An artificial intelligence-based fault interpretation method

Abstract

The application relates to an artificial intelligence-based fault interpretation method, which comprises the following steps: converting three-dimensional seismic data into instantaneous phase bodies (S1); intercepting a plurality of instantaneous phase profiles perpendicular to the strike of a fault indication (S2); extracting phase points from each instantaneous phase profile at a certain phase interval, and assigning natural numbers to the points of different phases (S3); determining an interpretation point extraction distance A by using an iterative method according to the distance between two points with the same value (S4); extracting the points with the same value as the condition of meeting the distance A, and connecting the extracted points by lines to obtain a stratum interpretation line (S5); determining a breakpoint extraction distance B by using an iterative method according to the distance between the breakpoints of two different stratum interpretation lines (S6); extracting the breakpoints of each stratum interpretation line as the condition of meeting the distance B, and connecting the extracted breakpoints by lines to obtain a fault plane interpretation line (S7); and splicing the fault plane interpretation lines of each instantaneous phase profile to obtain a three-dimensional model of the fault (S8).

Description

TECHNICAL FIELD

[0001] The present application relates to the field of seismic interpretation, in particular to a fault interpretation method based on instantaneous phase. BACKGROUND

[0002] Fault is a geological structure formed by the fracture of crust under stress and the significant relative displacement of rock along the fracture, which may act as a channel or a barrier in the migration of oil and gas, and has important implications in the field of oil and gas exploration and development. Using seismic data to interpret faults is one of the main methods for finding faults in the process of oil and gas exploration and development. The conventional fault interpretation method is for interpreters to manually interpret faults on the vertical profile and horizontal slice of three-dimensional seismic data (amplitude), usually along the main line direction or any line perpendicular to the fault strike to track the fault, and then control the spatial correlation and extension of the fault through horizontal or layer slice, which is to realize fault interpretation by visually identifying the discontinuity of the reflection layer, and depends on the experience of interpreters and prior information of relevant geological knowledge, and has the problems of long cycle, great difficulty and strong subjectivity. For the geology with complex fault system and unknown fault strike in the work area, it is difficult to interpret faults, and the accuracy is also reduced.

[0003] Instantaneous phase data is a data volume obtained by processing seismic data, which reflects the change of boundary or the continuity of reflection. On the instantaneous phase profile, the top of the layer corresponds to the phase value near-180°, and the bottom corresponds to the phase value near +180°. Compared with the original seismic data, the boundary between layers is in the transition between trough and peak, and the phase change in the layer is continuous. This feature helps to make fine interpretation of faults, but cracks and small faults near the fault will also be identified, causing interference.

[0004] At the same time, the quality of seismic data is good or bad, and the change of strata lithology, pinchout and unconformity will all affect the instantaneous phase. With the existing computer processing method, it is still difficult to make correct judgment on faults. In addition, since the instantaneous phase data does not contain the information of seismic wave amplitude and frequency, and the difference between the real strata state is very large, it lacks intuitiveness, which also brings great difficulty to the work of interpreters.

[0005] The purpose of the present application is to provide a fault interpretation method based on artificial intelligence to improve the accuracy of interpreting faults using instantaneous phase data. SUMMARY

[0006] The technical scheme of the present application is a fault interpretation method based on artificial intelligence, characterized by the following steps: a transient phase calculation step (S1), converting three-dimensional seismic data into a three-dimensional transient phase volume through transformation; a transient phase profile acquisition step (S2), in the part with fault indications, a plurality of transient phase profiles are parallelly cut from the transient phase volume in a direction perpendicular to the strike of the fault indications, and the transient phase data of each transient phase profile is obtained; a phase point extraction and assignment step (S3), extracting phase points from each transient phase profile at a certain phase interval, assigning natural numbers to each extracted phase point from low to high or from high to low according to the phase size, and assigning the same value to the same phase.

[0007] An interpretation point extraction distance setting step (S4) sets the distance between any two points with the same value in the transient phase profile as the initial distance, draws a circle with all points with the same value as the center and the initial distance a as the radius, counts the proportion of circles with only two assigned points in all circles, and iteratively adjusts the radius of the circle until the proportion of circles with only two assigned points reaches a first threshold value, at which time the radius of the circle is taken as the interpretation point extraction distance A.

