A method for intelligent processing of measurement-while-drilling azimuthal resistivity

Abstract

The present application relates to the technical field of data processing, and more particularly to a method for intelligent processing of measurement-while-drilling azimuthal resistivity, which obtains a target resistivity sequence of each preset azimuth at a current depth in the process of drilling underground by measurement-while-drilling, and for any target resistivity sequence, screens initial suspected noise points through neighborhood distribution characteristics, removes noise points corresponding to real geological mutation interference from the initial suspected noise points to obtain target suspected noise points, further distinguishes real noise points from formation fracture signals through the discrete distribution characteristics of the clustering of the target suspected noise points, quantifies the noise existence degree in any target resistivity sequence, and dynamically adjusts the process noise covariance of Kalman filtering according to the noise existence degree, so that the filtering is enhanced when the noise risk is high and the filtering is weakened when the noise risk is low, thereby significantly improving the reliability of the measurement-while-drilling resistivity data and providing more accurate basis for trajectory adjustment of geosteering drilling and reservoir evaluation.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to an intelligent processing method for azimuth resistivity measured while drilling. Background Technology

[0002] Measurement-while-drilling (MWD) resistivity is the real-time acquisition of formation resistivity using downhole instruments during the drilling process. Its purpose is to promptly identify formation lithology, fluid properties, and reservoir boundary changes, providing a basis for wellbore adjustment and reservoir evaluation. MWD provides immediate feedback on formation information during drilling, which is of great significance for geosteering drilling. Due to various engineering environmental factors during drilling, such as drill string vibration and drill bit impact, the azimuth resistivity data obtained from MWD is prone to noise. For dynamic MWD scenarios, existing technologies typically employ Kalman filtering algorithms to remove data noise from the real-time acquired formation resistivity, thereby providing high-quality basic data for subsequent identification of formation lithology, fluid properties, and reservoir boundary changes.

[0003] However, due to the complex geological environment, directly applying Kalman filtering may identify real geological fluctuations (formation resistivity) during drilling as noise data, causing the filter to filter out the non-noise real fluctuation data. This leads to a decrease in the accuracy of Kalman filtering and, consequently, a reduction in the accuracy of formation resistivity. Therefore, improving the accuracy of Kalman filtering algorithms in the context of measurement-while-drilling resistivity is an urgent problem to be solved. Summary of the Invention

[0004] In view of this, embodiments of the present invention provide an intelligent processing method for azimuth resistivity measured while drilling, in order to solve the problem of how to improve the accuracy of the Kalman filter algorithm in the scenario of measuring resistivity while drilling.

[0005] This invention provides an intelligent processing method for azimuth resistivity measured while drilling, which includes the following steps: The target resistivity sequence at the current depth for each preset orientation during the underground drilling process was obtained by measurement while drilling. For any target resistivity sequence, based on the local neighborhood distribution characteristics, noise analysis is performed on each resistivity in the target resistivity sequence to obtain the initial suspected noise level of each resistivity, so as to screen out the initial suspected noise in the target resistivity sequence; stratigraphic variation interference analysis is performed on each of the initial suspected noise, and target suspected noise is screened out from all the initial suspected noise; One-dimensional clustering is performed on all suspected target noise points to obtain multiple clusters. The density of population distribution within each cluster is analyzed to obtain the credibility of noise points within each cluster, so as to identify credible noise points among all suspected target noise points. The distribution coverage and distribution quantity of all credible noise points are analyzed. Combined with the credibility of noise points within each cluster and the difference in resistivity between adjacent nodes, the degree of noise presence in any target resistivity sequence is obtained. When performing Kalman filtering on each target resistivity sequence, the noise covariance parameter is adaptively adjusted based on the noise level of each target resistivity sequence to obtain the corresponding optimized resistivity sequence.

[0006] Preferably, the step of performing noise analysis on each resistivity in any target resistivity sequence based on local neighborhood distribution characteristics to obtain the initial suspected noise level for each resistivity includes: For any resistivity in any target resistivity sequence, calculate the absolute value of the difference between any resistivity and its left and right adjacent resistivities, obtain the cumulative value of the absolute value of the difference, and normalize the cumulative value of the absolute value of the difference to obtain the resistivity change amplitude. Based on the slope of change between adjacent resistivities in any target resistivity sequence, determine the maximum number of resistivities that are continuous and have the same trend to the left and right of any resistivity, respectively denoted as the number of resistivities with the same trend on the left and the number of resistivities with the same trend on the right; calculate the cumulative value of the number of resistivities with the same trend on the left and the number of resistivities with the same trend on the right, and normalize the reciprocal of the cumulative value to obtain the extension distance of the same trend on both sides. The sum of the resistivity change magnitude and the distance of the same extension on both sides is used as the initial suspected noise level for any resistivity.

[0007] Preferably, the initial suspected noise points in any of the target resistivity sequences are obtained by screening, including: If the initial suspected noise level of any resistivity in any target resistivity sequence is greater than or equal to a preset initial suspected noise level threshold, then the any resistivity is confirmed as the initial suspected noise in the target resistivity sequence.

