Adaptive Threshold Constraint and Trend Fitting Based Micro-Logging Q Estimation Noise Suppression Method

Abstract

The present invention relates to the field of seismic data processing for oil and gas exploration, and particularly to a micro-log Q estimation noise suppression method based on adaptive threshold constraint and trend fitting. The method includes the following steps: estimating the Q value according to the micro-log data; performing a depth-variable Q value trend estimation; determining an adaptive depth-variable Q threshold; deleting the Q values outside the threshold range; performing polynomial fitting on the Q values; and replacing the deleted Q values with the fitting values. The method of the present invention can suppress the instability of Q estimation caused by data noise, improve the accuracy of micro-log Q estimation, and is beneficial to near-surface absorption attenuation investigation and Q modeling.

Description

Technical Field

[0001] The present invention relates to the field of seismic data processing for oil and gas exploration, and particularly to a micro-logging Q estimation noise suppression method based on adaptive threshold constraint and trend fitting. Background Art

[0002] Estimating the near-surface Q using micro-logging data and eliminating the influence of near-surface absorption attenuation is an important attempt in high-resolution seismic exploration in recent years.

[0003] Currently, the mainstream methods for micro-logging Q estimation are methods such as spectral ratio and centroid frequency shift. These methods all use the change in the spectrum of the first arrival wavelets excited by shot points at different depths to estimate the micro-logging Q and have been widely used. For example, the applicant's prior application CN110261904A discloses a method for inverting and classifying and evaluating the near-surface Q value based on the generalized S transform, including: Step 1, sorting adjacent two-channel data; Step 2, respectively applying the generalized S transform to obtain high-resolution time-frequency spectra; Step 3, extracting the instantaneous amplitude spectra of the corresponding first arrival signals; Step 4, inverting and fitting the Q value between two channels using the logarithmic spectral ratio method; Step 5, repeating steps (1)-(4) to invert the Q value between adjacent channels in all micro-logging data; Step 6, generating a curve of the Q value changing with depth at each well position; Step 7, calculating the average attenuation effect factor in combination with each layer thickness and layer velocity; Step 8, performing near-surface classification and evaluation based on the average attenuation effect factor. Chinese patent application CN102109617A discloses a method for measuring the near-surface formation Q value using the double-well micro-logging technology. This method uses double-well micro-logging data, selects the bottom and surface seismic records of the shot point A located in the high-velocity layer and the shot point B at the bottom interface of the low-velocity layer, intercepts the first arrival part of the record, performs Fourier transform, and obtains the peak frequencies of each record; then uses the peak frequency shift method to estimate the Q value formula, establishes a system of equations for the change in the peak frequency of the seismic signal and solves it to obtain Q1 and Q2; finally, for the Q1 and Q2 values obtained at all double-well micro-logging points in the work area, perform horizontal interpolation and extrapolation according to the layered results interpreted by single-well micro-logging to obtain the near-surface Q1 and Q2 values of the whole area. This method solves the problem that the original double-well micro-logging technology can only be used to measure the velocity and thickness of each near-surface layer;

[0004] With continuous in-depth research, the inventor found that: when the data contains no noise, these methods have good effects. However, actual micro-logging data contains noise, which will affect the spectrum of the first arrival wavelet, and further affect the micro-logging Q estimation. Especially as the depth of the excitation well increases, the received first arrival signal becomes weaker, the signal-to-noise ratio decreases, and the Q estimation error increases, greatly affecting the accuracy and reliability of the results and unable to meet the requirements of high-resolution seismic exploration. And there is currently no solution to this problem. Summary of the Invention

[0005] The main object of the present invention is to provide a micro-logging Q estimation noise suppression method based on adaptive threshold constraint and trend fitting. This method can suppress the instability of Q estimation caused by data noise, improve the accuracy of micro-logging Q estimation, and is beneficial to near-surface absorption attenuation investigation and Q modeling.

[0006] To achieve the above object, the present invention adopts the following technical solutions:

[0007] The present invention provides a micro-logging Q estimation noise suppression method based on adaptive threshold constraint and trend fitting, which includes the following steps: estimating the Q value according to the micro-logging data; performing a depth-varying Q value trend estimation; determining an adaptive depth-varying Q threshold; deleting the Q values outside the threshold range; performing a polynomial fitting on the Q values; and replacing the deleted Q values with the fitting values.