[0008] A stratigraphic interpretation line generation step (S5) searches and extracts assigned points with the same value in the transient phase profile under the condition that the interpretation point extraction distance A is satisfied, extracts the assigned points as interpretation points of the strata, and connects the interpretation points with the same value with the shortest line as the stratigraphic interpretation line; a breakpoint extraction distance setting step (S6) sets the distance between any two breakpoints of the stratigraphic interpretation line in the transient phase profile as the initial distance b, draws a circle with all the breakpoints as the center and the initial distance b as the radius, counts the proportion of circles with only two breakpoints in all circles, and iteratively adjusts the radius of the circle until the proportion of circles with only two breakpoints in all circles reaches a second threshold value, at which time the radius of the circle is taken as the breakpoint extraction distance B.

[0009] A cross-section line generation step (S7) searches and extracts the breakpoints of different stratigraphic interpretation lines in the transient phase profile under the condition that the breakpoint extraction distance B is satisfied, extracts the breakpoints as interpretation points of the cross section, and connects the interpretation points with the shortest line as the cross-section interpretation line; a fault model generation step (S8) splices the cross-section interpretation lines of each transient phase profile to obtain a three-dimensional model of the fault.

[0010] In the present application, by converting the three-dimensional seismic data volume into an instantaneous phase volume, using the principle that the seismic wave phase is discontinuous when encountering a fault, the fault under complex geological conditions is identified, especially the small fault can be correctly identified. Through the extraction of the interpretation point extraction distance A and the breakpoint extraction distance B set by the point extraction distance setting step (S4) and the breakpoint extraction distance setting step (S6), the interpretation points of the same fault and the breakpoints of the same fault can be correctly extracted according to the actual situation of each instantaneous phase profile, the interference of the lithology change of the stratum near the fault and the small crack is excluded, and the interpretation accuracy of the fault is improved.

[0011] Preferably, the phase range of the instantaneous phase profile is -180°~+180°, and in the phase point extraction and assignment step (S3), the phase interval is a positive divisor of 360°, and the phase points are extracted from each instantaneous phase profile.

[0012] Since the phase interval is a positive divisor of 360°, the phase points are extracted from each instantaneous phase profile, the distance between the assigned adjacent points can be pulled apart, and the extraction accuracy of the interpretation points is improved.

[0013] Preferably, in the phase point extraction and assignment step (S3), when the phase points are extracted from each instantaneous phase profile, the phase points adjacent to the phase points are also extracted, and the extracted phase points and the phase points adjacent to the phase points are assigned with the same value.

[0014] Since the extracted phase points and the phase points adjacent to the phase points are assigned with the same value, even if some phase points are missing, it will not affect the extraction of the interpretation points, and the success rate of the interpretation point extraction is improved.

[0015] Preferably, in the interpretation point extraction distance setting step (S4), the interpretation point extraction distance A is generated separately by each instantaneous phase profile.

[0016] Therefore, the interpretation point extraction distance A can be determined according to the actual state of each instantaneous phase profile, the success rate of the interpretation point extraction is improved, and the possibility of false extraction is reduced.

[0017] Preferably, in the breakpoint extraction distance setting step (S6), the breakpoint extraction distance B is generated separately by each instantaneous phase profile.

[0018] Therefore, the breakpoint extraction distance B can be determined according to the actual state of each instantaneous phase profile, the success rate of the breakpoint extraction is improved, and the possibility of false extraction is reduced.

[0019] Preferably, in the instantaneous phase calculation step (S1), the three-dimensional seismic data is converted into a three-dimensional instantaneous phase volume by using Hilbert transform.

[0020] Preferably, in the explanation point extraction distance setting step (S4), the first threshold value is determined according to geological investigation.

