Single well sedimentary microfacies automatic identification method based on optimization model

Abstract

The invention discloses a single well sedimentary microfacies automatic identification method based on an optimization model, and the method comprises the steps: dividing a target layer section into lithology combinations according to the lithology characteristics of a conventional logging curve; dividing the target layer section into basic primitives according to the depth according to the shape features of the conventional logging curve; constructing a composite layer corresponding to the lithology combination according to the basic primitive layer; selecting a function corresponding to the expected function image as a target function, solving a fitting result of the composite image layer to the target function, and evaluating the matching degree of the composite image layer and the fitting result; when the matching degree is not less than a threshold value, or the composite layer only comprises one basic layer; according to a graph fitting result of the composite graph layer, combining adjacent graph layers satisfying the complex graph; otherwise, selecting the feature points, segmenting the composite layer, and re-evaluating the matching degree between the composite layer and the fitting result; and constructing the sedimentary microfacies according to the depth range of the composite layer, determining the type of the sedimentary microfacies, and completing automatic identification of the sedimentary microfacies of the single well.

Description

Technical Field

[0001] The present invention belongs to the field of geophysical logging, and particularly relates to a method for automatically identifying single-well sedimentary microfacies based on an optimization model. Background Technique

[0002] Sedimentary facies are stratigraphic units that reflect specific sedimentation processes and complex sedimentary environments, and potential sedimentary environments and their geological changes can be determined based on characteristics such as the lithology of rocks, fossil content, and sedimentary structures. In recent years, the analysis and research of sedimentary facies have become an indispensable and important part of the oil and gas exploration and development process. With the in-depth understanding of sedimentary facies, the general view is to define facies as the combination of sedimentary environments and the characteristics of sediments formed in these environments. Microfacies are the most basic constituent units in sedimentary systems and reflect sedimentary rocks formed under basically the same sedimentation conditions. The concept of logging facies is proposed to evaluate or interpret sedimentary facies using logging data. Logging data have a high vertical resolution, contain rich underground geological characteristics and sedimentary environment information, and have the advantages of low cost and short operation cycle, etc. Therefore, using logging curves for related research on sedimentary facies is one of the essential means.

[0003] After the emergence of the automatic identification method for sedimentary facies, explorations on the automatic identification method have been carried out in two directions. The first is the artificial intelligence direction, attempting to apply the image recognition and other capabilities of artificial intelligence algorithms in automatic identification. The second is the statistical analysis method, establishing a general process framework for automatic stratification (such as geometric features like variance within a layer), feature extraction (such as cluster analysis), and type discrimination (such as Bayesian determination).

[0004] With the emergence and progress of artificial intelligence technology, automatic recognition based on artificial intelligence methods has made great development. Especially in the past five years, almost all academic papers and degree research have focused on this direction. These studies on automatic recognition using artificial intelligence methods have deeply explored the expression of longitudinal curve features, the selection of training feature parameters, the improvement of algorithm models, and the construction of training sample sets. In contrast, automatic recognition based on statistical analysis methods is relatively less, but the application of morphological factors in each link has been developed. The automatic stratification method has become more precise. For example, the approximate position of the layer interface is first determined by an index function, and then the layer interface is finely calibrated using geometric parameters such as variance. Feature extraction methods referring to geometric parameters such as inflection points, curvature, thickness, and texture have emerged. Type discrimination has started to use geometric parameters such as the slope and variance of the curves within the layer to represent common shapes such as bell-shaped, box-shaped, funnel-shaped, finger-shaped, serrated, and composite shapes. In recent years, some studies have analyzed the graphical features of seismic data to complete sedimentary facies characterization, and these studies are aimed at the texture attributes of two-dimensional seismic data images. It is worth noting that in recent years, a small number of studies have analyzed the geometric features of well logging curve data and applied them to the practice of automatic recognition. Matching and fitting linear curved structures with function curves (such as the sine function family, second moment equivalent ellipse function, etc.) to express the influence of geometric features on the determination of sedimentary microfacies types has a certain effect on optimizing the recognition efficiency and result accuracy.

