley wave velocities are automatically acquired in both cable and LWD environments.

[0110] By converting the array waveform signals obtained from full-wavelength acoustic logging into frequency-domain processing to obtain the velocities of various mode waves, this method overcomes the problems of traditional frequency-domain processing methods, such as the need for manual interaction and slow processing efficiency, enabling simultaneous processing of acoustic logging data and achieving real-time acquisition of formation sound velocity. This application can be applied to real-time acquisition of various acoustic wave velocities, such as longitudinal waves, shear waves, and Stoneley waves, in wireline logging and logging-while-drilling scenarios. It can become a powerful tool for field data processing and real-time analysis and decision-making, while overcoming the inaccurate and unstable sound velocity processing results found in existing technologies and addressing the shortcomings of time-domain real-time sound velocity processing methods.

[0111] Figure 11 FIG. 1 is a flow chart of a method for acquiring borehole acoustic wave velocity based on frequency domain analysis according to another embodiment of the present application. Figure 11 As shown in the figure, full-wave acoustic ranging is performed in the target layer, and the array waveform signal is collected by the receiver. Then, the array waveform signal is processed by fast Fourier transform to obtain the waveform spectrum signal. Based on the preset dispersion extraction method, the mode wave dispersion signal is extracted from the waveform spectrum signal ( Figure 1The noisy dispersion signal in the embodiment shown is used), and the dispersion noise reduction model is used to perform intelligent noise reduction processing on the above-mentioned mode wave dispersion signal to obtain an effective dispersion signal. For the effective dispersion signal, it can be determined whether fitting processing is required. If fitting processing is required, a theoretical simulation dispersion signal can be obtained by using a dispersion fast simulation network. The effective dispersion signal and the theoretical simulation dispersion signal are fitted to determine the wellbore mode wave sound velocity. If fitting processing is not required, the wellbore mode wave sound velocity can be determined by frequency statistics.

[0112] Figure 12 FIG. 1 shows a structural block diagram of a wellbore acoustic wave velocity acquisition device based on frequency domain analysis according to an embodiment of the present application. Figure 12 As shown, the device includes:

[0113] An acquisition module 1201 is adapted to acquire an array waveform signal obtained by performing full-wavelength acoustic logging on a target layer interval;

[0114] The Fourier transform module 1202 is adapted to perform Fourier transform on the array waveform signal to obtain a waveform spectrum signal;

[0115] Extraction module 1203, adapted to extract the noisy dispersion signal from the waveform spectrum signal;

[0116] A noise reduction processing module 1204 is adapted to perform noise reduction processing on the noisy dispersion signal to obtain an effective dispersion signal;

[0117] The signal analysis and processing module 1205 is adapted to perform signal analysis and processing on the effective dispersion signal to obtain the borehole acoustic wave velocity.

[0118] Optionally, the signal analysis and processing module is further adapted to: if the effective dispersion signal is a Stoneley wave dispersion signal and / or a longitudinal wave dispersion signal, perform frequency distribution statistics on the Stoneley wave dispersion signal and / or the longitudinal wave dispersion signal, and determine the velocity corresponding to the maximum frequency as the Stoneley wave sound velocity and / or the longitudinal wave sound velocity;

[0119] If the effective dispersion signal is a bending wave dispersion signal and / or a spiral wave dispersion signal, a fitting process is performed on the bending wave dispersion signal and / or the spiral wave dispersion signal to obtain the shear wave speed.

[0120] Optionally, the signal analysis and processing module is further adapted to: input wellbore acoustic parameters into a pre-trained dispersion fast simulation network to obtain corresponding theoretical simulation dispersion signals, wherein the wellbore acoustic parameters include: wellbore mud sound velocity, well diameter, and formation shear wave sound velocity;

[0121] The shear wave velocity of the formation at which the difference between the bending wave dispersion signal and the theoretical simulation dispersion signal is minimum is determined as the shear wave velocity, and / or the shear wave velocity of the formation at which the difference between the spiral wave dispersion signal and the theoretical simulation dispersion signal is minimum is determined as the shear wave velocity.