[0021] Preferably, in the breakpoint extraction distance setting step (S6), the second threshold value is determined according to geological investigation. BRIEF DESCRIPTION OF DRAWINGS

[0022] Figure 1 is a step schematic diagram of the artificial intelligence-based fault interpretation method;

[0023] Figure 2 (A) is a profile of a three-dimensional data volume;

[0024] Figure 2 (B) is a profile of a transient phase volume

[0025] Figure 3 is an explanatory diagram of phase point extraction, assignment, and stratigraphic interpretation lines;

[0026] Figure 4 is an explanatory diagram of connecting breakpoints to obtain fault interpretation lines;

[0027] Figure 5 is an explanatory diagram of forming fault interpretation lines from transient phase profiles;

[0028] Figure 6 is a formation explanatory diagram of stratigraphic interpretation lines;

[0029] Figure 7 is a flowchart of generating stratigraphic interpretation lines;

[0030] Figure 8 is a flowchart of determining an explanation point extraction distance. DETAILED DESCRIPTION

[0031] The preferred embodiments of the present application will be described in detail below with reference to the accompanying drawings, so that the advantages and features of the present application can be more easily understood in a suitable environment.

[0032] In this embodiment, the basic information used is three-dimensional seismic data, Figure 2 (A) is a profile of a three-dimensional seismic data volume obtained by software processing of three-dimensional seismic data. The acquisition of three-dimensional seismic data is a prior art, and the specific acquisition method will not be described here.

[0033] Figure 1 is a step schematic diagram of the artificial intelligence-based fault interpretation method.

[0034] As Figure 1 shown, in the transient phase calculation step S1, the three-dimensional seismic data is converted into a three-dimensional transient phase volume using Hilbert transform. Figure 2(B) is a cross-sectional view of the instantaneous phase volume, where the brightness of the colors represents the magnitude of the phase value. Furthermore, the continuous areas of color in the image represent discrete phase points in the data. Figure 2 Compared to the 3D seismic data volume profile in (A), the instantaneous phase volume profile does not reflect information such as seismic wave amplitude and frequency. However, information reflecting boundaries, such as layer interfaces, faults, and cracks, is emphasized, resulting in a more refined image quality than the 3D seismic data volume profile. In the figure, Ng, Ed, Es1, and Es2 are stratigraphic labels indicating potential for oil and gas exploration.

[0035] Instantaneous phase profile acquisition step S2, as follows Figure 2 (A) shows the location 100 with fault signs identified in the 3D seismic data profile, truncated parallel to the instantaneous phase volume perpendicular to the strike of the fault signs. Figure 2 (B) shows multiple instantaneous phase profiles, and instantaneous phase data from -180° to +180° are obtained in each instantaneous phase profile.

[0036] In step S3, phase point extraction and assignment involves extracting phase points from each instantaneous phase profile at regular intervals, using a positive divisor of 360° as the phase interval. Each extracted phase point is assigned a natural number, ordered from lowest to highest phase magnitude, with identical phases assigned the same value. For example, using a phase interval of 60°, six phase points are extracted: -180°, -120°, -60°, 0°, +60°, +120°, and +180°, and assigned values ​​of 1, 2, 3, 4, 5, 6, and 7 respectively, in ascending order. To address data incompleteness, additional phase points can be extracted near each extracted phase. For instance, near 0°, multiple phase points of -1°, -2°, +1°, and +2° can be extracted, with the five phase points including 0° assigned the uniform value of 4. Other extracted phase points undergo the same processing. In this way, even if some phase point data is missing, it will not affect the extraction of interpretation points as explained below, and can improve the success rate of interpretation point extraction.

[0037] In step S4, the distance setting for interpreting point extraction involves using the distance between any two points with the same assigned value in the instantaneous phase profile as the initial distance. Circles are drawn with all points with the same assigned value as centers and the initial distance 'a' as the radius. The proportion of circles containing only two assigned points is counted among all circles. The radius of these circles is iteratively adjusted until the proportion of circles with only two assigned points reaches a first threshold. The radius of this first threshold is taken as the distance A for interpreting point extraction. The first threshold is determined based on nearby geological survey results or experience; in this embodiment, it is set to 90%. The setting of the distance for interpreting point extraction is explained in detail below. Figure 8 ).

[0038] The stratum interpretation line generation step S5 searches and extracts the assignment points with the same assignment in the instantaneous phase profile under the condition that the extraction distance A for the interpretation point is satisfied, and the extracted assignment points are used as the interpretation points of the stratum, and the shortest connecting lines are used to connect the interpretation points with the same assignment to form the stratum interpretation lines.