[0005] Artificial intelligence has advantages, but it also has its disadvantages. One of them is that it requires a relatively large training sample set, and another disadvantage is that it requires the features reflected in the training sample set to be as comprehensive as possible. If there are biases in the training set, the trained model may have large errors. For sedimentary facies research, on the one hand, the formation and evolution processes of geological environments in different regions and different blocks vary greatly; on the other hand, due to cost factors, the proportion of core-taking and other operations for actual verification in the same block is very limited. Therefore, the scale of the training set that can be obtained is small, and it is difficult to achieve a comprehensive sample set. This has a great impact on the application of artificial intelligence in sedimentary facies automatic recognition. Although there are currently many studies focusing on solving the sample set problem, the solutions are more about obtaining more derivative sample sets from the direct sample set through mathematical methods to expand the sample set scale, while the evaluation and solution of the deviation degree of the sample set are very difficult.

[0006] At present, the automatic identification of sedimentary facies in the direction of statistical analysis is divided into two small directions: 1. Numerical analysis of geometric morphological features; 2. Function curve matching and fitting. The numerical analysis of geometric morphological features regards a section of curve as a point set, extracts numerical features from this data set, such as calculating slope, curvature, inflection point, variance, thickness, etc., and then applies these feature values to various links of layering, analysis, and discrimination. This method ignores the sequential correlation between data points in the data set. In the research of sedimentary facies identification tasks, not only the distribution characteristics of sedimentary facies belts on the plane should be considered, but also the sequential correlation in space during the sedimentation process of sedimentary rocks should be emphasized. The research direction of function curve matching and fitting is more comprehensive in the morphological analysis of linear curved structures and expresses the sequential correlation in space to a certain extent. However, the current idea is restricted by the framework of automatic layering, feature extraction, and type discrimination. Automatic layering and type discrimination are independent of each other, and the morphological analysis can only be applied in the type discrimination stage, which limits its application in the automatic layering stage and affects the further optimization of automatic sedimentary facies identification. Summary of the Invention

[0007] In order to solve the problems existing in the prior art, the present invention provides a method for automatically identifying single-well sedimentary microfacies based on an optimization model, which is used to solve the problem of insufficient analysis of the sequential correlation of curve morphology or data in space in the existing framework or method during layering, so as to improve the accuracy of automatic identification of single-well sedimentary facies.

[0008] To achieve the above object, the present invention provides the following technical solutions:

[0009] A method for automatically identifying single-well sedimentary microfacies based on an optimization model includes the following processes.

[0010] Divide the target interval into lithologic combinations according to the lithologic characteristics of conventional logging curves; and divide the target interval into basic primitive units according to the shape characteristics of conventional logging curves by depth.

[0011] Construct a composite layer corresponding to the lithologic combination based on the basic primitive unit layer.

[0012] Select the function corresponding to the desired function image as the target function, obtain the fitting result of the composite layer for the target function, and evaluate the matching degree between the composite layer and the fitting result.

[0013] When the matching degree is not less than the threshold, or, when the composite layer only contains one basic layer; merge adjacent layers that meet the complex graph according to the graph fitting result of the composite layer; otherwise, select feature points, split the composite layer, and re-evaluate the matching degree between the composite layer and the fitting result.

[0014] Construct sedimentary microfacies according to the depth range of the composite layer and determine the types of sedimentary microfacies to complete the automatic identification of sedimentary microfacies for a single well.

[0015] Preferably, divide the specified formation into lithologic combinations according to the value range of conventional curves, which specifically includes the following steps.

[0016] Step 1.1, calculate the shale classification value of the natural gamma curve, and then classify each sampling point p in the target interval according to the natural gamma sampling value v(d) at depth d; if v(d)>shale classification value, the type of sampling point p corresponding to depth d is a shale sampling point; otherwise, the type of sampling point p is a non-shale sampling point.

[0017] Step 1.2, group all consecutive sampling points p with the same type in Step 1.1 together. If the type of the sampling point is a shale sampling point, it is grouped into a shale layer; if the type of the sampling point is a non-shale sampling point, it is grouped into a non-shale layer.