[0122] Optionally, the apparatus further comprises: a training module for a fast dispersion simulation network, adapted to generate a preset number of theoretical simulation dispersion signals based on a dispersion equation of the wellbore acoustic field;

[0123] The neural network is trained using the borehole acoustic sample parameters, including wellbore mud velocity, wellbore diameter, and formation shear wave velocity, as training dataset features and theoretically simulated dispersion signals as training dataset labels to obtain a fast dispersion simulation network.

[0124] Optionally, the noise reduction processing module is further adapted to: perform noise reduction processing on the noisy dispersion signal using a pre-trained dispersion noise reduction model to obtain an effective dispersion signal;

[0125] The device also includes: a training module of a dispersion noise reduction model, suitable for obtaining a noisy sample dispersion signal and a valid sample dispersion signal;

[0126] Use the noisy sample dispersion signal to train the deep learning model and obtain the model output signal;

[0127] Calculate the loss function between the model output signal and the effective sample dispersion signal;

[0128] Adjust the model parameters of the deep learning model based on the loss function;

[0129] The model training steps are iterated until the preset training termination conditions are met.

[0130] Optionally, the extraction module is further adapted to extract the noisy dispersion signal from the waveform spectrum signal using a matrix bundle method.

[0131] In summary, this application transforms array waveforms from the time domain to the frequency domain and calculates the noisy borehole mode wave dispersion signal from the waveform spectrum. Secondly, a targeted, trained and optimized deep learning model is used to achieve efficient data cleaning of the dispersion signal without human interaction. Finally, through a dispersion correction technology driven by both data and models, P-wave, S-wave, and Stoneley wave velocities are automatically acquired in both cable and LWD environments.

[0132] By converting the array waveform signals obtained from full-wavelength acoustic logging into frequency-domain processing to obtain the velocities of various mode waves, this method overcomes the problems of traditional frequency-domain processing methods, such as the need for manual interaction and slow processing efficiency, enabling simultaneous processing of acoustic logging data and achieving real-time acquisition of formation sound velocity. This application can be applied to real-time acquisition of various acoustic wave velocities, such as longitudinal waves, shear waves, and Stoneley waves, in wireline logging and logging-while-drilling scenarios. It can become a powerful tool for field data processing and real-time analysis and decision-making, while overcoming the inaccurate and unstable sound velocity processing results found in existing technologies and addressing the shortcomings of time-domain real-time sound velocity processing methods.

[0133] An embodiment of the present application provides a non-volatile computer storage medium, which stores at least one executable instruction or computer program, which enables a processor to perform operations corresponding to the method for acquiring borehole acoustic wave velocity based on frequency domain analysis in any of the above method embodiments.

[0134] An embodiment of the present application provides a computer program product, which includes at least one executable instruction or computer program, which can enable a processor to perform operations corresponding to the method for obtaining wellbore acoustic wave velocity based on frequency domain analysis in any of the above method embodiments.

[0135] Figure 13 A schematic structural diagram of a computing device according to an embodiment of the present application is shown. The specific embodiment of the present application does not limit the specific implementation of the computing device.

[0136] like Figure 13 As shown, the computing device may include a processor 1302 , a communication interface 1304 , a memory 1306 , and a communication bus 1308 .

[0137] The processor 1302 , the communication interface 1304 , and the memory 1306 communicate with each other via a communication bus 1308 .

[0138] The communication interface 1304 is used to communicate with other devices such as clients or other servers.

[0139] The processor 1302 is configured to execute the program 1310 , and specifically to execute the relevant steps in the above-mentioned embodiment of the method for acquiring borehole acoustic wave velocity based on frequency domain analysis.

[0140] Specifically, the program 1310 may include program codes, which include computer operation instructions.

