Pump noise suppression method and system for continuous wave mud pulse signals while drilling

Abstract

The invention belongs to the technical field of oil exploration, and relates to a pump noise suppression method and system for a while-drilling continuous wave mud pulse signal, and the method comprises the steps: collecting an original mud pulse signal containing a continuous wave pulse signal; pump noise fundamental waves are determined according to the recognized fundamental wave frequency of the pump noise in the original mud pulse signals; constructing a linear time-invariant state space model of the k-order harmonics according to an unsteady state pump noise model constructed by utilizing the pump noise fundamental waves, and constructing a linear time-invariant measurement model of the k-order harmonics; performing pump noise prediction on the constructed linear time-invariant state space model by using adaptive Kalman filtering to obtain an optimal pump noise posteriori estimated value at each moment; reconstructing a pump noise signal in the continuous wave mud pulse observation signal calculated by the linear time-invariant measurement model through the pump noise posterior estimation value to obtain a reconstructed pump noise signal containing all order harmonics; and subtracting the reconstructed noise signal from the original mud pulse signal to obtain a continuous wave mud pulse signal after the pump noise is suppressed. According to the invention, unstable pump noise signals can be efficiently and reliably inhibited.

Description

Technical Field

[0001] The invention belongs to the technical field of oil and gas exploration, and relates to a measurement while drilling technology, and in particular to a pump noise suppression method and system for a continuous wave mud pulse signal while drilling. Background Art

[0002] In the process of oil exploration, the measurement of downhole information is a key link to ensure accurate and efficient drilling operations. Through the measurement while drilling (MWD) technology, downhole information can be obtained in real time by measuring short circuits during drilling, and these downhole information can be transmitted to the ground system through certain specific media through coding and modulation. The ground system will process, store and display the signals, restore the downhole transmission data, and improve drilling efficiency. With the continuous development of measurement while drilling technology, the scope of measurement objects is also expanding. From the initial basic measurement data such as well inclination, azimuth and tool face parameters, it has gradually developed to the measurement of geological parameters such as formation pressure, gamma rays, formation density, neutron porosity, etc., and real-time logging and guidance functions can be achieved through this information.

[0003] In measurement while drilling, the data transmission method is crucial. At present, common signal transmission methods mainly include cable transmission, mud pulse transmission, acoustic transmission and electromagnetic transmission. Among them, mud pulse transmission technology is the most widely used. It uses drilling mud as the transmission medium and transmits the downhole measurement data to the ground system in real time through the positive pulse, negative pulse or continuous wave pulse generated by the downhole pulse generator. Among these transmission methods, continuous wave mud pulse transmission has become the mainstream measurement while drilling signal transmission method due to its advantages such as stability, flexibility and reliability.

[0004] However, due to the complex downhole environment, continuous wave mud pulse transmission technology still faces many challenges. First, when the mud pump provides power, it will generate strong periodic pressure noise with extremely high amplitude, which is the main noise source affecting the signal transmission quality. Secondly, when the drill bit cuts the rock formation or collides with the well wall, it will generate strong mechanical vibrations, which will also trigger additional noise signals. Therefore, the mud pulse signal received by the ground pressure sensor often contains a large amount of background noise, resulting in a significant reduction in the signal-to-noise ratio of the downhole information. What is more serious is that when the speed of the mud pump is unstable, the harmonic spectrum of the pump noise will change accordingly, and the unstable spectrum overlaps with the pulse signal, further increasing the difficulty of signal extraction. The weak pulse signal is easily annihilated in the complex and changeable background noise, which seriously affects the accurate analysis of the downhole information by the ground system and even makes the measurement while drilling system fail. Therefore, how to effectively reduce noise interference and improve the quality and transmission rate of the signal is still a technical problem that needs to be solved in the field of measurement while drilling technology.

[0005] A Chinese patent with publication number CN 116955941 B discloses a method for denoising continuous wave signals during drilling measurement. The method uses a bandpass filter to remove pump noise and other noise outside the pulse signal frequency band, and uses an unscented Kalman filter to reconstruct and remove pump noise within the frequency band. This method can remove mud pump noise in real time under extremely low signal-to-noise ratio conditions, accurately identify and extract continuous wave signals, and still has effective denoising capabilities when the signal and noise frequencies interfere with each other at the same frequency. The denoising method of this patent requires multiple iterations to adjust the unscented Kalman filter parameters, and the algorithm complexity is relatively high, which limits its application in high real-time environments.

[0006] A Chinese patent with publication number CN 111535802 B discloses a mud pulse signal processing method, which sequentially processes the mud pulse signal collected by a single pressure sensor through pump noise state space model construction, Kalman filtering, random noise removal, baseline drift removal, and adaptive threshold setting to obtain an obvious effective transmission pulse signal. The patented method is simple and reliable, and can quickly and reliably identify effective pulse signals, but the method is highly dependent on the setting of Kalman filter parameters. If the signal characteristics suddenly change, the filtering effect may be affected.