[0018] Step 1.3, calculate the sandstone classification value of the natural gamma curve, and then classify each sampling point in the non-shale layer interval according to the natural gamma sampling value v(d) at depth d; if v(d)>sandstone classification value, the type of sampling point p corresponding to depth d is a mixed rock sampling point; otherwise, the type of sampling point p is a sandstone sampling point.

[0019] Step 1.4, group all consecutive sampling points p with the same type in Step 1.3 together. If the type of the sampling point is a mixed rock sampling point, it is grouped into a mixed rock layer; if the type of the sampling point is a sandstone layer point, it is grouped into a sandstone layer.

[0020] Step 1.5, traverse all layers from bottom to top in the sedimentary order, and calculate the thin layer index t according to the ratio a of the current layer thickness to the thicknesses of the upper and lower adjacent layers and the ratio b of the median value of the natural gamma value of the current layer to the median values of the natural gamma values of the upper and lower adjacent layers; if t<400, the current layer is an ultra-thin layer; otherwise, the current layer is not an ultra-thin layer.

[0021] Step 1.6, if all layers in Step 1.5 are not ultra-thin layers, regard each layer as a lithologic combination, and then go to the next step.

[0022] Step 1.7, if the current layer is an ultra-thin layer, correct the type of the current layer according to the types of its upper and lower adjacent layers; if neither of the upper and lower adjacent layers is an ultra-thin layer and they have the same type, correct the type of the current ultra-thin layer to be the same as that of the upper and lower adjacent layers; if neither of the upper and lower adjacent layers is an ultra-thin layer and they have different types, take the median value of the layer thickness and natural gamma as the characteristic value, select the layer that is most similar to the current ultra-thin layer among the upper and lower adjacent layers, and correct the type of the previous ultra-thin layer to be the type of this layer; if the upper and lower adjacent layers contain ultra-thin layers, then select the ultra-thin layer with the smallest thin layer index among the consecutive ultra-thin layers to correct the type.

[0023] Step 1.8, traverse all layers, merge consecutive layers of the same type together, and if there are still ultra-thin layers, go to Step 1.5.

[0024] Furthermore, in Step 1.5, the thin layer index t = a 2 + a × b + b 2 。

[0025] Preferably, divide the specified formation into basic primitives according to the shape characteristics of the conventional curves, which specifically includes the following steps.

[0026] Step 2.1, traverse each sampling depth d in the target interval from bottom to top according to the sedimentation order, calculate the slope k(p) of the natural gamma curve at the depth d, if k(p) is less than -threshold, then p is a decreasing point; if -threshold < k(p) < threshold, then p is a stable point; if k(p) > threshold, then p is an increasing point.

[0027] Step 2.2, consecutive decreasing points form a decreasing primitive in the basic primitive, consecutive stable points form a stable primitive in the basic primitive, and consecutive increasing points form an increasing primitive in the basic primitive.

[0028] Preferably, construct a composite layer corresponding to the lithology combination based on the basic primitive layer, which specifically includes the following steps.

[0029] Step 3.1, traverse all lithology combinations from bottom to top according to the sedimentation order.

[0030] Step 3.2, for each lithology combination, find all basic primitives that have an intersection with it in depth.

[0031] Step 3.3, construct a composite layer from the set of basic primitives found in Step 3.2.

[0032] Preferably, Step 4: Select a graphical model and solve the fitting result and fitting degree of the composite layer, which specifically includes the following steps.

[0033] Step 4.1, traverse all composite layers from bottom to top according to the sedimentation order.

[0034] Step 4.2: Select the function model y = kx + b of the straight line as the graphic model for the image;

[0035] Step 4.3: For each composite layer L, obtain k(L) corresponding to the fitting result of y = kx + b, and calculate the variance δ(L) of the composite layer when k = k(L);

[0036] Take the reciprocal of the variance δ(L) as the fitting degree of the composite layer L with respect to the straight line graphic model y = kx + b.