[0141] Processor 1302 may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application. The one or more processors included in a computing device may be processors of the same type, such as one or more CPUs, or processors of different types, such as one or more CPUs and one or more ASICs.

[0142] The memory 1306 is used to store the program 1310. The memory 1306 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk storage.

[0143] Program 1310 can specifically be used to cause processor 1302 to execute the method for acquiring borehole acoustic wave velocity based on frequency domain analysis described in any of the aforementioned method embodiments. The specific implementation of each step in program 1310 can be found in the descriptions of the corresponding steps and units in the aforementioned method for acquiring borehole acoustic wave velocity based on frequency domain analysis, and will not be repeated here. Those skilled in the art will clearly understand that, for ease and brevity of description, the specific operating processes of the aforementioned devices and modules can refer to the corresponding process descriptions in the aforementioned method embodiments, and will not be repeated here.

[0144] The algorithms and displays provided herein are not inherently related to any particular computer, virtual system, or other device. Various general-purpose systems may also be used together with the teachings herein. Based on the above description, it is apparent that the structure required for constructing such systems is suitable. In addition, the present application is not directed to any specific programming language. It should be understood that various programming languages may be utilized to implement the present application described herein, and the description of the specific languages above is provided for the purpose of disclosing the preferred embodiment of the present application.

[0145] In the description provided herein, a large number of specific details are described. However, it is understood that the embodiments of the present application can be practiced without these specific details. In some instances, well-known methods, structures, and techniques are not shown in detail so as not to obscure the understanding of this description.

[0146] Similarly, it should be understood that in order to streamline the present disclosure and aid understanding of one or more of the various inventive aspects, in the above description of the exemplary embodiments of the present application, various features of the present application are sometimes grouped together into a single embodiment, figure, or description thereof. However, this disclosed method should not be interpreted as reflecting an intention that the claimed application requires more features than are expressly recited in each claim. Rather, as reflected in the claims, inventive aspects lie in less than all the features of the individual embodiments disclosed above. Accordingly, the claims that follow the detailed description are hereby expressly incorporated into this detailed description, with each claim standing on its own as a separate embodiment of the present application.

[0147] Those skilled in the art will appreciate that the modules in the devices of the embodiments can be adaptively modified and installed in one or more devices different from the embodiments. The modules, units, or components in the embodiments can be combined into a single module, unit, or component, and furthermore, they can be divided into multiple sub-modules, sub-units, or sub-components. All features disclosed in this specification (including the accompanying claims, abstract, and drawings), and all processes or units of any method or device disclosed therein, can be combined in any combination, unless at least some of such features and / or processes or units are mutually exclusive. Unless expressly stated otherwise, each feature disclosed in this specification (including the accompanying claims, abstract, and drawings) may be replaced by an alternative feature that serves the same, equivalent, or similar purpose.

[0148] Furthermore, those skilled in the art will appreciate that although some embodiments herein include certain features included in other embodiments but not other features, combinations of features from different embodiments are intended to be within the scope of this application and to form different embodiments. For example, in the claims, any of the claimed embodiments may be used in any combination.

[0149] The various component embodiments of the present application can be implemented in hardware, or as software modules running on one or more processors, or a combination thereof. Those skilled in the art will appreciate that in practice, a microprocessor or digital signal processor (DSP) can be used to implement some or all of the functionality of some or all of the components according to the embodiments of the present application. The present application can also be implemented as a device or apparatus program (e.g., a computer program or computer program product) for performing part or all of the methods described herein. Such a program implementing the present application can be stored on a computer-readable medium or in the form of one or more signals. Such signals can be downloaded from an Internet website, provided on a carrier signal, or provided in any other form.

[0150] It should be noted that the above embodiments illustrate rather than limit the present application, and that a person skilled in the art may devise alternative embodiments without departing from the scope of the appended claims. In the claims, any reference signs placed between brackets should not be construed as limiting the claims. The word "comprising" does not exclude the presence of elements or steps not listed in the claims. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The present application may be implemented by means of hardware comprising several different elements and by means of a suitably programmed computer. In a unit claim enumerating several means, several of these means may be embodied by the same item of hardware. The use of the words first, second and third etc. does not indicate any order. These words may be interpreted as names.