[0007] The Chinese patent application with publication number CN 111832332 A discloses a mud pulse signal processing method and device, which uses frame synchronization wavelet denoising to perform first-level forced denoising on the mud pulse sampling signal, and then uses command wavelet denoising to perform second-level forced denoising on the original command signal to obtain a denoised command signal, thereby improving the denoising effect of the mud pulse signal. However, the denoising method of the patent application requires precise frame synchronization and command signal detection, has too high requirements on signal quality, and the denoising performance is greatly affected by the selected wavelet basis function and threshold.

[0008] A Chinese patent with publication number CN 115059458 A discloses a method for generating and identifying mud pulse signals for downhole measurement while drilling. The method generates signals of different waveforms by rotating the encoder disk and records the rotation angle. The high-frequency ground acquisition signal is low-pass filtered and smoothed to extract the effective mud pulse pressure signal. The amplitude comparison approximation method and the gray correlation algorithm are used to identify the signal coding and measure downhole information in real time. This patent application effectively obtains downhole measurement information through the dual coding form of frequency and amplitude, thereby improving the identifiability of downhole signals. However, the encoder disk structure of this patent application is complex and easily affected by the downhole environment. The gray correlation algorithm and phase analysis process are complex, which increases the difficulty of real-time processing.

[0009] From the above, it can be seen that although the existing methods for removing pump noise interference can eliminate pump impulse noise, they still have problems such as poor real-time performance, over-reliance on filtering parameters and relatively poor denoising effect. Summary of the invention

[0010] In view of the above-mentioned problems of poor denoising effect in the prior art, the present invention provides a simple and reliable pump noise suppression method and system for while-drilling continuous wave mud pulse signals, which can effectively suppress unstable pump noise signals and adapt to different working conditions without excessive adjustment of filtering parameters.

[0011] In order to achieve the above object, the present invention provides a method for suppressing pump noise of a continuous wave mud pulse signal while drilling, the steps of which are as follows:

[0012] S1, collecting the original mud pulse signal including the continuous wave pulse signal;

[0013] S2, identifying the fundamental frequency of the pump noise in the original mud pulse signal, and determining the fundamental wave of the pump noise according to the fundamental frequency;

[0014] S3, constructing a non-steady-state pump noise model using the pump noise fundamental wave, constructing a linear time-invariant state space model of the kth harmonic based on the non-steady-state pump noise model, and constructing a linear time-invariant measurement model of the kth harmonic based on the set measurement noise;

[0015] S4, using adaptive Kalman filtering to predict pump noise on the constructed linear time-invariant state space model to obtain the best a posteriori estimate of pump noise at each moment;

[0016] S5, reconstructing the pump noise signal in the continuous wave mud pulse observation signal calculated by the linear time-invariant measurement model through the pump noise posterior estimation value to obtain a reconstructed pump noise signal containing all order harmonics;

[0017] S6. The original mud pulse signal minus the reconstructed noise signal is used to obtain a continuous wave mud pulse signal after suppressing pump noise.

[0018] In some embodiments, in step S2, the method for identifying the fundamental frequency of the pump noise in the original mud pulse signal is: performing a fast Fourier transform on the original mud pulse signal to obtain the original mud pulse signal spectrum, and finding the fundamental frequency of the pump noise from the spectrum; when the low-frequency noise covers the fundamental frequency of the pump noise, performing a cepstrum analysis on the original mud pulse signal to obtain a cepstrum, and obtaining the fundamental frequency by solving the inverse of the horizontal coordinate time at the peak value in the cepstrum.

[0019] In some embodiments, in step S2, the method for determining the pump noise fundamental wave according to the fundamental wave frequency is:

[0020] Calculate the fundamental wave angular frequency according to the fundamental wave frequency;

[0021] ω0=2πf0

[0022] In the formula, ω0 is the fundamental angular frequency, f0 is the fundamental frequency;

[0023] Calculate the pump noise fundamental wave according to the fundamental wave angular frequency;

[0024]

[0025] In the formula, is the fundamental wave of pump noise at time n; n is the continuous sampling point, n=1,2,...,N, N is the total number of sampling points; A m is the amplitude of the fundamental wave of pump noise; φ is the phase of the fundamental wave of pump noise.

[0026] In some embodiments, in step S3, the method of constructing a non-steady-state pump noise model using the pump noise fundamental wave is:

[0027] Using the trigonometric equation, the pump noise fundamental is expressed as:

[0028]

[0029] In the formula, is the fundamental wave of pump noise at time n+1; is the fundamental wave of pump noise at time n-1;

[0030] Introduce a zero-mean random error w n , the non-steady-state pump noise model is constructed using the pump noise fundamental wave and is expressed as:

[0031]

[0032] In some embodiments, in step S3, the method of constructing a linear time-invariant state-space model of the kth harmonic according to the unsteady-state pump noise model is:

[0033] Assume that the state vector of the pump noise containing k-order harmonics is Among them, each element is represented as:

[0034] The linear time-invariant state space model of the kth harmonic is constructed based on the unsteady-state pump noise model and is expressed as:

[0035] p n =Ap n-1 +Bw n

[0036] In the formula, p n-1 is the state vector of the pump noise at time n-1; A is the state transfer matrix, B is the control matrix, T is transpose.