[0037] Preferably, select feature points and split the composite layer, which specifically includes the following steps:

[0038] Step 6.1: Traverse the boundaries between all adjacent basic primitives in the composite layer to obtain the deepest point p(max) with the highest curvature at all boundaries;

[0039] Step 6.2: Split the composite layer into two new composite layers at the deepest point p(max). For each new composite layer, iterate repeatedly until all composite layers meet the requirements, and then continue to evaluate the matching degree between the composite layer and the fitting result.

[0040] Preferably, merge adjacent layers that meet the complex graphics according to the graphic fitting result of the composite layer, which specifically includes the following steps:

[0041] Step 7.1: Traverse the fitting slope k(L) of the function model y = kx + b corresponding to all composite layers from bottom to top;

[0042] Step 7.2: For the fitting results of continuously adjacent composite layers of the same lithology from bottom to top, if they form combinations of k(L) < -1.5, -1.5 < k(L) < 1.5, k(L) > 1.5, then merge them into box-shaped composite layers; if they form combinations of k(L) < -1.5, k(L) > 1.5, then merge them into finger-shaped composite layers;

[0043] Step 7.3: Do not merge in other cases. If k(L) < -1.5, the composite layer is a funnel-shaped composite layer; if k(L) > 1.5, the composite layer is a bell-shaped composite layer.

[0044] Preferably, construct sedimentary microfacies based on the depth range of the composite layer and determine the sedimentary microfacies type, which specifically includes the following steps:

[0045] Step 8.1: Determine the depth range of the sedimentary microfacies according to the depth range of the composite layer;

[0046] Step 8.2: Determine the type of sedimentary microfacies according to the graphic type of the composite layer, combined with the sedimentary subfacies type corresponding to the depth range.

[0047] Compared with the prior art, the present invention has the following beneficial technical effects:

[0048] The present invention provides a method for automatically identifying single-well sedimentary microfacies based on an optimization model. By fitting and matching conventional logging curves with an objective function reflecting graphic features, the maximum stratification result satisfying the feature matching degree constraint is found, and the type of sedimentary facies within the stratification is determined according to the graphic features of the objective function with the highest matching degree, realizing the automatic identification of sedimentary facies. The stratification of the method of the present invention is to solve an optimization model based on lithologic features of numerical statistics and shape features of the objective function, solving the problem of insufficient analysis of the sequence correlation of curve morphology or data in space during the stratification process of existing sedimentary microfacies identification methods based on statistical analysis.

[0049] The objective function in the present invention can express the graphic features corresponding to its function image and has sequence correlation. The stratification process of this method is completed on the basis of preliminary lithologic property analysis with the premise of satisfying the feature matching degree constraint. Therefore, compared with all existing methods, the graphic features including lithologic features and sequence correlation are simultaneously reflected during the stratification process, and the stratification result has a higher credibility. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 It is a schematic flow chart of a method for automatically identifying single-well sedimentary microfacies based on an optimization model of the present invention.

[0051] Figure 2 It is an example diagram of lithologic combination division.

[0052] Figure 3 It is an example diagram of basic lithologic primitive division.

[0053] Figure 4 It is an example diagram of composite primitive.

[0054] Figure 5 It is a schematic diagram of basic primitive. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0055] The following further describes the present invention in detail with specific embodiments, which is an explanation rather than a limitation of the present invention.

[0056] The present invention proposes a method for automatically identifying single - well sedimentary microfacies based on an optimization model. A function with a function image as the basic graph is used as the objective function, and the optimization model in mathematical modeling is adopted. In the iterative process of curve data fitting, fitting result matching degree analysis, and curve graph segmentation, the maximum stratification result satisfying the feature matching degree constraint is solved, and the sedimentary facies type is determined according to the extracted function graph, so as to solve the problem of insufficient analysis of the sequence correlation of curve morphology or data in space in the existing framework or method during stratification, thereby improving the accuracy of automatic identification of single - well sedimentary facies.

[0057] As Figure 1 shown, a method for automatically identifying single - well sedimentary microfacies based on an optimization model of the present invention includes the following processes.