[0008] Preferably, the step of performing stratigraphic variation interference analysis on each of the initial suspected noise points and screening out target suspected noise points from all the initial suspected noise points includes: For any initial suspected noise point, the stratigraphic change node characteristic degree of the initial suspected noise point is obtained based on the resistivity change in the left and right neighborhoods of the initial suspected noise point in the target resistivity sequence. If the stratigraphic change node characteristic degree of the initial suspected noise point is less than a preset stratigraphic change node characteristic degree threshold, then the initial suspected noise point is determined to be a target suspected noise point.

[0009] Preferably, the step of obtaining the stratigraphic change node characteristic degree of any initial suspected noise point based on the resistivity changes in the left and right neighborhoods of any initial suspected noise point in any target resistivity sequence includes: In any target resistivity sequence, calculate the absolute value of the difference between any initial suspected noise point and each resistivity in its left neighborhood, and obtain the mean of the absolute values ​​of the differences, which is denoted as the left-side difference mean of any initial suspected noise point; obtain the right-side difference mean of any initial suspected noise point, take the minimum value between the left-side difference mean and the right-side difference mean, and normalize the reciprocal of the minimum value to obtain the one-sided neighborhood data similarity. Calculate the mean resistivity of all resistivity in the left neighborhood of any initial suspected noise point, and denot it as the mean resistivity of the left neighborhood of any initial suspected noise point. Obtain the mean resistivity of the right neighborhood of any initial suspected noise point. Normalize the absolute value of the difference between the mean resistivity of the left neighborhood and the mean resistivity of the right neighborhood to obtain the overall neighborhood distribution difference. The sum of the similarity of the unilateral neighborhood data and the difference in the overall neighborhood distribution is calculated as the stratigraphic change node characteristic degree of any initial suspected noise point.

[0010] Preferably, the analysis of the intra-cluster population density of each cluster to obtain the intra-cluster noise confidence level of each cluster includes: For any cluster, if the number of suspected target noise points in any cluster is greater than 1, then calculate the sampling time interval between each suspected target noise point in any cluster and its left and right adjacent suspected target noise points respectively, and obtain the cumulative value of the sampling time interval, which is recorded as the two-way interval distance of the corresponding suspected target noise point. Based on the lateral spacing distance between each suspected noise point in any cluster, the mean and variance of the lateral spacing distances are calculated. The mean of the lateral spacing distances is normalized to obtain a first normalized value, and the variance of the lateral spacing distances is normalized to obtain a second normalized value. The sum of the first normalized value and the second normalized value is used as the intra-cluster noise confidence of any cluster.

[0011] Preferably, the analysis of the distribution coverage and number of all credible noise points, combined with the intra-cluster noise credibility and adjacent resistivity differences corresponding to each credible noise point, to obtain the noise presence degree in any target resistivity sequence, includes: Based on the sampling time corresponding to each credible noise point, the interval between the maximum and minimum sampling times is calculated, and the interval is normalized to obtain the noise distribution coverage; the number of credible noise points is counted, and the number is normalized to obtain the noise percentage. In any target resistivity sequence, the absolute value of the resistivity difference between each credible noise point and its left adjacent resistivity is obtained. The intra-cluster noise confidence level corresponding to each credible noise point is used as a weight. The absolute values ​​of the resistivity differences of all credible noise points are summed in a weighted manner to obtain a weighted sum result. The weighted sum result is then normalized to obtain the overall noise difference level. The product of the noise percentage and the overall noise difference is obtained, and the sum of the product and the noise distribution coverage is taken as the noise presence degree in any target resistivity sequence.

[0012] Preferably, the step of adaptively adjusting the process noise covariance parameter based on the noise level of each target resistivity sequence to obtain the corresponding optimized resistivity sequence includes: For any target resistivity sequence, the difference between constant 2 and the noise level of the target resistivity sequence is calculated as an adaptive adjustment coefficient. The product of the adaptive adjustment coefficient and the basic process noise covariance parameter in the Kalman filter algorithm is calculated as the adaptive process noise covariance parameter. Based on the adaptive process noise covariance parameter, Kalman filtering is performed on the target resistivity sequence to obtain the corresponding optimized resistivity sequence.

[0013] The beneficial effects of the embodiments of the present invention compared with the prior art are as follows: Compared to traditional methods that directly use Kalman filtering to process azimuth resistivity data while drilling, the core advantage of this invention lies in its adaptive optimization of the filtering parameters (process noise covariance parameter Q). Specifically, it first filters initial suspected noise points based on neighborhood distribution characteristics. Then, combining the characteristics of formation change points—"high similarity in one-sided neighborhoods and large differences in two-sided neighborhoods"—noise points corresponding to real geological abrupt changes are removed from the initial suspected noise points to obtain target suspected noise points, avoiding misjudging formation boundaries and fracture zones as noise. Furthermore, it distinguishes real noise points from formation fracture signals by the discrete distribution characteristics of clusters of target suspected noise points, accurately quantifying the noise presence in any target resistivity sequence, and dynamically adjusting the process noise covariance of the Kalman filter accordingly. Filtering is enhanced when the noise risk is high and weakened when the risk is low. This effectively filters out random noise caused by drilling vibrations while preserving real geological fluctuation information to the greatest extent, significantly improving the reliability of the resistivity data while drilling, and providing a more accurate basis for trajectory adjustment and reservoir evaluation in geological-guided drilling. Attached Figure Description