[0037] In some embodiments, in step S3, a linear time-invariant measurement model of the kth harmonic is constructed according to the set measurement noise and is expressed as:

[0038] y n =H T p n +v n

[0039] In the formula, y n is the measured value of the continuous wave mud pulse signal; H is the measurement transfer matrix, T is transpose; v n To measure noise.

[0040] In some embodiments, in step S4, the specific method of using adaptive Kalman filtering to predict the posterior estimate of pump noise is:

[0041] State update step: Calculate the prior estimate based on the state transfer matrix A

[0042]

[0043] In the formula, is the posterior estimate at time n-1;

[0044] Calculate the measurement value y according to the measurement transfer matrix H n and the prior estimate The residual

[0045]

[0046] Take L as the window size and calculate the residual The mean of the variance

[0047]

[0048] Assume T λ is the threshold of the adaptive factor, and T λ ≥0, calculate the adaptive factor λ:

[0049]

[0050] In the formula, is the posterior estimation error covariance at time n-1, q n is the process noise covariance, r nTo measure the noise covariance, the adaptive factor λ is adaptively changed according to the following formula to adjust the process noise covariance q n , adjust the notch properties of different harmonics and calculate the prior estimation error covariance

[0051]

[0052] Kalman gain calculation steps: Estimation of error covariance based on prior Calculate the Kalman gain K n ;

[0053]

[0054] Measurement update step: Based on the prior estimate and the Kalman gain K n Update the posterior estimate

[0055]

[0056] According to the prior estimate error covariance and the Kalman gain K n Update the posterior estimate error covariance

[0057]

[0058] By looping and recursively repeating the state update step to the measurement update step, the best a posteriori estimate of the pump noise at each moment can be obtained. In some embodiments, in step S5, the reconstructed pump noise signal is expressed as:

[0059]

[0060] In the formula, B n To reconstruct the pump noise signal.

[0061] In a second aspect, the present invention provides a pump noise suppression system for a continuous wave mud pulse signal while drilling, which is used to implement the pump noise suppression method for a continuous wave mud pulse signal while drilling described in the first aspect of the present invention, comprising:

[0062] A data acquisition device for acquiring original mud pulse signals including continuous wave pulse signals;

[0063] A pump noise fundamental wave determination module is used to identify the fundamental wave frequency of the pump noise in the original mud pulse signal and determine the pump noise fundamental wave according to the fundamental wave frequency;

[0064] The model building module uses the pump noise fundamental wave to build a non-steady-state pump noise model, builds a linear time-invariant state-space model of the kth harmonic based on the non-steady-state pump noise model, and builds a linear time-invariant measurement model of the kth harmonic based on the set measurement noise; the pump noise prediction module uses an adaptive Kalman filter to predict the pump noise of the constructed linear time-invariant state-space model to obtain the best a posteriori estimate of the pump noise at each moment;

[0065] A reconstruction module reconstructs the pump noise signal in the continuous wave mud pulse observation signal calculated by the linear time-invariant measurement model through the pump noise posterior estimation value to obtain a reconstructed pump noise signal containing all order harmonics;

[0066] The pump noise suppression module performs a subtraction between the collected original mud pulse signal and the reconstructed noise signal to remove the reconstructed noise signal and obtain a continuous wave mud pulse signal after suppressing the pump noise.

[0067] In some embodiments, the pump noise prediction module includes:

[0068] The state update module calculates the prior estimation value according to the state transfer matrix, calculates the residual of the measured value and the prior estimation value according to the measurement transfer matrix, calculates the adaptive factor according to the variance average of the residual, adaptively changes the process noise covariance according to the adaptive factor, adjusts the notch properties of different harmonics, and calculates the prior estimation error covariance;

[0069] A Kalman gain calculation module calculates the Kalman gain based on the prior estimation error covariance;

[0070] The measurement update module updates the a posteriori estimate and the a posteriori estimate error covariance according to the a priori estimate and the Kalman gain. Compared with the prior art, the advantages and positive effects of the present invention are:

[0071] (1) The pump noise suppression method and system of the continuous wave mud pulse signal while drilling provided by the present invention, based on the generation and output of the continuous wave mud pulse signal while drilling, takes into account the characteristics of spectrum expansion that may occur under actual conditions, identifies the fundamental frequency of the collected mud pulse signal in turn, constructs a non-steady-state noise model based on the pump noise fundamental wave determined by the fundamental frequency, constructs a state space model based on the non-steady-state noise model, uses adaptive Kalman filtering to predict the pump noise of the constructed linear time-invariant state space model to obtain the best a posteriori estimate of the pump noise at each moment, removes the pump noise signal reconstructed according to the a posteriori estimate of the pump noise, and effectively suppresses unstable pump noise signals.