[0039] The breakpoint extraction distance setting step S6 sets the distance between the breakpoints of any two stratum interpretation lines in the instantaneous phase profile as the initial distance b, draws a circle with all the breakpoints as the center and the initial distance b as the radius, counts the proportion of the circles with only two breakpoints in all the circles, and iteratively adjusts the radius of the circle until the proportion of the circles with only two breakpoints in all the circles reaches a second threshold value, at which time the radius of the circle is used as the breakpoint extraction distance B. The second threshold value is determined according to the nearby geological survey results or experience, and in this embodiment, 90% is taken.

[0040] The section line generation step S7 searches and extracts the breakpoints of different stratum interpretation lines in the instantaneous phase profile under the condition that the breakpoint extraction distance B is satisfied, and the extracted breakpoints are used as the interpretation points of the section, and the shortest connecting lines are used to connect the breakpoints to form the section interpretation lines (fault lines).

[0041] Figure 5 is a diagram illustrating the formation of the fault interpretation lines in the instantaneous phase profile, as shown in Figure 5 11 stratum interpretation lines The breakpoints of each stratum interpretation line are connected by the shortest lines to obtain 5 section interpretation lines (fault lines) FL1, FL2, FL3, FL4, and FL5.

[0042] The fault model generation step S8 splices the section interpretation lines of each instantaneous phase profile to obtain a three-dimensional model of the fault.

[0043] The principle on which the present application is based is that when seismic waves propagate in a homogeneous medium, the instantaneous phase (hereinafter referred to as phase) calculated by Hilbert transform is continuous; when the seismic waves encounter cracks, faults, and other propagation medium mutations during transmission, the phase will change significantly at the mutation position of the medium, and will present discontinuity in the instantaneous phase profile. Therefore, when the breakpoints of phase discontinuity appear in the instantaneous phase profile, it can be judged that a fault exists at this position.

[0044] Since the instantaneous phase has very sensitive seismic properties, when it is used to interpret faults, nearby small fractures and other geological structure changes are usually also interpreted as faults, resulting in interpretation errors.

[0045] In the present application, the three-dimensional seismic data body obtained by processing the information of seismic exploration is used as the basic information, the three-dimensional seismic data is converted into three-dimensional instantaneous phase body by using Hilbert transform, and in the part with fault indication, a plurality of instantaneous phase sections are parallelly cut from the instantaneous phase body in the direction perpendicular to the strike of the fault indication. The instantaneous phase is a seismic attribute parameter independent of seismic amplitude, and is also a measure of the continuity and direction of the same phase axis on the seismic section, and is not affected by the strength of seismic amplitude and energy. The instantaneous phase body is the distribution of discrete values of seismic wave instantaneous phase on the section, and each discrete data point on the instantaneous phase section contains a phase degree and coordinates.

[0046] Before Hilbert transform, the three-dimensional seismic data body can be divided into horizons and structural positions with oil and gas prospects according to the interpretation results of time-depth conversion and well-seismic calibration, and then wavelet transform frequency division processing is performed to obtain the data body with appropriate horizons, appropriate frequency bands and appropriate waveforms, and then Hilbert transform is performed to obtain the instantaneous phase body. That is, by performing the above-mentioned preprocessing on the seismic data, the interference caused by the change of stratum lithology, pinch-out and small fault zone can be excluded to a certain extent, and the interpretation accuracy of the fault can be improved.

[0047] Figure 3 is a schematic diagram of phase point extraction, assignment and connection. Figure 3 is a schematic diagram of the part in the circle in (B) for fault interpretation. As shown in the enlarged part in the figure, Figure 2 Figure 3 is a schematic diagram of the formation of each different phase stratum interpretation line. Each stratum interpretation point on the same phase with a distance A meeting the extraction condition is connected by the shortest line to form a stratum interpretation line.

[0048] Figure 4 is a schematic diagram of connecting the breakpoints to obtain the fault interpretation line. As shown in the figure, Figure 4

[0049] The formation of the stratum interpretation point and the fault section interpretation line is described below.