[0014] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0015] Figure 1 This is a flowchart of a method for intelligent processing of azimuth resistivity measured while drilling, provided in Embodiment 1 of the present invention. Detailed Implementation

[0016] Embodiments of this disclosure are described in detail below, with examples of these embodiments illustrated in the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this disclosure, and should not be construed as limiting it.

[0017] It should be noted that the terms "first," "second," etc., used in this disclosure and the accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this disclosure described herein can be implemented in orders other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this disclosure. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this disclosure.

[0018] To illustrate the technical solution of the present invention, specific embodiments are described below.

[0019] See Figure 1 This is a flowchart of a method for intelligent processing of azimuth resistivity measured while drilling, provided in Embodiment 1 of the present invention. Figure 1 As shown, the method may include: Step S101: Obtain the target resistivity sequence at the current depth for each preset orientation during the underground drilling process through measurement while drilling.

[0020] During underground drilling, a significant amount of noise data is generated, interfering with the accuracy of resistivity monitoring. Although the Kalman filter algorithm can effectively filter out noise when filtering the resistivity measured while drilling, there is a risk that the resistivity corresponding to geological changes may be weakened by the filter. Therefore, in this embodiment of the invention, the filter parameters in the Kalman filter algorithm are optimized and adjusted according to the distribution characteristics of the resistivity monitored in real time, so as to reduce the risk that the resistivity corresponding to the actual geological changes may be weakened by the filter.

[0021] Measuring instruments, including electromagnetic transmitting antennas, receiving antennas, signal processing modules, and data transmission modules, are installed inside the logging-while-drilling (LWD) collar, located a few meters to a dozen meters above the drill bit. During drilling, the electromagnetic transmitting antennas inside the measuring instruments continuously emit low-frequency electromagnetic waves of a specific frequency into the surrounding formation. The receiving antennas at each azimuth receive the reflected signals, thus obtaining the resistivity at that azimuth and depth. It should be noted that each fan-shaped area viewed from above when the drill bit is upright is used as a preset azimuth, i.e., the preset azimuths are 0°, 45°, 90°, 135°, 180°, 225°, 270°, and 315°. Simultaneously, the resistivity measured while drilling at each preset azimuth is stored in a separate sequence according to the measurement acquisition order. The resistivity acquisition frequency is 10Hz, which is not limited here.

[0022] Based on the above-mentioned resistivity acquisition method, in this embodiment of the invention, taking the current depth as an example, the target resistivity sequence at the current depth at each preset orientation during the underground drilling process is obtained by measurement while drilling. The target resistivity sequence refers to a time sequence consisting of resistivity acquired within 1 minute at the current moment. One preset orientation corresponds to one target resistivity sequence, providing basic data for subsequent noise analysis.

[0023] Step S102: For any target resistivity sequence, based on the local neighborhood distribution characteristics, perform noise analysis on each resistivity in the target resistivity sequence to obtain the initial suspected noise level of each resistivity, so as to screen out the initial suspected noise in the target resistivity sequence; perform stratigraphic variation interference analysis on each initial suspected noise, and screen out the target suspected noise from all the initial suspected noise.

[0024] For noisy data occurring during drilling, compared to normal resistivity data obtained through measurements while drilling, noisy data typically exhibits drastic abrupt changes within a short timeframe. Specifically, the changes occur in a relatively shorter timeframe and with a relatively larger amplitude. Therefore, in this embodiment of the invention, a target resistivity sequence at a preset orientation is used as an example, denoted as any target resistivity sequence. The analysis method for target resistivity sequences at other preset orientations is the same. Furthermore, based on the local neighborhood distribution characteristics, noise analysis is performed on each resistivity in any target resistivity sequence to obtain the initial suspected noise level for each resistivity.

[0025] Taking the q-th resistivity of any target resistivity sequence as an example, calculate the absolute value of the difference between the q-th resistivity and its left adjacent resistivity (the (q-1-th resistivity). Calculate the absolute value of the difference between the q-th resistivity and its right-hand adjacent resistivity (the (q+1-th resistivity)). The sum of the absolute values ​​of the differences is obtained. The accumulated absolute values ​​of the differences are normalized to obtain the resistivity change amplitude. Based on the slope of the change between adjacent resistivities in any target resistivity sequence, the number of maximum resistivities that are continuous and have the same trend to the left of the q-th resistivity is determined and denoted as the number of resistivities with the same trend to the left. Similarly, to the right of the q-th resistivity, determine the number of maximum resistivity that is continuous with it and has the same trend (monotonically increasing, monotonically decreasing, or constant), denoted as the number of resistivity with the same trend to the right. ; Calculate the cumulative value of the number of resistivities with the same trend on the left and the number of resistivities with the same trend on the right. The reciprocal of the accumulated quantity is normalized to obtain the same trend extension distance on both sides; the sum of the resistivity change amplitude and the same trend extension distance on both sides is used as the initial suspected noise level of the q-th resistivity.