[0072] (2) The pump noise suppression method and system for the continuous wave mud pulse signal while drilling provided by the present invention adopts an adaptive Kalman filter with a notch characteristic. By introducing an adaptive factor, the covariance parameter value can be adaptively adjusted according to the signal characteristics, and then the Kalman gain can be adjusted to adapt to the different notch depths and widths of harmonics of each order. The method and system have strong adaptive ability and can adapt to different working conditions without excessive adjustment of the filter parameters. The complexity of the adaptive Kalman filter is simplified, and efficient and reliable suppression of unstable pump noise signals is achieved, which has great practical value.

[0073] (2) The pump noise suppression method and system for the continuous wave mud pulse signal while drilling provided by the present invention are simple to operate and suitable for processing narrow-band widened pump noise. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] Figure 1 It is a flowchart of a method for suppressing pump noise of a continuous wave mud pulse signal while drilling according to an embodiment of the present invention;

[0075] Figure 2 This is a spectrum diagram of the original mud pulse signal according to an embodiment of the present invention;

[0076] Figure 3 The inverse spectrum diagram of the original mud pulse signal according to the embodiment of the present invention;

[0077] Figure 4 This is a flow chart of an adaptive Kalman filter according to an embodiment of the present invention;

[0078] Figure 5 It is a structural block diagram of the pump noise suppression system of the continuous wave mud pulse signal while drilling according to the embodiment of the present invention;

[0079] Figure 6 This is a diagram of an original mud pulse signal containing a continuous wave pulse signal according to an embodiment of the present invention;

[0080] Figure 7 This is a diagram of a continuous wave mud pulse signal after suppressing pump noise according to an embodiment of the present invention;

[0081] Figure 8 This is a spectrum diagram of a continuous wave mud pulse signal after suppressing pump noise according to an embodiment of the present invention;

[0082] Fig. 9 This is a diagram of a continuous wave mud pulse signal after random noise suppression of a continuous wave mud pulse signal after suppressing pump noise signal using a conventional random noise removal method.

[0083] In the figure, 1. Data acquisition device, 2. Pump noise fundamental wave determination module, 3. Model construction module, 4. Pump noise prediction module, 41. State update module, 42. Kalman gain calculation module, 43. Measurement update module, 5. Reconstruction module, 6. Pump noise suppression module, An , the original mud pressure signal containing the continuous wave mud pulse signal, C n , continuous wave mud pulse signal after suppressing pump noise, F1, original mud pulse signal spectrum, F2, original mud pulse signal inverse spectrum, F3, continuous wave mud pulse signal spectrum after suppressing pump noise, P1, pump noise fundamental wave, P2, pump noise second harmonic, P3, pump noise third harmonic, P6, pump noise sixth harmonic, P9, pump noise ninth harmonic, P 12 , the twelfth harmonic of pump noise. DETAILED DESCRIPTION

[0084] The present invention is described in detail below by way of exemplary embodiments. However, it should be understood that elements, structures, and features in one embodiment may also be beneficially combined in other embodiments without further description.

[0085] In view of the problems of poor denoising effect in the existing methods for removing pump noise interference, the present invention provides a method and system for suppressing pump noise of continuous wave mud pulse signals while drilling, taking into account the characteristics of spectrum expansion in actual situations, identifying the fundamental frequency of the collected mud pulse signal, building a non-steady-state noise model based on the fundamental wave of the pump noise, and establishing a state space model based on the non-steady-state noise model according to the characteristics of unstable pump noise and BFSK modulated continuous wave signals. Adaptive Kalman filtering with notch characteristics is used to adaptively adjust the covariance parameter value to adapt to the different notch depths and widths of each order harmonic, so as to achieve efficient and accurate suppression of pump noise signals and improve the quality and accuracy of the signal. The following is a detailed description of a method and system for suppressing pump noise of continuous wave mud pulse signals while drilling provided by the present invention in conjunction with the accompanying drawings.

[0086] See also Figure 1 According to a first aspect of the present invention, there is provided a method for suppressing pump noise of a continuous wave mud pulse signal while drilling, the steps of which are as follows:

[0087] S1. Collecting the original mud pulse signal including the continuous wave pulse signal.

[0088] Specifically, in some embodiments, the raw mud pulse signal is collected by a pressure sensor or an acceleration sensor. The sampling frequency is not less than 100 Hz to ensure accurate signal collection, especially in an environment where low-frequency noise and pulse signals are mixed. A higher sampling frequency helps to improve measurement accuracy and signal quality after filtering.

[0089] S2. Identify the fundamental frequency of the pump noise in the original mud pulse signal, and determine the fundamental wave of the pump noise according to the fundamental frequency.