[0058] First, the target interval is divided into lithology groups according to the lithology characteristics of conventional logging curves (such as shale baseline, etc.), and the target interval is divided into basic graphic elements according to shape characteristics (such as slope, curvature, etc.). The lithology group is a section of strata with the same lithology and is the largest division of sedimentary microfacies. The basic graphic element is the smallest graphic unit with the same graphic characteristics and is the smallest division of sedimentary microfacies. The basic graphic elements that are continuous in depth and belong to the same lithology group are combined into composite graphic elements.

[0059] Select the function corresponding to the desired function image as the objective function, obtain the fitting result of the composite graphic layer for this objective function, and evaluate the matching degree between the composite graphic layer and the fitting result. If the matching degree of the composite graphic layer meets the set matching degree threshold, or the composite graphic layer contains only one basic graphic layer, then proceed to the next step. Otherwise, select the depth point that has the greatest impact on the matching degree, and divide the composite graphic layer into two new composite graphic layers. Iterate repeatedly until the matching degree of all composite graphic layers meets the set matching degree threshold, or the composite graphic layer contains only one basic graphic layer.

[0060] Traverse the composite graphic layers from bottom to top in the sedimentary order, and merge adjacent composite graphic layers that can form complex graphs according to the function graph characteristics corresponding to the fitting results.

[0061] Determine the depth range of the sedimentary microfacies according to the depth range of the composite graphic layer, and determine the sedimentary microfacies type according to the simple function image or complex function image corresponding to the fitting result of the composite graphic layer.

[0062] The automatic identification method of single - well sedimentary microfacies based on an optimization model proposed by the present invention realizes the automatic identification of sedimentary facies by fitting and matching conventional logging curves with an objective function reflecting graphic features, finding the maximum stratification result that meets the feature matching degree constraint, and determining the type of sedimentary facies within the stratification according to the graphic features of the objective function with the highest matching degree. The stratification of this method is to solve an optimization model based on lithologic features of numerical statistics and shape features of the objective function, solving the problem of insufficient analysis of the sequence correlation of curve morphology or data in space during the stratification process of existing sedimentary microfacies identification methods based on statistical analysis.

[0063] The objective function in the method of the present invention can express the graphic features corresponding to its function image and has sequence correlation. The stratification process of this method is completed on the premise of meeting the feature matching degree constraint based on the preliminary lithologic property analysis. Therefore, compared with all existing methods, the graphic features including lithologic features and sequence correlation are simultaneously reflected during the stratification process, and the credibility of the stratification result is higher.

[0064] Embodiment

[0065] An embodiment of the automatic identification method of single - well sedimentary microfacies based on an optimization model of the present invention is as follows: including the following steps,

[0066] S1: Divide the specified formation into lithologic combinations according to the value range of conventional curves.

[0067] S1.1: Calculate the shale classification value 80 of the natural gamma curve, and then classify each sampling point p in the target interval according to the natural gamma sampling value v(d) at depth d. If v(d)>80, the type of the sampling point p corresponding to depth d is a shale sampling point; otherwise, the type of the sampling point p is a non - shale sampling point.

[0068] S1.2: Combine all consecutive sampling points p with the same type in S1.1. If the type of the sampling point is a shale sampling point, combine them into a shale layer; if the type of the sampling point is a non - shale sampling point, combine them into a non - shale layer.

[0069] S1.3: Calculate the sandstone classification value 65 of the natural gamma curve, and then classify each sampling point in the non - shale layer interval according to the natural gamma sampling value v(d) at depth d. If v(d)>65, the type of the sampling point p corresponding to depth d is a mixed - rock sampling point; otherwise, the type of the sampling point p is a sandstone sampling point.

[0070] S1.4: Combine all consecutive sampling points p with the same type in S1.3. If the type of the sampling point is a mixed - rock sampling point, combine them into a mixed - rock layer; if the type of the sampling point is a sandstone sampling point, combine them into a sandstone layer.