[0026] The formula for calculating the initial suspected noise level of the q-th resistivity is as follows: ; in, This indicates the initial suspected noise level at the q-th resistivity. The maximum and minimum value normalization function is defined as the normalization range of all resistivities in any target resistivity sequence, corresponding to a value range of [0, 1]. It is worth noting that in all embodiments of the present invention, the value range of all values ​​processed by the maximum and minimum value normalization is [0, 1].

[0027] It should be noted that the greater the difference between the data value of the q-th resistivity and its adjacent resistivity, the greater the magnitude of resistivity change at the q-th resistivity, and the more it exhibits the characteristics of drastically changing noise data. The number of resistivity with the same trend to the left or right of the q-th resistivity is used to characterize the distance from the q-th resistivity to the resistivity where the trend changes positive or negative. The shorter the distance of the same trend extension on both sides of the q-th resistivity, the shorter the change time at the q-th resistivity, and the more it exhibits the characteristics of abrupt noise data. Therefore, the q-th resistivity is more likely to be noise, and the corresponding initial suspected noise level is higher.

[0028] Similarly, the initial suspected noise level of each resistivity in any target resistivity sequence is obtained. Since there must be noise in the normally acquired and monitored resistivity, the initial suspected noise level can initially divide the noise and normal resistivity at both ends of the value range of the initial suspected noise level. Therefore, the middle value of the value range, 1, is taken as the preset initial suspected noise level threshold. If the initial suspected noise level of any resistivity in any target resistivity sequence is greater than or equal to the preset initial suspected noise level threshold, then the any resistivity is confirmed as the initial suspected noise in the any target resistivity sequence. Thus, all the initial suspected noise in any target resistivity sequence are obtained.

[0029] However, during drilling, the resistivity exhibits abrupt changes when crossing formation boundaries. For example, the resistivity is 45 at a depth of 1000m, 47 at 1000.1m, 46 at 1000.2m, 48 at 1000.3m, and 12 at 1000.4m. As the drill bit approaches a new formation, the resistivity changes rapidly due to geological differences. Directly analyzing the drastic changes might misidentify the resistivity data of boundary changes as noise. Therefore, in this embodiment of the invention, the differences in data characteristics between real noise and boundary change data points are further distinguished to filter out target suspected noise from all initial suspected noise points, thereby eliminating initial suspected noise points caused by formation node interference.

[0030] Noise points appear relatively isolated and differ significantly from the regular data in their adjacent regions. Even if other noise points exist, their amplitude and duration of fluctuations differ from the current noise point due to the strong randomness of the noise itself. This leads to significant differences between the noise point and the data in its adjacent regions. In contrast, the resistivity monitored at the site of stratigraphic change maintains the level of the old stratigraphic layer before the change and remains at the level of the new layer after the change. Therefore, the resistivity will at least show a high degree of similarity to the data in its adjacent regions on one side. On the other hand, noise points typically return to their original level after abrupt changes, while the resistivity remains at the new level after a stratigraphic change. Therefore, the data in the adjacent regions on both sides of the noise point will generally have a relatively high degree of similarity, while the data in the adjacent regions on both sides of the resistivity corresponding to the stratigraphic change will generally show significant differences.

[0031] Based on the above feature analysis, a stratigraphic variation interference analysis is performed on each of the initial suspected noise points, and the target suspected noise points are obtained by screening from all the initial suspected noise points: Taking the p-th initial suspected noise point as an example, we set the number of resistivities contained in its left and right neighborhoods to 10. Then, in any target resistivity sequence, we calculate the absolute value of the difference between the p-th initial suspected noise point and each resistivity in its left neighborhood, and obtain the mean of the absolute values ​​of the differences, which is denoted as the mean left difference of the p-th initial suspected noise point. ,in, This represents the resistivity corresponding to the p-th initial suspected noise point. This represents the resistivity of the i-th resistivity within the left neighborhood of the p-th initial suspected noise point, where || denotes the absolute value sign. This represents the resistivity of the left neighborhood of the p-th initial suspected noise point. The smaller the mean difference on the left, the closer the resistivity of the p-th initial suspected noise point is to that of its left neighborhood.