[0090] Specifically, in some embodiments, the method for identifying the fundamental frequency of the pump noise in the original mud pulse signal is: performing a fast Fourier transform on the original mud pulse signal to obtain the original mud pulse signal spectrum F1 as shown in FIG. Figure 2 As shown, the fundamental frequency of the pump noise is found from the spectrum; when the low-frequency noise covers the fundamental frequency of the pump noise, the inverse spectrum F2 is obtained by performing inverse spectrum analysis on the original mud pulse signal as shown in Figure 3 As shown, the fundamental frequency is obtained by solving the inverse of the horizontal axis time at the peak in the inverse spectrum. It should be noted that the fundamental frequency is usually caused by the periodic noise of the pump system. Therefore, the prominent fundamental frequency component in the spectrum is identified as the fundamental frequency of the noise.

[0091] It should also be noted that when low-frequency noise covers the fundamental frequency of pump noise, it cannot be accurately identified through Fourier transform. In order to highlight the most obvious periodic characteristic component in the collected original mud pulse signal, low-frequency noise interference is effectively removed through inverse spectrum analysis. The obtained inverse spectrum signal can more clearly identify the fundamental frequency of pump noise. The inverse spectrum method helps to improve the accuracy of fundamental frequency identification, especially in complex noise environments.

[0092] Specifically, in some embodiments, the method for determining the fundamental wave of the pump noise according to the fundamental wave frequency is:

[0093] Calculate the fundamental wave angular frequency according to the fundamental wave frequency;

[0094] ω0=2πf0

[0095] In the formula, ω0 is the fundamental angular frequency, f0 is the fundamental frequency;

[0096] Calculate the pump noise fundamental wave according to the fundamental wave angular frequency;

[0097]

[0098] In the formula, is the fundamental wave of pump noise at time n; n is the continuous sampling point, n=1,2,...,N, N is the total number of sampling points; A m is the amplitude of the fundamental wave of pump noise; φ is the phase of the fundamental wave of pump noise.

[0099] S3. Use the pump noise fundamental wave to construct a non-steady-state pump noise model, construct a linear time-invariant state space model of the kth harmonic based on the non-steady-state pump noise model, and construct a linear time-invariant measurement model of the kth harmonic based on the set measurement noise.

[0100] Specifically, in some embodiments, the method of constructing a non-steady-state pump noise model using the pump noise fundamental wave is:

[0101] Using the trigonometric equation, the pump noise fundamental is expressed as:

[0102]

[0103] In the formula, is the fundamental wave of pump noise at time n+1; is the fundamental wave of pump noise at time n-1;

[0104] Introduce a zero-mean random error w n , the non-steady-state pump noise model is constructed using the pump noise fundamental wave and is expressed as:

[0105]

[0106] It should be noted that since the actually generated fundamental wave may have slight deviations in frequency, amplitude and phase, a zero-mean random error is introduced when constructing the non-steady-state pump noise model to represent the errors that may occur in the measurement process, simulate the uncertainty of measurement noise and signals in the real environment, and ensure that the model is more flexible and adaptable to changes in different working environments.

[0107] The constructed unsteady pump noise model adopts time-varying characteristics, which can adapt to the dynamics of pump noise changing over time during operation. The unsteady nature of the unsteady pump noise model ensures that it can be adjusted in real time to adapt to fluctuations in the external environment.

[0108] Specifically, in some embodiments, a method for constructing a linear time-invariant state-space model of the kth harmonic according to the unsteady-state pump noise model is:

[0109] Assume that the state vector of the pump noise containing k-order harmonics is Among them, each element is represented as:

[0110] The linear time-invariant state space model of the kth harmonic is constructed based on the unsteady-state pump noise model and is expressed as:

[0111] p n =Ap n-1 +Bw n

[0112] In the formula, p n-1 is the state vector of the pump noise at time n-1; A is the state transfer matrix, B is the control matrix, T is transpose.

[0113] Specifically, in some embodiments, the method for constructing a linear time-invariant measurement model of the kth harmonic according to the set measurement noise is:

[0114] Assume that the state vector of the pump noise containing k-order harmonics is Among them, each element is represented as:

[0115] The linear time-invariant measurement model of the kth harmonic is constructed according to the set measurement noise and is expressed as:

[0116] y n =H T p n +v n

[0117] In the formula, y n is the measured value of the continuous wave mud pulse signal; H is the measurement transfer matrix, T is transpose; v n To measure noise.

[0118] A linear time-invariant measurement model of the pump noise is constructed based on the set measurement noise to ensure that the noise reconstruction process accurately reflects the measurement errors and environmental changes.

[0119] S4. Use adaptive Kalman filtering to predict pump noise on the constructed linear time-invariant state space model to obtain the best a posteriori estimate of pump noise at each moment.