[0008] Further, the Q value of the micro-logging data is estimated by using the spectral ratio method, which satisfies the following formula:

[0009]

[0010] where Q evt (i) is the formation Q value between the i-th shot and the (i + 1)-th shot, f is the frequency, Δt i,i+1 is the first arrival time difference between the i-th shot and the (i + 1)-th shot, W i (f) is the first arrival wavelet amplitude spectrum of the i-th shot, and ln represents taking the logarithm.

[0011] Further, the Alpha-Trim mean filter is used to perform a depth-varying Q value trend estimation on the Q estimation value Q evt . The Alpha-Trim mean filter takes the Q value at the i-th point as the center and has a window length of w to perform a time window processing on the Q result. The Q values within the window are sorted from small to large, and α proportion of the maximum and minimum values of the total number of points within the window are discarded, and the average value of the remaining data within the window is calculated as the filtering output at the center position i of the window. The window is slid point by point along the depth direction, and the calculated result is used as the depth-varying trend estimation value Q trend of the Q value.

[0012] Further, the method for determining the adaptive depth-varying Q threshold: for the Q estimation value Q evt , using the time window with a length of w in the Alpha-Trim mean filter, after the Q values within the time window are sorted from small to large, α proportion of the maximum and minimum values of the total number of points within the time window are discarded, and the mean square error after discarding the outliers is calculated for the remaining data within the window:

[0013]

[0014] where n is the number of remaining data after discarding α proportion of the maximum and minimum values in the time window corresponding to the i-th point, Q trend(i) The Q trend value of the i-th point calculated by Alpha-Trim mean filtering; and given a threshold control parameter p, calculate the adaptive threshold range of the Q value varying with depth [Q min (i), Q min (i)], where Q min (i) and Q max (i) are calculated by the following formulas respectively,

[0015] Q min (i) = Q trend (i) - p * Q σ (i),

[0016] Q max (i) = Q trend (i) + p * Q σ (i).

[0017] Furthermore, polynomial fitting of the Q value is preferably carried out by polynomial fitting of the Q value.

[0018] Compared with the prior art, the present invention has the following advantages:

[0019] The method of the present invention can effectively suppress the instability of Q estimation caused by data noise, improve the accuracy of micro-logging Q estimation, and is beneficial to near-surface absorption attenuation investigation and Q modeling. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The accompanying drawings forming a part of the present invention are used to provide a further understanding of the present invention. The schematic embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention.

[0021] Figure 1 It is a schematic flow chart of the micro-logging Q estimation noise suppression method based on adaptive threshold constraint and trend fitting described in Embodiment 1 of the present invention;

[0022] Figure 2 It is the forward modeling result of the first arrival wave of the micro-logging data described in Embodiment 1;

[0023] Figure 3 It is the trace equalization display of the forward modeling result of the first arrival wave of the micro-logging data described in Embodiment 1;

[0024] Figure 4 It is the logarithmic spectrum ratio of the first arrival waves of two adjacent traces of the micro-logging data described in Embodiment 1 and its regression fitting curve graph;

[0025] Figure 5 It is the Q estimation result and its estimation error graph of the micro-logging data under the condition of no noise described in Embodiment 1;

[0026] Figure 6Forward result display of the trace equalization for the noisy micro-logging data described in Example 1;

[0027] Figure 7 Logarithmic spectrum ratio of the first arrival waves of two adjacent traces and its regression fitting curve graph for the noisy micro-logging data described in Example 1;

[0028] Figure 8 Q estimation result and its estimation error graph for the noisy micro-logging data described in Example 1;

[0029] Figure 9 Adaptive threshold determination based on Alpha-Trim mean filtering and abnormal Q value rejection result graph under noisy conditions for the micro-logging data described in Example 1;

[0030] Figure 10 Trend fitting result graph after rejecting abnormal Q values under noisy conditions for the micro-logging data described in Example 1;

[0031] Figure 11 Comparison graph of the Q estimated value, the Q value after suppression and correction, and the theoretical value under noisy conditions for the micro-logging data described in Example 1;

[0032] Figure 12 Error comparison graph of the Q estimated value and the Q value after suppression and correction under noisy conditions for the micro-logging data described in Example 1. Detailed implementation method

[0033] It should be noted that the following detailed description is exemplary and is intended to provide further illustration of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which the present invention belongs.

[0034] It should be noted that the terms used herein are only for describing specific implementation manners and are not intended to limit the exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular forms are also intended to include the plural forms. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, and / or combinations thereof.