[0032] Similarly, based on the absolute value of the difference between the p-th initial suspected noise point and each resistivity in its right neighborhood, the mean difference value to the right of the p-th initial suspected noise point is obtained. Take the minimum value between the mean difference on the left and the mean difference on the right. The reciprocal of the minimum value is normalized to obtain the one-sided neighborhood data similarity. Calculate the mean resistivity of all resistivity in the left neighborhood of the p-th initial suspected noise point, and denote it as the mean resistivity of the left neighborhood of the p-th initial suspected noise point. Similarly, based on the average resistivity of all resistivity in the left neighborhood of the p-th initial suspected noise point, the average resistivity of the right neighborhood of the p-th initial suspected noise point is obtained. The absolute value of the difference between the mean resistivity of the left neighborhood and the mean resistivity of the right neighborhood is normalized to obtain the overall neighborhood distribution difference. The sum of the unilateral neighborhood data similarity and the overall neighborhood distribution difference is calculated as the stratigraphic change node characteristic degree of the p-th initial suspected noise point: ; in, This indicates the degree of stratigraphic change node characteristics of the p-th initial suspected noise point. This represents the maximum and minimum value normalization function, whose normalization range refers to all initial suspected noise points, and | represents the absolute value sign.

[0033] It should be noted that, This is used to obtain the minimum value of the mean difference between the two sides of the p-th initial suspected noise point. The smaller the minimum value, the more similar the neighborhood data on one side of the p-th initial suspected noise point is, and the stronger the stratigraphic change node characteristics of the data point. Therefore, an inverse proportional form is used. ; The overall difference between the data values ​​in the left and right neighborhoods of the p-th initial suspected noise point is the greater the overall difference. The larger the overall difference, the more the p-th initial suspected noise point matches the neighborhood distribution characteristics of the stratigraphic change node, corresponding to the stratigraphic change node characteristic degree of the p-th initial suspected noise point.

[0034] Similarly, the stratigraphic change node characteristic degree of each initial suspected noise point is obtained. Since historical data collected and monitored under normal conditions will inevitably contain both noise and stratigraphic changes, the stratigraphic change node characteristic degree can divide noise and stratigraphic change nodes into two ends of the value range. Therefore, the value range boundary value of 1 is taken as the preset threshold for the stratigraphic change node characteristic degree. If the stratigraphic change node characteristic degree of any initial suspected noise point is less than the preset threshold, then the initial suspected noise point is determined to be a target suspected noise point. This yields all target suspected noise points in any target resistivity sequence that have been excluded from stratigraphic node interference.

[0035] Step S103: Perform one-dimensional clustering on all suspected target noise points to obtain multiple clusters. Analyze the density of the population distribution within each cluster to obtain the credibility of the noise points within each cluster, so as to identify credible noise points among all suspected target noise points. Analyze the distribution coverage and distribution quantity of all credible noise points, and combine the credibility of the noise points within each cluster corresponding to each credible noise point with the difference in resistivity between adjacent clusters to obtain the degree of noise presence in any target resistivity sequence.

[0036] After obtaining all the suspected noise points in any target resistivity sequence, the noise presence in any target resistivity sequence is evaluated based on the population distribution characteristics of the suspected noise points. The larger the distribution coverage of the suspected noise points in any target resistivity sequence, the greater the number of suspected noise points, and the stronger the degree of change at each suspected noise point, the higher the noise presence in any target resistivity sequence.

[0037] However, formation drilling exhibits distinct regional characteristics. For example, in homogeneous layers, the high degree of geological homogeneity results in highly uniform resistivity measurements. In fractured zones, the presence of numerous pores leads to significant fluctuations in resistivity measurements. Directly basing measurements on resistivity variations can easily misidentify these fractured zones as abrupt noise, thus masking the geological features of the fractured areas during noise filtering. Therefore, analyzing individual resistivity values ​​is insufficient; a comprehensive analysis of adjacent resistivity domains is necessary. Fracture formation is stress-controlled; if fractures exist, subsequent stress preferentially propagates along these weak surfaces. Thus, fractures exhibit inheritance; closely spaced fractures increase the likelihood of more fractures appearing in neighboring areas, forming concentrated fracture clusters. Consequently, in fractured formations, suspected anomalies in resistivity tend to be more concentrated, while environmental noise from drilling vibrations and shocks exhibits a more random and discrete distribution within the same formation. Therefore, in this embodiment of the invention, all suspected noise points are clustered to assess whether they are noise points or stratigraphic feature monitoring points in fracture zones based on the internal discrete characteristics of the clusters, thereby assigning the credibility of real noise points to different suspected noise points.

[0038] Specifically, one-dimensional K-means clustering is performed on all suspected target noise points to obtain multiple clusters. The K value is determined using the elbow method, which is a current technique. Then, the density of population distribution within each cluster is analyzed to obtain the credibility of noise points within each cluster, thereby identifying credible noise points among all suspected target noise points. When the distribution of suspected noise points within a cluster is denser and the sampling time interval between suspected noise points is closer, it indicates that the suspected noise points within the cluster better match the population distribution characteristics of resistivity corresponding to formation fractures in the fracture zone. Conversely, the greater the sampling time interval between suspected noise points within a cluster and the worse the uniformity of their differences, the better the suspected noise points within the cluster match the discrete and random population distribution characteristics of real noise points, and the higher the confidence level that the suspected noise points within the cluster are real noise points.