[0120] Specifically, in some embodiments, see Figure 4 The specific method of using adaptive Kalman filtering to predict the posterior estimate of pump noise is:

[0121] S41, state update step: calculate the prior estimate according to the state transfer matrix A

[0122]

[0123] In the formula, is the posterior estimate at time n-1;

[0124] Calculate the measurement value y according to the measurement transfer matrix H n and the prior estimate The residual

[0125]

[0126] Take L as the window size and calculate the residual The mean of the variance

[0127]

[0128] Assume T λ is the threshold of the adaptive factor, and T λ ≥0, calculate the adaptive factor λ:

[0129]

[0130] In the formula, is the posterior estimation error covariance at time n-1, q n is the process noise covariance, r n To measure the noise covariance, the adaptive factor λ is adaptively changed according to the following formula to adjust the process noise covariance q n , adjust the notch properties of different harmonics and calculate the prior estimation error covariance

[0131]

[0132] S42, Kalman gain calculation steps: according to the prior estimation error covariance Calculate the Kalman gain K n ;

[0133]

[0134] S43, measurement update step: based on the prior estimate and the Kalman gain K n Update the posterior estimate

[0135]

[0136] According to the prior estimate error covariance and the Kalman gain K n Update the posterior estimate error covariance

[0137]

[0138] By cyclically recursively performing steps S41 to S43, the best a posteriori estimation value of the pump noise at each moment can be obtained.

[0139] It should be noted that the adjustment of the adaptive factor depends on the noise intensity of the current signal. If the noise is strong, the adaptive factor is increased, the notch depth is increased, and the noise suppression effect is enhanced.

[0140] Through the adaptive Kalman filter with notch characteristics, combined with the adaptive factor, the covariance parameter value can be adaptively adjusted according to the signal characteristics, and then the Kalman gain can be adjusted to adapt to the different notch depths and widths of each order harmonic, adapt to different working conditions, and significantly improve the pump noise suppression effect. Especially in the case of high noise interference, it can better adapt to the changing noise spectrum and improve the robustness.

[0141] S5. Reconstruct the pump noise signal in the continuous wave mud pulse observation signal calculated by the linear time-invariant measurement model through the pump noise posterior estimation value to obtain a reconstructed pump noise signal containing all order harmonics.

[0142] Specifically, the reconstructed pump noise signal is expressed as:

[0143]

[0144] In the formula, B n To reconstruct the pump noise signal.

[0145] S6. The original mud pulse signal minus the reconstructed noise signal is used to obtain a continuous wave mud pulse signal after suppressing pump noise.

[0146] See also Figure 5 The second embodiment of the present invention provides a pump noise suppression system for a continuous wave mud pulse signal while drilling, which is used to implement the pump noise suppression method for a continuous wave mud pulse signal while drilling described in the first aspect of the present invention, comprising:

[0147] A data acquisition device 1 is used to collect original mud pulse signals including continuous wave pulse signals;

[0148] A pump noise fundamental wave determination module 2 is used to identify the fundamental wave frequency of the pump noise in the original mud pulse signal and determine the pump noise fundamental wave according to the fundamental wave frequency;

[0149] Model construction module 3, using the pump noise fundamental wave to construct a non-steady-state pump noise model, constructing a linear time-invariant state-space model of the kth harmonic based on the non-steady-state pump noise model, and constructing a linear time-invariant measurement model of the kth harmonic based on the set measurement noise; pump noise prediction module 4, using adaptive Kalman filtering to predict pump noise on the constructed linear time-invariant state-space model to obtain the best a posteriori estimate of pump noise at each moment;

[0150] Reconstruction module 5, reconstructs the pump noise signal in the continuous wave mud pulse observation signal calculated by the linear time-invariant measurement model through the pump noise posterior estimation value to obtain a reconstructed pump noise signal containing all order harmonics;

[0151] The pump noise suppression module 6 performs subtraction between the collected original mud pulse signal and the reconstructed noise signal to remove the reconstructed noise signal and obtain a continuous wave mud pulse signal after suppressing the pump noise.

[0152] Each module works together through data flow. The data acquisition device transmits the signal to the pump noise fundamental wave determination module; the fundamental wave information is transmitted to the model construction module to generate the state space model; the pump noise prediction module dynamically updates the model output estimate value for use by the reconstruction module; and finally the reconstructed pump noise signal is removed from the original mud pulse signal through the suppression module. The overall structure of the system is tightly coordinated, which can achieve dynamic suppression and signal enhancement of complex pump noise.

[0153] In some embodiments, the pump noise prediction module 4 includes:

[0154] A state updating module 41 calculates a priori estimation value according to a state transfer matrix, calculates a residual of a measurement value and a priori estimation value according to a measurement transfer matrix, calculates an adaptive factor according to a variance average of the residual, adaptively changes a process noise covariance according to the adaptive factor, adjusts notch properties of different harmonics, and calculates a priori estimation error covariance;

[0155] A Kalman gain calculation module 42 is used to calculate the Kalman gain according to the prior estimation error covariance;

[0156] The measurement update module 43 updates the a posteriori estimate and the a posteriori estimate error covariance according to the prior estimate and the Kalman gain. The pump noise prediction module uses an adaptive Kalman filter to dynamically estimate the constructed linear time-invariant state space model and achieves the best filtering effect by adjusting the noise covariance. Specifically, in some embodiments, the pump noise prediction module is an adaptive Kalman filter.