[0035] In order to enable those skilled in the art to more clearly understand the technical solution of the present invention, the technical solution of the present invention will be described in detail below in conjunction with specific embodiments.

[0036] Example 1

[0037] As Figure 1 shown, the micro-logging Q estimation noise suppression method based on adaptive threshold constraint and trend fitting includes the following steps:

[0038] Step 1: Estimate the Q value based on micro-log data;

[0039] Select the spectral ratio method to estimate the micro-log Q value, which satisfies the following formula:

[0040]

[0041] where Q evt (i) is the formation Q value between the i-th shot and the (i + 1)-th shot, f is the frequency, and Δt i,i+1 is the first arrival time difference between the i-th shot and the (i + 1)-th shot, and W i (f) is the amplitude spectrum of the first arrival wavelet of the i-th shot, and ln represents taking the logarithm.

[0042] Step 2: Trend estimation of depth-variable Q value based on Alpha-Trim mean filtering;

[0043] On the basis of completing Step 1, select an appropriate window length w of Alpha-Trim mean filtering and a control parameter α for the rejection ratio of maximum and minimum values, and perform Alpha-Trim mean filtering on the estimated Q value. Specifically, sort the Q value data within the window with length w centered at point i from small to large, discard the points with the maximum and minimum values accounting for α proportion of the total number of points in the window, calculate the average value of the remaining Q values in the window, and use it as the estimated depth-variable Q trend value at point i. Slide the window along the depth to obtain the estimated Q depth-variable trend value Q trend .

[0044] Step 3: Adaptive method for determining the depth-variable Q threshold;

[0045] On the basis of completing Step 2, use the window length parameter w and the control parameter α for the rejection ratio of maximum and minimum values in Step 2 to sort the Q value data within the window with length w centered at point i from small to large, discard the points with the maximum and minimum values accounting for α proportion of the total number of points in the window, and calculate the mean square deviation after discarding outliers for the remaining points in the window using the following formula:

[0046]

[0047] where n is the number of remaining data after discarding the maximum and minimum values accounting for α proportion of the total data in the time window corresponding to point i, and Q trend (i) is the Q trend value at point i calculated by Alpha-Trim mean filtering. And given a threshold control parameter p, calculate the adaptive threshold range [Q min (i), Q min (i)], where Q min (i) and Q max (i) are calculated by the following formulas (3) and (4) respectively,

[0048] Q min (i) = Q trend (i) - p * Q σ (i),

[0049] Q max (i) = Q trend (i) + p * Q σ (i).

[0050] Step 4: Delete Q values outside the threshold range;

[0051] Step 5: Perform polynomial fitting on the Q values;

[0052] On the basis of completing Step 4, further perform polynomial fitting on the micro-logging Q values. Preferably, the fitting adopts the method of piecewise Q value fitting.

[0053] Step 6: Replace the deleted Q values with the fitting values.

[0054] Thus, the micro-logging Q values after suppression and correction can be obtained.

[0055] The suppression method provided by the present invention will be specifically described below in combination with typical data. As Figures 2 - 12 shown, among them, Figure 2 is the forward modeling result of the first arrival wave of a micro-logging data. As the trace number increases and the shot hole depth increases, the amplitude of the first arrival wave received on the ground becomes weaker; Figure 3 is the trace equalization display of the forward modeling result of the first arrival wave of the micro-logging data; Figure 4 is the logarithmic spectrum ratio of the first arrival waves of two adjacent traces of the micro-logging data and its regression fitting curve graph. The spectrum ratio method uses this logarithmic spectrum ratio relationship to estimate the micro-logging Q value. This relationship is theoretically linear, and the logarithmic spectrum ratio without noise and the fitted linear regression curve fit well; Figure 5 is the Q estimation result of the micro-logging data and its estimation error graph under the condition of no noise. The theoretical value and the estimated value basically have the same variation trend with depth. From the estimation error graph, the error between the estimated value and the theoretical value is very small and close to zero; Figure 6 is the trace equalization display of the forward modeling result of the micro-logging data with noise. The effective signal of the first arrival wave is affected by noise. The larger the trace number and the deeper the shot hole depth, the more serious the influence; Figure 7 is the logarithmic spectrum ratio of the first arrival waves of two adjacent traces of the micro-logging data with noise and its regression fitting curve graph. This logarithmic spectrum ratio relationship is affected by noise and seriously deviates from the linear relationship, showing a certain difference from the fitted linear regression curve; Figure 8 is the Q estimation result of the micro-logging data with noise and its estimation error graph. Affected by noise, the Q estimation result has an oscillating error, and the deeper the shot depth, the more serious the influence of the noise; Figure 9Adaptive threshold determination and abnormal Q-value rejection results graph based on Alpha-Trim mean filtering in the presence of noise. The Q trend value estimated based on Alpha-Trim mean filtering is less affected by outliers, can reflect the trend of Q value changing with depth, and the calculated adaptive threshold can reject most of the outliers.