[0050] Figure 6 is a schematic diagram of the formation of the stratum interpretation line. Figure 6 ​​​In the figure, the phase interval of the extracted phase points is 60°, the phase values of the phase points Pll a, Pll b are -180°, and the values are assigned as 1; the phase values of the phase points P12a, P12b are -120°, and the values are assigned as 2; the phase values of the phase points P13a, P13b are -60°, and the values are assigned as 3. The difference between the phase values of the phase points Pll a(b) and P12a(b) is 60°, and the difference between the phase values of the phase points P12a(b) and P13a(b) is also 60°. The solid lines in the figure are the connecting lines of the adjacent phase points with the same phase values, and the hollow marked phase points Pll b (fl lb), P12b (f12b) are the break points. … are the connecting lines of the adjacent phase points with the same phase values, and the hollow marked phase points Pll b (fl lb), P12b (f12b) are the break points.

[0051] Figure 7 is a flow chart for generating the stratigraphic interpretation line, that is, the stratigraphic interpretation line generating step S5 includes the following steps. As shown in Figure 7 , a preparation sub-step S501 is performed to establish a coordinate system for calculation in the instantaneous phase profile with geographical coordinates as the horizontal axis and seismic wave propagation time as the vertical axis, and to obtain the coordinates of the assigned phase points.

[0052] A circle making sub-step S502 is performed to read the set interpretation point extraction distance A (interpretation point extraction distance setting step S4), and to take a phase point with the minimum value as the center of the circle according to the value size, and to make a circle with the distance A as the radius.

[0053] A judgment sub-step S503 is performed to judge whether the assigned points with the same value as the value of the center of the circle can be read in the circle, that is, the assigned points not connected with the center of the circle. If yes (YES), a connecting line sub-step S504 is performed, and if no (NO), a step S506 is entered to judge whether there are other assigned points with the same value outside the circle.

[0054] The connecting line sub-step S504 connects the center of the circle with the nearest unconnected assigned point read.

[0055] A circle making sub-step S505 is performed to take the assigned point connected with the center of the circle in the last step as the new center of the circle, and to make a circle with the distance A as the radius, and to return to the judgment sub-step S503 to judge whether the unconnected assigned points with the same value as the value of the center of the circle can be read in the new circle. The above steps are repeatedly performed until all the phase points with the same phase value and the mutual distance of A (or less than A) are connected into a connecting line, that is, the stratigraphic interpretation line (i.e., Figure 5 In the figure ).

[0056] The unconnected assignment point judgment sub-step S506 judges whether there are other unconnected assignment points. If there are (YES), the circle drawing sub-step S502 is performed to start connecting as a new stratigraphic interpretation line. If there are not (NO), the assignment points of the assignment are all connected, and the output break point sub-step S507 is entered to save the end point information of all connected lines (stratigraphic interpretation lines).

[0057] The unconnected assignment judgment sub-step S508 judges whether there are unconnected assignments in the instantaneous phase profile. If there are (YES), the step S509 is entered to change the assignment point to the next assignment point, and the step S502 is returned to repeat the above steps until all assignment points of the assignment are processed.

[0058] Figure 8 The flowchart is a process of determining the interpretation point extraction distance, that is, the interpretation point extraction distance setting step S4 includes the following steps.

[0059] The distance between any two points with the same assignment in the instantaneous phase profile is taken as the initial distance (S301a), and the assignment points with the same assignment are read (S301b). The circles with the points with the same assignment as the center and the initial distance a as the radius are drawn (S301c). The proportion of the circles with only two assignment points inside in all circles, that is, the probability P, is counted (S301d).

[0060] It is judged whether the probability P is greater than 90%. If it is not (NO), the set distance D is corrected in the direction of increasing probability (S301f). Through continuous iteration processing, the distance D is finally obtained as the interpretation point extraction distance A after the proportion of the circles with only two assignment points in the total number of circles reaches 90%.

[0061] The above method can connect the points with the same phase and adjacent to each other. That is, the problem that the distance between the phase points is not equal due to the complexity of the stratum geology, and it is difficult for the computer to determine whether the adjacent standard is difficult to judge can be solved.

[0062] The length of the distance is modified each time, for example, which can be calculated according to the following formula.

[0063] D0=D0+α(P2-P 设 )·(P1-P3)·D0

[0064] In the above formula, D0 is the initial distance, a is a learning rate parameter set according to the actual situation, P 设is a first preset value, P1 is a sampling probability of falling into 1 or below phase points in a circle, P2 is a sampling probability of falling into 2 phase points in the circle, and P3 is a sampling probability of falling into 3 or above phase points in the circle. Through iterative calculation by the above formula, the step length of iteration can be enlarged or reduced according to the size of error, so as to prevent the iterative result from swinging around the correct value with a fixed step length, thereby preventing the calculation from entering a dead loop.