Claims

1. A method for obtaining borehole acoustic wave velocity based on frequency domain analysis, characterized in that: The method comprises: Acquire array waveform signals obtained by full-wavelength acoustic logging of the target layer; Performing Fourier transform on the array waveform signal to obtain a waveform spectrum signal; A noisy dispersion signal is extracted from the waveform spectrum signal, and a pre-trained dispersion denoising model is used to perform denoising on the noisy dispersion signal to obtain an effective dispersion signal, wherein the dispersion denoising model is a deep learning model obtained by training the noisy sample dispersion signal and the effective sample dispersion signal, and the deep learning model includes a one-dimensional convolution layer, a batch normalization layer, an activation function, a pooling layer, and a deconvolution layer. The forward propagation processing process of the deep learning model includes: an encoder composed of a one-dimensional convolution layer extracts local features and adjusts the dimension of the input noisy sample dispersion signal; the pooling layer performs a local maximum operation on the features output by the encoder and extracts features; a combination of multiple encoders is used to extract signal features of different resolutions, and the deconvolution layer processes the extracted signal features of different resolutions to restore the length of the noisy sequence; and the output of the transposed convolution layer is limited to [0, 1] by a Sigmoid activation function. The output of the dispersion denoising model has the same shape as the noisy dispersion signal. The size of the output value represents the validity probability of the data point in the noisy dispersion signal. The set of data points with a validity probability greater than 0.9 constitutes the effective dispersion signal after denoising. If the effective dispersion signal is a Stoneley wave dispersion signal and / or a longitudinal wave dispersion signal, frequency distribution statistics are performed on the Stoneley wave dispersion signal and / or the longitudinal wave dispersion signal, and a velocity corresponding to a maximum frequency is determined as the Stoneley wave sound velocity and / or the longitudinal wave sound velocity; If the effective dispersion signal is a bending wave dispersion signal and / or a spiral wave dispersion signal, the wellbore acoustic parameters are input into a pre-trained dispersion fast simulation network to obtain a corresponding theoretical simulation dispersion signal, wherein the wellbore acoustic parameters include: wellbore mud sound velocity, well diameter and formation shear wave sound velocity; the formation shear wave sound velocity at which the difference between the bending wave dispersion signal and the theoretical simulation dispersion signal is minimized is determined as the shear wave sound velocity, and / or the formation shear wave sound velocity at which the difference between the spiral wave dispersion signal and the theoretical simulation dispersion signal is minimized is determined as the shear wave sound velocity.

2. The method for obtaining borehole acoustic wave velocity based on frequency domain analysis according to claim 1, characterized in that: The training process of the fast dispersion simulation network includes: generating a preset number of theoretical simulation dispersion signals based on the dispersion equation of the wellbore acoustic field; Wellbore acoustic sample parameters are obtained, and the neural network is trained using the wellbore acoustic sample parameters as training data set features and the theoretically simulated dispersion signals as training data set labels to obtain a dispersion fast simulation network, wherein the wellbore acoustic sample parameters include: wellbore mud velocity, wellbore diameter, and formation shear wave velocity.

3. The method for acquiring borehole acoustic wave velocity based on frequency domain analysis according to claim 1 or 2, characterized in that: The performing noise reduction processing on the noisy dispersion signal to obtain an effective dispersion signal further comprises: Using a pre-trained dispersion denoising model to perform denoising on the noisy dispersion signal to obtain an effective dispersion signal; The training process of the dispersion denoising model includes: Obtaining a noisy sample dispersion signal and a valid sample dispersion signal; Use the noisy sample dispersion signal to train the deep learning model and obtain the model output signal; Calculating a loss function between the model output signal and the effective sample dispersion signal; Adjusting model parameters of the deep learning model based on the loss function; The model training steps are iterated until the preset training termination conditions are met.