[0056] Figure 10 Trend fitting result graph after rejecting abnormal Q-values in the presence of noise. The fitting values from well depth 0 - 30 meters shown are basically completely consistent with the estimated Q values. The result fitted using the Q estimated values after rejecting outliers from well depth 30 - 50 meters can better reflect the trend of Q value changing with depth and reduce the oscillation between adjacent values. Figure 11 Comparison graph of Q estimated values, Q values after suppression correction, and theoretical values in the presence of noise. It shows that from well depth 30 - 50 meters, the Q values after suppression correction significantly reduce the oscillation error of the Q value estimation results caused by noise, and they are in good agreement with the theoretical values, improving the accuracy of the Q values estimated from the noisy environment. Figure 12 Error comparison graph of Q estimated values and Q values after suppression correction in the presence of noise. This graph further shows that the error of the corrected estimated values is significantly lower than the original Q estimated values, indicating that the oscillation error of Q estimation caused by noise is suppressed.

[0057] The above results show that: the micro-log Q estimation noise suppression method based on threshold constraint and trend fitting provided by the present invention can suppress the instability of Q estimation caused by data noise and improve the micro-log Q estimation accuracy.

[0058] The above embodiments are preferred embodiments of the present invention, but the embodiments of the present invention are not limited to the above embodiments. Any other changes, modifications, substitutions, combinations, and simplifications made without departing from the spirit and principle of the present invention shall be equivalent replacement methods and are all included in the protection scope of the present invention.

Claims

1. Micro-logging Q estimation noise suppression method based on adaptive threshold constraint and trend fitting, Characterized in that, It includes the following steps: Estimate the Q value according to the micro-logging data; Perform deep-variable Q value trend estimation; determine the adaptive deep-variable Q threshold; Delete Q values outside the threshold range; Perform polynomial fitting on the Q values; use the fitting values to replace the deleted Q values; Use the spectral ratio method to estimate the Q value of the micro-logging data, satisfying the following formula: Among them, Q evt (i) is the Q value of the formation between the i-th shot and the (i + 1)-th shot, f is the frequency, and Δt i,i+1 is the first arrival time difference between the i-th shot and the (i + 1)-th shot, W i (f) is the first arrival wavelet amplitude spectrum of the i-th shot, and ln represents taking the logarithm; Use Alpha-Trim mean filtering for the Q estimate value Q evt Perform deep-variant Q value trend estimation. Alpha-Trim mean filtering takes the Q value at the i-th point as the center and has a window length of w to perform a time window processing on the Q result. The Q values within the window are sorted from small to large, and α proportion of the maximum and minimum values of the total number of points within the window are discarded, and the average value of the remaining data within the window is calculated as the filtering output at the center position i of the window. Slide the window point by point along the depth direction, and the calculated result is used as the deep-variant trend estimate value Q of the Q value trend ; Adaptive Deep-Variable Q Threshold Determination Method: For the Q estimated value Q evt , using a time window of length w in Alpha-Trim mean filtering, after sorting the Q values in the time window from small to large, discard the maximum and minimum values that account for α proportion of the total number of points in the time window, and then calculate the mean square error after discarding the outliers for the remaining data in the window: where n is the number of remaining data after discarding the maximum and minimum values of α proportion in the time window corresponding to the i-th point, and Q trend (i) is the Q trend value of the i-th point calculated by Alpha-Trim mean filtering; and a threshold control parameter p is given to calculate the adaptive threshold range of the Q value varying with depth [Q min (i), Q min (i)], where Q min (i) and Q max (i) are calculated by the following formulas respectively, Q min (i) = Q trend (i) - p * Q σ (i), Q max (i) = Q trend (i) + p * Q σ (i).

2. The method according to claim 1, Characterized in that, The fitting adopts the method of piecewise Q value fitting.

Images