[0065] It should be noted that the above embodiments illustrate the application rather than limit the application, and alternative embodiments can be designed by those skilled in the art without departing from the scope of the appended claims. In the claims, any reference signs located between parentheses shall not be construed as limiting the claims.

Claims

1. An artificial intelligence-based fault interpretation method, characterized by: The method comprises the following steps: a transient phase calculation step (S1) for converting three-dimensional seismic data into a three-dimensional transient phase volume through transformation; a transient phase profile acquisition step (S2) for obtaining transient phase data of each transient phase profile by parallelly cutting a plurality of transient phase profiles from the transient phase volume at positions with fault indications and in a direction perpendicular to the strike of the fault indications; a phase point extraction and assignment step (S3) for extracting phase points from each transient phase profile at a certain phase interval, assigning each extracted phase point with a natural number according to the phase size from low to high or from high to low, and assigning the same phase with the same value; an explanation point extraction distance setting step (S4) for setting the distance between any two points with the same value in the transient phase profile as an initial distance, drawing a circle with all the points with the same value as the center and the initial distance a as the radius, counting the proportion of the circle with only two assigned points in all the circles, and iteratively adjusting the radius of the circle until the proportion of the circle with only two assigned points reaches a first threshold, and the radius of the circle at this time is taken as the explanation point extraction distance A; a stratum explanation line generation step (S5) for searching and extracting assigned points with the same value in the transient phase profile under the condition that the explanation point extraction distance A is satisfied, taking the extracted assigned points as explanation points of the stratum, and connecting the explanation points with the same value with the shortest line as the explanation line of the stratum; a breakpoint extraction distance setting step (S6) for setting the distance between any two breakpoints of the explanation lines of the stratum in the transient phase profile as an initial distance b, drawing a circle with all the breakpoints as the center and the initial distance b as the radius, counting the proportion of the circle with only two breakpoints in all the circles, iteratively adjusting the radius of the circle, until the proportion of the circle with only two breakpoints in all the circles reaches a second threshold, and the radius of the circle at this time is taken as the breakpoint extraction distance B; a section line generation step (S7) for searching and extracting the breakpoints of different explanation lines of the stratum in the transient phase profile under the condition that the breakpoint extraction distance B is satisfied, taking the extracted breakpoints as explanation points of the section, and connecting the explanation points with the shortest line as the explanation line of the section; a fault model generation step (S8) for splicing the section explanation lines of each transient phase profile to obtain a three-dimensional model of the fault.

2. The AI-based fault interpretation method according to claim 1, wherein: the phase range of the transient phase profile is -180° to +180°, and in the phase point extraction and assignment step (S3), the phase interval is a positive divisor of 360°, and the phase points are extracted from each transient phase profile.

3. The AI-based fault interpretation method according to claim 1, wherein: in the phase point extraction and assignment step (S3), the phase points are extracted from each transient phase profile while simultaneously extracting phase-adjacent phase points, and the extracted phase points and the phase-adjacent phase points are assigned with the same value, and in the explanation point extraction distance setting step (S4), the explanation point extraction distance A is generated separately from each transient phase profile. ​ ​ ​ 4.The artificial intelligence-based fault interpretation method of claim 1, wherein: ​ 5. The artificial intelligence-based fault interpretation method of claim 1, wherein: In the breakpoint extraction distance setting step (S6), the breakpoint extraction distance B is generated separately by each instantaneous phase profile.

6. The artificial intelligence-based fault interpretation method of claim 1, wherein: In the instantaneous phase calculation step (S1), the Hilbert transform is used to convert the three-dimensional seismic data into a three-dimensional instantaneous phase volume.

7. The artificial intelligence-based fault interpretation method according to any one of claims 1 to 6, characterized in that: In the interpretation point extraction distance setting step (S4), the first threshold is determined according to geological investigation.

8. The artificial intelligence-based fault interpretation method of claim 7, wherein: In the breakpoint extraction distance setting step (S6), the second threshold is determined according to geological investigation.

Images