4. The method for acquiring borehole acoustic wave velocity based on frequency domain analysis according to claim 1 or 2, characterized in that: The extracting of the noisy dispersion signal from the waveform spectrum signal further comprises: A noisy dispersion signal is extracted from the waveform spectrum signal using a matrix bundle method.

5. A device for acquiring borehole acoustic wave velocity based on frequency domain analysis, characterized in that: The device comprises: An acquisition module, adapted to acquire array waveform signals obtained by performing full-wavelength acoustic logging on a target layer; A Fourier transform module, adapted to perform Fourier transform on the array waveform signal to obtain a waveform spectrum signal; An extraction module, adapted to extract a noisy dispersion signal from the waveform spectrum signal; The denoising processing module is adapted to perform denoising on the noisy dispersion signal using a pre-trained dispersion denoising model to obtain an effective dispersion signal, wherein the dispersion denoising model is a deep learning model obtained by training the noisy sample dispersion signal and the effective sample dispersion signal, and the deep learning model includes a one-dimensional convolution layer, a batch normalization layer, an activation function, a pooling layer, and a deconvolution layer. The forward propagation processing process of the deep learning model includes: an encoder composed of a one-dimensional convolution layer extracts local features and adjusts the dimension of the input noisy sample dispersion signal; the pooling layer performs a local maximum operation on the features output by the encoder and extracts features; a combination of multiple encoders is used to extract signal features of different resolutions, and the deconvolution layer processes the extracted signal features of different resolutions to restore the length of the noisy sequence; and the output of the transposed convolution layer is limited to [0, 1] by a Sigmoid activation function. The output of the dispersion denoising model has the same shape as the noisy dispersion signal. The size of the output value represents the validity probability of the data point in the noisy dispersion signal. The set of data points with a validity probability greater than 0.9 constitutes the effective dispersion signal after denoising. The signal analysis and processing module is suitable for, if the effective dispersion signal is a Stoneley wave dispersion signal and / or a longitudinal wave dispersion signal, performing frequency distribution statistics on the Stoneley wave dispersion signal and / or the longitudinal wave dispersion signal, and determining the velocity corresponding to the maximum frequency as the Stoneley wave sound velocity and / or the longitudinal wave sound velocity; if the effective dispersion signal is a flexural wave dispersion signal and / or a spiral wave dispersion signal, inputting the wellbore acoustic parameters into a pre-trained dispersion fast simulation network to obtain a corresponding theoretical simulation dispersion signal, wherein the wellbore acoustic parameters include: wellbore mud sound velocity, wellbore diameter and formation shear wave sound velocity; determining the formation shear wave sound velocity when the difference between the flexural wave dispersion signal and the theoretical simulation dispersion signal is the smallest as the shear wave sound velocity, and / or determining the formation shear wave sound velocity when the difference between the spiral wave dispersion signal and the theoretical simulation dispersion signal is the smallest as the shear wave sound velocity.

6. A computing device, characterized in that include: A processor, a memory, a communication interface, and a communication bus, wherein the processor, the memory, and the communication interface communicate with each other via the communication bus; The memory is used to store at least one executable instruction, and the executable instruction enables the processor to perform operations corresponding to the method for acquiring borehole acoustic wave velocity based on frequency domain analysis according to any one of claims 1 to 4.

7. A computer storage medium, characterized in that The computer storage medium stores at least one executable instruction, which enables the processor to perform operations corresponding to the method for acquiring borehole acoustic wave velocity based on frequency domain analysis according to any one of claims 1 to 4.

8. A computer program product, characterized in that The method comprises at least one executable instruction, wherein the executable instruction enables a processor to execute operations corresponding to the method for acquiring borehole acoustic wave velocity based on frequency domain analysis according to any one of claims 1 to 4.

Images