[0157] Specifically, in some embodiments, the data acquisition device is a pressure sensor or an acceleration sensor. The sampling frequency of the above sensor is not less than 100 Hz to ensure accurate acquisition of the signal, especially in an environment where low-frequency noise and pulse signals are mixed. A higher sampling frequency helps to improve the measurement accuracy and the quality of the filtered signal.

[0158] The effectiveness of the pump noise suppression method and system for the continuous wave mud pulse signal while drilling will be described below in conjunction with specific embodiments.

[0159] Example: Figure 6 A section of the original mud pressure signal A containing a continuous wave pulse signal collected by the pressure sensor at the surface riser n (i.e. the original mud pulse signal) as an example. The original mud pressure signal A n The data frequency is 1024Hz, the bit rate is 4bps, and the useful pulse signal is a continuous phase BFSK modulation signal with main frequencies of 8Hz and 12Hz.

[0160] The pump noise suppression method and system of the continuous wave mud pulse signal while drilling according to the embodiment of the present invention are used to suppress the original mud pressure signal A.n Perform pump noise suppression to obtain a continuous wave pulse signal C after suppressing pump noise n . After suppressing the pump noise, the continuous wave pulse signal C n Perform fast Fourier transform to obtain the continuous wave pulse signal spectrum F3 after suppressing pump noise. Figure 7 As shown. Figure 7 It can be seen that the pump noise is effectively suppressed, and the continuous wave mud pulse signal characteristics with 8 Hz and 12 Hz as the main frequencies are clearly presented in the spectrum. It can be considered that the purpose of pump noise removal has been achieved.

[0161] The continuous wave mud pulse signal C after suppressing pump noise n Conventional random noise removal methods are used to further suppress random noise, such as digital filters, wavelet transforms, etc. According to the characteristics of BFSK modulated continuous wave signals, a bandpass filter containing two carrier frequencies is selected to suppress random noise and extract the continuous wave mud pulse signal D n like Figure 8 As shown, the BFSK continuous wave signal waveform can be clearly seen.

[0162] The above embodiments are used to explain the present invention rather than to limit the present invention. Any modification and change made to the present invention within the spirit of the present invention and the protection scope of the claims shall fall within the protection scope of the present invention.

Claims

1. A method for suppressing pump noise of continuous wave mud pulse signal while drilling, characterized in that: The steps are: S1, collecting the original mud pulse signal including the continuous wave pulse signal; S2, identifying the fundamental frequency of the pump noise in the original mud pulse signal, and determining the fundamental wave of the pump noise according to the fundamental frequency; S3, constructing a non-steady-state pump noise model using the pump noise fundamental wave, constructing a linear time-invariant state space model of the kth harmonic based on the non-steady-state pump noise model, and constructing a linear time-invariant measurement model of the kth harmonic based on the set measurement noise; S4, using adaptive Kalman filtering to predict pump noise on the constructed linear time-invariant state space model to obtain the best a posteriori estimate of pump noise at each moment; S5, reconstructing the pump noise signal in the continuous wave mud pulse observation signal calculated by the linear time-invariant measurement model through the pump noise posterior estimation value to obtain a reconstructed pump noise signal containing all order harmonics; S6. The original mud pulse signal minus the reconstructed noise signal is used to obtain a continuous wave mud pulse signal after suppressing pump noise.

2. The method for suppressing pump noise of continuous wave mud pulse signal while drilling according to claim 1, characterized in that: In step S2, the method for identifying the fundamental frequency of the pump noise in the original mud pulse signal is as follows: performing fast Fourier transform on the original mud pulse signal to obtain the spectrum of the original mud pulse signal, and finding the fundamental frequency of the pump noise from the spectrum; when the low-frequency noise covers the fundamental frequency of the pump noise, performing inverse spectrum analysis on the original mud pulse signal to obtain the inverse spectrum, and obtaining the fundamental frequency by solving the inverse of the horizontal coordinate time at the peak value in the inverse spectrum.

3. The pump noise suppression method of continuous wave mud pulse signal while drilling according to claim 2, characterized in that: In step S2, the method for determining the pump noise fundamental wave according to the fundamental wave frequency is: Calculate the fundamental wave angular frequency according to the fundamental wave frequency; ω0=2πf0 In the formula, ω0 is the fundamental angular frequency, f0 is the fundamental frequency; Calculate the pump noise fundamental wave according to the fundamental wave angular frequency; In the formula, is the fundamental wave of pump noise at time n; n is the continuous sampling point, n=1,2,...,N, N is the total number of sampling points; A m is the amplitude of the fundamental wave of pump noise; φ is the phase of the fundamental wave of pump noise.