[0071] S1.5: Traverse all layers from bottom to top in the deposition order. According to the ratio a of the current layer thickness to the thicknesses of the upper and lower adjacent layers, and the ratio b of the median value of the natural gamma value of the current layer to the median values of the natural gamma values of the upper and lower adjacent layers, calculate the thin layer index t = a 2 + a×b + b 2 . If t < 400, the current layer is an ultra-thin layer; otherwise, the current layer is not an ultra-thin layer.

[0072] S1.6: If none of the layers are ultra-thin layers, consider each layer as a lithologic combination, and then go to S2. Note: There is no requirement for the order of S1 and S2 here, and they can be carried out concurrently.

[0073] S1.7: If the current layer is an ultra-thin layer, correct the type of the current layer according to the types of its upper and lower adjacent layers. If neither the upper nor the lower adjacent layer is an ultra-thin layer and they have the same type, correct the type of the current ultra-thin layer to be the same as that of the upper and lower adjacent layers; if neither the upper nor the lower adjacent layer is an ultra-thin layer and they have different types, take the layer thickness and the median value of the natural gamma value as characteristic values, select the layer in the upper and lower adjacent layers that is most similar to the current ultra-thin layer, and correct the type of the former ultra-thin layer to be the type of that layer; if the upper and lower adjacent layers contain ultra-thin layers, then select the ultra-thin layer with the smallest thin layer index among consecutive ultra-thin layers to correct the type.

[0074] S1.8: Traverse all layers, merge consecutive layers of the same type together. If there are still ultra-thin layers, go to S1.5.

[0075] S2: Divide the specified formation into basic primitives according to the shape characteristics of the conventional curves.

[0076] S2.1: Traverse each sampling depth d in the target layer segment from bottom to top in the deposition order, and calculate the slope k(p) of the natural gamma curve at the depth d. Taking the slope error 3 as the threshold, if k(p) is less than -3, then p is a decreasing point; if -3 < k(p) < 3, then p is a stable point; if k(p) > 3, then p is an increasing point.

[0077] S2.2: Consecutive decreasing points form a decreasing primitive, consecutive stable points form a stable primitive, and consecutive increasing points form an increasing primitive. As Figure 5 shown in the "basic primitive" track in the schematic diagram of the basic primitive, where the red segment is the decreasing primitive, the blue segment is the stable primitive, and the green segment is the increasing primitive. In this embodiment, the decreasing primitive, the stable primitive, and the increasing primitive form the basic primitive, and there can also be other classification methods such as: {accelerated decreasing, uniform decreasing, decelerated decreasing, steady, decelerated increasing, uniform increasing, accelerated increasing}, {uniform decreasing, uniform steady, uniform increasing, non-uniform decreasing, non-uniform steady, non-uniform increasing}, etc.

[0078] S3: Construct a composite layer corresponding to the lithologic combination based on the basic primitive layer.

[0079] S3.1: Traverse all lithologic combinations from bottom to top according to the sedimentary sequence.

[0080] S3.2: For each lithologic combination, find all basic primitives that intersect with it in depth.

[0081] S3.3: Construct a composite layer from the set of basic primitives found in S3.2.

[0082] S4: Select a graphical model and solve the fitting result and fitting degree of the composite layer.

[0083] S4.1: Traverse all composite layers from bottom to top according to the sedimentary sequence.

[0084] S4.2: Select the function model y = kx + b where the image is a straight line as the graphical model. In the steps of this embodiment, a linear model is selected. In addition to the linear model, a periodic model, a logarithmic model, etc. can be selected. The selection of the model is determined according to the formation lithology characteristics.

[0085] S4.3: For each composite layer L, obtain k(L) corresponding to the fitting result of its y = kx + b, and calculate the variance δ(L) of the composite layer when k = k(L). Take the reciprocal of the variance δ(L) as the fitting degree of the composite layer L with respect to the straight line graphical model y = kx + b.

[0086] S5: If the matching degree of the composite layer is greater than or equal to the matching degree threshold set by the user, or the composite layer contains only one basic layer, then go to S7; otherwise, go to S6.

[0087] S6: Select feature points and split the composite layer.