[0039] Therefore, for any cluster, if the number of suspected target noise points within any cluster is 1, it indicates that the cluster cannot be the resistivity corresponding to formation fractures; it is a real noise point, and the confidence level of the noise point within the cluster is 1. Conversely, if the number of suspected target noise points in any cluster is greater than 1, the sampling time interval between each suspected target noise point in the cluster and its left and right adjacent suspected target noise points is calculated, and the accumulated value of the sampling time interval is recorded as the distance between the two sides of the corresponding suspected target noise point. For example, for the j-th suspected target noise point in the c-th cluster, based on the sampling time of each suspected target noise point in the c-th cluster, the sampling time interval between the j-th suspected target noise point and its left adjacent suspected target noise point (the (j-1)-th suspected target noise point) is calculated. Similarly, calculate the sampling time interval between the j-th suspected target noise point and its right-side adjacent suspected target noise point (the (j+1)-th suspected target noise point). , sampling time interval With sampling time interval The sum between them is used as the distance between the two sides of the j-th suspected noise point. .

[0040] Based on the lateral spacing distances of each suspected noise point in any cluster, the mean and variance of the lateral spacing distances are calculated. The mean of the lateral spacing distances is normalized to obtain a first normalized value, and the variance of the lateral spacing distances is normalized to obtain a second normalized value. The sum of the first normalized value and the second normalized value is used as the intra-cluster noise confidence of any cluster. ; in, This represents the confidence level of intra-cluster noise in the c-th cluster. This represents the maximum and minimum value normalization function. This represents the distance between the two sides of the j-th suspected noise point. This represents the number of suspected noise points in the c-th cluster. This represents the variance of the two-sided spacing distance among all suspected noise points in the c-th cluster.

[0041] It should be noted that, The smaller the value, the more the distance between the two sides of the suspected noise point of the j-th target conforms to the dense distribution characteristics of formation fractures. The larger the value, the more it matches the discrete distribution characteristics of the spacing distance of real noise points. Therefore, The larger the value, the greater the probability that the suspected noise in the c-th cluster is a real noise, and the higher the credibility of the corresponding noise in the cluster. The larger the value, the more discrete the distance between the two sides of the suspected noise point in the c-th cluster, which better matches the distribution characteristics of the group where noise points appear randomly. The higher the confidence level that the suspected noise point in the cluster is a real noise point, the higher the confidence level of the noise point in the cluster.

[0042] Similarly, the intra-cluster noise confidence level of each cluster is obtained to distinguish the confidence evaluation value of noise points from the data points in the formation fracture zone. The higher the intra-cluster noise confidence level, the more likely the target suspected noise point in the corresponding cluster is a real noise point. The value range of intra-cluster noise confidence level is [0, 2]. The resistivity of the suspected formation feature and the resistivity of the suspected noise point can be divided at both ends of the value range. The middle value of 1 is taken as the intra-cluster noise confidence level threshold. When the intra-cluster noise confidence level of a cluster is greater than or equal to 1, it means that the target suspected noise point in the cluster is a noise point with better confidence. Then, all the target suspected noise points in the cluster are regarded as reliable noise points, and the intra-cluster noise confidence level of the cluster is assigned to each of the reliable noise points it contains.

[0043] Furthermore, after identifying reliable noise points in any target resistivity sequence and assigning different intra-cluster noise reliability to each reliable noise point, the distribution coverage and number of all reliable noise points are analyzed. Combined with the intra-cluster noise reliability corresponding to each reliable noise point and the difference in adjacent resistivity, the presence of noise in any target resistivity sequence is evaluated, thereby obtaining the degree of noise presence in any target resistivity sequence. Based on the sampling time corresponding to each credible noise point, the interval between the maximum and minimum sampling times is calculated, and the interval is normalized to obtain the noise distribution coverage; the number of credible noise points is counted, and the number is normalized to obtain the noise percentage. In any target resistivity sequence, the absolute value of the resistivity difference between each credible noise point and its left adjacent resistivity is obtained. The intra-cluster noise confidence level corresponding to each credible noise point is used as a weight. The absolute values ​​of the resistivity differences of all credible noise points are summed in a weighted manner to obtain a weighted sum result. The weighted sum result is then normalized to obtain the overall noise difference level. The product of the noise percentage and the overall noise difference is obtained, and the sum of the product and the noise distribution coverage is taken as the noise presence degree in any target resistivity sequence.

[0044] In one embodiment, the formula for calculating the degree of noise presence in any target resistivity sequence is: ; in, This indicates the degree of noise presence in any target resistivity sequence. This represents the maximum and minimum value normalization function. This represents the sampling time interval between the most recent and the last reliable noise points. This represents the number of reliable noise points in any target resistivity sequence. This represents the intra-cluster noise confidence level corresponding to the i-th reliable noise point. It represents the absolute value of the resistivity difference between the i-th reliable noise point and its left adjacent resistivity.

[0045] It should be noted that, The larger the value, the greater the coverage of the reliable noise in any target resistivity sequence, and the greater the number of reliable noise points. The more noise a target resistivity sequence contains, and the stronger the resistivity variation of each high-noise confidence point, the higher the degree of noise presence.