4. The method for suppressing pump noise of continuous wave mud pulse signal while drilling according to claim 3, characterized in that: In step S3, the method of constructing the non-steady-state pump noise model using the pump noise fundamental wave is: Using the trigonometric equation, the pump noise fundamental is expressed as: In the formula, is the fundamental wave of pump noise at time n+1; is the fundamental wave of pump noise at time n-1; Introduce a zero-mean random error w n , the non-steady-state pump noise model is constructed using the pump noise fundamental wave and is expressed as:

5. The method for suppressing pump noise of continuous wave mud pulse signal while drilling according to claim 4, characterized in that: In step S3, the method of constructing a linear time-invariant state-space model of the kth harmonic according to the unsteady-state pump noise model is: Assume that the state vector of the pump noise containing k-order harmonics is Among them, each element is represented as: The linear time-invariant state space model of the kth harmonic is constructed based on the unsteady-state pump noise model and is expressed as: p n =Ap n-1 +Bw n In the formula, p n-1 is the state vector of the pump noise at time n-1; A is the state transfer matrix, B is the control matrix, T is transpose.

6. The method for suppressing pump noise of continuous wave mud pulse signal while drilling according to claim 5, characterized in that: In step S3, a linear time-invariant measurement model of the kth harmonic is constructed according to the set measurement noise and is expressed as: y n =H T p n +v n In the formula, y n is the measured value of the continuous wave mud pulse signal; H is the measurement transfer matrix, T is transpose; v n To measure noise.

7. The method for suppressing pump noise of continuous wave mud pulse signal while drilling according to claim 6, characterized in that: In step S4, the specific method of using adaptive Kalman filtering to predict the posterior estimate of pump noise is: State update step: Calculate the prior estimate based on the state transfer matrix A In the formula, is the posterior estimate at time n-1; Calculate the measurement value y according to the measurement transfer matrix H n and the prior estimate The residual Take L as the window size and calculate the residual The mean of the variance Assume T λ is the threshold of the adaptive factor, and T λ ≥0, calculate the adaptive factor λ: In the formula, is the posterior estimation error covariance at time n-1, q n is the process noise covariance, r n To measure the noise covariance, the adaptive factor λ is adaptively changed according to the following formula to adjust the process noise covariance q n , adjust the notch properties of different harmonics and calculate the prior estimation error covariance Kalman gain calculation steps: Estimation of error covariance based on prior Calculate the Kalman gain K n ; Measurement update step: Based on the prior estimate and the Kalman gain K n Update the posterior estimate According to the prior estimate error covariance and the Kalman gain K n Update the posterior estimate error covariance By cyclically recursively repeating the state update step to the measurement update step, the best a posteriori estimate of the pump noise at each moment can be obtained.

8. The method for suppressing pump noise of continuous wave mud pulse signal while drilling according to claim 7, characterized in that: In step S5, the reconstructed pump noise signal is expressed as: In the formula, B n To reconstruct the pump noise signal.

9. A pump noise suppression system for a continuous wave mud pulse signal while drilling, used to implement the pump noise suppression method for a continuous wave mud pulse signal while drilling as claimed in any one of claims 1 to 8, characterized in that: include: A data acquisition device for acquiring original mud pulse signals including continuous wave pulse signals; A pump noise fundamental wave determination module is used to identify the fundamental wave frequency of the pump noise in the original mud pulse signal and determine the pump noise fundamental wave according to the fundamental wave frequency; The model building module uses the pump noise fundamental wave to build a non-steady-state pump noise model, builds a linear time-invariant state-space model of the kth harmonic based on the non-steady-state pump noise model, and builds a linear time-invariant measurement model of the kth harmonic based on the set measurement noise; the pump noise prediction module uses an adaptive Kalman filter to predict the pump noise of the constructed linear time-invariant state-space model to obtain the best a posteriori estimate of the pump noise at each moment; A reconstruction module reconstructs the pump noise signal in the continuous wave mud pulse observation signal calculated by the linear time-invariant measurement model through the pump noise posterior estimation value to obtain a reconstructed pump noise signal containing all order harmonics; The pump noise suppression module performs a subtraction between the collected original mud pulse signal and the reconstructed noise signal to remove the reconstructed noise signal and obtain a continuous wave mud pulse signal after suppressing the pump noise.

10. The pump noise suppression system for continuous wave mud pulse signal while drilling according to claim 9, characterized in that: The pump noise prediction module comprises: The state update module calculates the prior estimation value according to the state transfer matrix, calculates the residual of the measured value and the prior estimation value according to the measurement transfer matrix, calculates the adaptive factor according to the variance average of the residual, adaptively changes the process noise covariance according to the adaptive factor, adjusts the notch properties of different harmonics, and calculates the prior estimation error covariance; A Kalman gain calculation module calculates the Kalman gain based on the prior estimation error covariance; The measurement update module updates the posterior estimate and the posterior estimate error covariance according to the prior estimate and the Kalman gain.

Images