[0088] S6.1: Traverse the boundaries between all adjacent basic primitives in the composite layer to obtain the depth points p(max) with the highest curvature at all boundaries.

[0089] S6.2: Split the composite layer into two new composite layers at the depth point p(max). For each new composite layer, execute S4 and iterate repeatedly until the matching degree of all composite layers meets the set matching degree threshold, or the composite layer contains only one basic layer; re-solve the fitting result and fitting degree of the composite layer.

[0090] S7: Merge adjacent layers that meet the complex graphics according to the graphical fitting result of the composite layer.

[0091] S7.1: Traverse the fitting slope k(L) of the function model y = kx + b corresponding to all composite layers from bottom to top.

[0092] S7.2: For the fitting results of continuously adjacent lithology composite layers from bottom to top, if they form combinations of k(L) < -1.5, -1.5 < k(L) < 1.5, and k(L) > 1.5, then merge them into box-shaped composite layers; if they form combinations of k(L) < -1.5 and k(L) > 1.5, then merge them into finger-shaped composite layers.

[0093] S7.3: Do not merge in other cases. If k(L) < -1.5, then the composite layer is a funnel-shaped composite layer; if k(L) > 1.5, then the composite layer is a bell-shaped composite layer.

[0094] S8: Construct sedimentary microfacies based on the depth range of the composite layer and determine the sedimentary microfacies type.

[0095] S8.1: Determine the depth range of the sedimentary microfacies based on the depth range of the composite layer.

[0096] S8.2: Based on the graphic type of the composite layer, such as funnel-shaped, bell-shaped, box-shaped, finger-shaped, etc., combined with the sedimentary subfacies type corresponding to the depth range, determine the type of sedimentary microfacies. For example, for the same finger-shaped, if the depth range is in the braided river delta front subfacies, the type of sedimentary microfacies is determined as underwater distributary channel; if the depth range is in the shore-shallow lake subfacies, the type of sedimentary microfacies is determined as shore-shallow lake sand.

[0097] S9: End.

[0098] Figure 2 It is a result map of dividing the specified formation into lithology combinations based on the values of conventional curves. The stratification basis in this map is the numerical statistics of the curves, which can better reflect the lithological characteristics of the formation. Existing various frameworks and methods achieve stratification in the same way.

[0099] Figure 3 It is a result map of dividing the specified formation into basic primitives based on the shape characteristics of conventional curves. The stratification basis in this map is the fitting result of the objective function of the curves, which can better reflect the morphological characteristics of the logging curves. The linear fitting results of the curves in each layer segment are plotted in the left track of the map.

[0100] Figure 4 It is an intermediate result map of constructing a composite layer based on basic primitives. There are three segments of fitting lines in the "core change rate" track in the map. The matching degree evaluation result of the uppermost fitting line is that further splitting is required, and the final splitting result is consistent with the stratification result of the "standard sedimentary microfacies" on the left. The matching degree evaluation results of the middle and lower fitting lines are that no splitting is required, and the effect of being consistent with the stratification result of the "standard sedimentary microfacies" on the left has been achieved.

[0101] Figure 5It is the division result of basic primitive elements, as shown in the "basic primitive elements" path in the schematic diagram, where the red segment is the decreasing primitive element, the blue segment is the stable primitive element, and the green segment is the increasing primitive element.

Claims

1. An automatic identification method for single - well sedimentary microfacies based on an optimization model, characterized in that, it includes the following processes, divide the target interval into lithologic combinations according to the lithologic characteristics of conventional logging curves; and divide the target interval into basic primitive units according to the shape characteristics of conventional logging curves by depth; Based on the basic primitive unit layer, construct a composite layer corresponding to the lithologic combination; Select the function corresponding to the desired function image as the objective function, obtain the fitting result of the composite layer for the objective function, and evaluate the matching degree between the composite layer and the fitting result; When the matching degree is not less than the threshold, or, when the composite layer contains only one basic layer; merge adjacent layers that meet the complex graph according to the graphic fitting result of the composite layer; Otherwise, select feature points, split the composite layer, and re - evaluate the matching degree between the composite layer and the fitting result; Construct sedimentary microfacies according to the depth range of the composite layer and determine the sedimentary microfacies type to complete the automatic identification of single - well sedimentary microfacies.