[0046] Similarly, following the method for obtaining the noise level in any target resistivity sequence, the noise level in the target resistivity sequence at each preset location is obtained.

[0047] Step S104: When performing Kalman filtering on each target resistivity sequence, the covariance parameter of the Kalman filtering algorithm is adaptively adjusted based on the noise level of each target resistivity sequence to obtain the corresponding optimized resistivity sequence.

[0048] The higher the level of noise in the target resistivity sequence, the higher the risk of the real-time monitored resistivity being a true noise source. In this case, historical predictions should be trusted more, the process noise covariance parameter Q in the Kalman filter algorithm should be reduced, the filtering should be strengthened, and real-time outliers should be suppressed. Conversely, the lower the level of noise in the target resistivity sequence, the more stable the current environment is, and the actual fluctuations are more in line with real geological changes. The lower the risk of the real-time monitored resistivity being a true noise source, the more trust should be placed in the real-time monitored resistivity. In this case, the process noise covariance parameter Q in the Kalman filter algorithm should be increased, the filtering should be weakened, and the real signal should not be filtered out as a result.

[0049] Specifically, taking any target resistivity sequence as an example, when performing Kalman filtering on any target resistivity sequence, the difference between the constant 2 and the noise level of the target resistivity sequence is calculated as the adaptive adjustment coefficient. The product of the adaptive adjustment coefficient and the basic process noise covariance parameter in the Kalman filtering algorithm is calculated as the adaptive process noise covariance parameter. Based on the adaptive process noise covariance parameter, Kalman filtering is performed on any target resistivity sequence to obtain the corresponding optimized resistivity sequence. The formula for calculating the adaptive process noise covariance parameter of any resistivity sequence is as follows: ,in, This represents the basic process noise covariance parameter in the Kalman filter algorithm. Let represent the adaptive process noise covariance parameter of any resistivity sequence, E represent the degree of noise presence of any target resistivity sequence, and 2 represent a constant.

[0050] It should be noted that since the noise level E of any target resistivity sequence has a range of [0, 2], using 2-E as the adaptive adjustment coefficient can achieve the following when E=0: The filtering is weak when E=2. It has extremely strong filtering capabilities, thus achieving "strengthening model confidence when there is a lot of noise and strengthening observation confidence when the signal is clean" by automatically adjusting Q through E.

[0051] This completes the adaptive Kalman filtering of the target resistivity sequence for each preset orientation, and obtains the corresponding optimized resistivity sequence. The optimized resistivity sequence refers to the sequence after filtering out a large amount of noise in the drilling measurement, while preventing real formation fluctuation data from being misidentified as noise. Then, all optimized resistivity sequences are transmitted to the data terminal for formation geological assessment during the formation drilling process through more accurate resistivity monitoring data.

[0052] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A method for intelligent processing of azimuth resistivity measured while drilling, characterized in that, The method includes: The target resistivity sequence at the current depth for each preset orientation during the underground drilling process was obtained by measurement while drilling. For any target resistivity sequence, based on the local neighborhood distribution characteristics, noise analysis is performed on each resistivity in the target resistivity sequence to obtain the initial suspected noise level of each resistivity, so as to screen out the initial suspected noise in the target resistivity sequence; stratigraphic variation interference analysis is performed on each of the initial suspected noise, and target suspected noise is screened out from all the initial suspected noise; One-dimensional clustering is performed on all suspected target noise points to obtain multiple clusters. The density of population distribution within each cluster is analyzed to obtain the credibility of noise points within each cluster, so as to identify credible noise points among all suspected target noise points. The distribution coverage and distribution quantity of all credible noise points are analyzed. Combined with the credibility of noise points within each cluster and the difference in resistivity between adjacent nodes, the degree of noise presence in any target resistivity sequence is obtained. When performing Kalman filtering on each target resistivity sequence, the noise covariance parameter is adaptively adjusted based on the noise level of each target resistivity sequence to obtain the corresponding optimized resistivity sequence.

2. The intelligent processing method for azimuth resistivity measured while drilling according to claim 1, characterized in that, The step of performing noise analysis on each resistivity in any target resistivity sequence based on local neighborhood distribution characteristics to obtain the initial suspected noise level for each resistivity includes: For any resistivity in any target resistivity sequence, calculate the absolute value of the difference between any resistivity and its left and right adjacent resistivities, obtain the cumulative value of the absolute value of the difference, and normalize the cumulative value of the absolute value of the difference to obtain the resistivity change amplitude. Based on the slope of change between adjacent resistivities in any target resistivity sequence, determine the maximum number of resistivities that are continuous and have the same trend to the left and right of any resistivity, respectively denoted as the number of resistivities with the same trend on the left and the number of resistivities with the same trend on the right; calculate the cumulative value of the number of resistivities with the same trend on the left and the number of resistivities with the same trend on the right, and normalize the reciprocal of the cumulative value to obtain the extension distance of the same trend on both sides. The sum of the resistivity change magnitude and the distance of the same extension on both sides is used as the initial suspected noise level for any resistivity.

3. The intelligent processing method for azimuth resistivity measured while drilling according to claim 1, characterized in that, The initial suspected noise points obtained from the screening of any target resistivity sequence include: If the initial suspected noise level of any resistivity in any target resistivity sequence is greater than or equal to a preset initial suspected noise level threshold, then the any resistivity is confirmed as the initial suspected noise in the target resistivity sequence.

4. The intelligent processing method for azimuth resistivity measured while drilling according to claim 1, characterized in that, The step involves performing stratigraphic variation interference analysis on each of the initial suspected noise points, and filtering out target suspected noise points from all initial suspected noise points, including: For any initial suspected noise point, the stratigraphic change node characteristic degree of the initial suspected noise point is obtained based on the resistivity change in the left and right neighborhoods of the initial suspected noise point in the target resistivity sequence. If the stratigraphic change node characteristic degree of the initial suspected noise point is less than a preset stratigraphic change node characteristic degree threshold, then the initial suspected noise point is determined to be a target suspected noise point.

5. The intelligent processing method for azimuth resistivity measured while drilling according to claim 4, characterized in that, The step of obtaining the stratigraphic change node characteristic degree of any initial suspected noise point based on the resistivity changes in the left and right neighborhoods of any initial suspected noise point in any target resistivity sequence includes: In any target resistivity sequence, calculate the absolute value of the difference between any initial suspected noise point and each resistivity in its left neighborhood, and obtain the mean of the absolute values ​​of the differences, which is denoted as the left-side difference mean of any initial suspected noise point; obtain the right-side difference mean of any initial suspected noise point, take the minimum value between the left-side difference mean and the right-side difference mean, and normalize the reciprocal of the minimum value to obtain the one-sided neighborhood data similarity. Calculate the mean resistivity of all resistivity in the left neighborhood of any initial suspected noise point, and denot it as the mean resistivity of the left neighborhood of any initial suspected noise point. Obtain the mean resistivity of the right neighborhood of any initial suspected noise point. Normalize the absolute value of the difference between the mean resistivity of the left neighborhood and the mean resistivity of the right neighborhood to obtain the overall neighborhood distribution difference. The sum of the similarity of the unilateral neighborhood data and the difference in the overall neighborhood distribution is calculated as the stratigraphic change node characteristic degree of any initial suspected noise point.

6. The intelligent processing method for azimuth resistivity measured while drilling according to claim 1, characterized in that, The analysis of the intra-cluster population density of each cluster to obtain the intra-cluster noise confidence level of each cluster includes: For any cluster, if the number of suspected target noise points in any cluster is greater than 1, then calculate the sampling time interval between each suspected target noise point in any cluster and its left and right adjacent suspected target noise points respectively, and obtain the cumulative value of the sampling time interval, which is recorded as the two-way interval distance of the corresponding suspected target noise point. Based on the lateral spacing distance between each suspected noise point in any cluster, the mean and variance of the lateral spacing distances are calculated. The mean of the lateral spacing distances is normalized to obtain a first normalized value, and the variance of the lateral spacing distances is normalized to obtain a second normalized value. The sum of the first normalized value and the second normalized value is used as the intra-cluster noise confidence of any cluster.

7. The intelligent processing method for azimuth resistivity measured while drilling according to claim 1, characterized in that, The analysis of the distribution coverage and number of all credible noise points, combined with the intra-cluster noise credibility and adjacent resistivity differences corresponding to each credible noise point, yields the degree of noise presence in any target resistivity sequence, including: Based on the sampling time corresponding to each credible noise point, the interval between the maximum and minimum sampling times is calculated, and the interval is normalized to obtain the noise distribution coverage; the number of credible noise points is counted, and the number is normalized to obtain the noise percentage. In any target resistivity sequence, the absolute value of the resistivity difference between each credible noise point and its left adjacent resistivity is obtained. The intra-cluster noise confidence level corresponding to each credible noise point is used as a weight. The absolute values ​​of the resistivity differences of all credible noise points are summed in a weighted manner to obtain a weighted sum result. The weighted sum result is then normalized to obtain the overall noise difference level. The product of the noise percentage and the overall noise difference is obtained, and the sum of the product and the noise distribution coverage is taken as the noise presence degree in any target resistivity sequence.

8. The intelligent processing method for azimuth resistivity measured while drilling according to claim 1, characterized in that, The step of adaptively adjusting the process noise covariance parameter based on the noise level of each target resistivity sequence to obtain the corresponding optimized resistivity sequence includes: For any target resistivity sequence, the difference between constant 2 and the noise level of the target resistivity sequence is calculated as an adaptive adjustment coefficient. The product of the adaptive adjustment coefficient and the basic process noise covariance parameter in the Kalman filter algorithm is calculated as the adaptive process noise covariance parameter. Based on the adaptive process noise covariance parameter, Kalman filtering is performed on the target resistivity sequence to obtain the corresponding optimized resistivity sequence.

Images