2. The automatic identification method for single - well sedimentary microfacies based on an optimization model according to claim 1, characterized in that, divide the specified formation into lithologic combinations according to the value range of the conventional curve, specifically including the following steps, Step 1.1, calculate the shale classification value of the natural gamma curve, and then classify each sampling point p in the target interval according to the natural gamma sampling value v(d) at depth d; if v(d)> shale classification value, then the type of the sampling point p corresponding to depth d is a shale sampling point; otherwise, the type of the sampling point p is a non - shale sampling point; Step 1.2, combine all consecutive sampling points p with the same type in Step 1.1 together. If the type of the sampling point is a shale sampling point, combine them into a shale layer; if the type of the sampling point is a non - shale sampling point, combine them into a non - shale layer; Step 1.3, calculate the sandstone classification value of the natural gamma curve, and then classify each sampling point in the non - shale layer section according to the natural gamma sampling value v(d) at depth d; if v(d)> sandstone classification value, then the type of the sampling point p corresponding to depth d is a mixed rock sampling point; otherwise, the type of the sampling point p is a sandstone sampling point; Step 1.4, combine all consecutive sampling points p with the same type in Step 1.3 together. If the type of the sampling point is a mixed rock sampling point, combine them into a mixed rock layer; if the type of the sampling point is a sandstone layer point, combine them into a sandstone layer; Step 1.5, traverse all layers from bottom to top in the sedimentary order, and calculate the thin - layer index t according to the ratio a of the current layer thickness to the thicknesses of the upper and lower adjacent layers and the ratio b of the median value of the natural gamma value of the current layer to the median values of the natural gamma values of the upper and lower adjacent layers; if t < 400, then the current layer is an ultra - thin layer; otherwise, the current layer is not an ultra - thin layer; Step 1.6, if none of the layers in Step 1.5 are ultra - thin layers, then regard each layer as a lithologic combination, and then go to the next step; Step 1.7, if the current layer is an ultra-thin layer, correct the type of the current layer according to the types of its upper and lower adjacent layers; if neither of the upper and lower adjacent layers is an ultra-thin layer and they have the same type, correct the type of the current ultra-thin layer to be the same as that of the upper and lower adjacent layers; if neither of the upper and lower adjacent layers is an ultra-thin layer and they have different types, take the median value of the layer thickness and natural gamma as the characteristic value, select the layer in the upper and lower adjacent layers that is most similar to the current ultra-thin layer, and correct the type of the previous ultra-thin layer to be the type of this layer; if the upper and lower adjacent layers contain ultra-thin layers, then select the ultra-thin layer with the smallest thin layer index among consecutive ultra-thin layers to correct the type; Step 1.8, traverse all layers, merge consecutive layers of the same type together, and if there are still ultra-thin layers, go to Step 1.

5.

3. A method for automatically identifying single-well sedimentary microfacies based on an optimization model according to claim 2, characterized in that, In step 1.5, the thin layer index t = a 2 + a × b + b 2 .

4. A method for automatically identifying single-well sedimentary microfacies based on an optimization model according to claim 1, characterized in that, divide the specified formation into basic primitive elements according to the shape characteristics of conventional curves, specifically including the following steps, Step 2.1, traverse each sampling depth d in the target interval from bottom to top in the sedimentation order, calculate the slope k(p) of the natural gamma curve at the depth d, if k(p) is less than -threshold, then p is a decreasing point; if -threshold < k(p) < threshold, then p is a stable point; if k(p) > threshold, then p is an increasing point; Step 2.2, consecutive decreasing points form a decreasing primitive element in the basic primitive elements, consecutive stable points form a stable primitive element in the basic primitive elements, and consecutive increasing points form an increasing primitive element in the basic primitive elements.

5. A method for automatically identifying single-well sedimentary microfacies based on an optimization model according to claim 1, characterized in that, construct a composite layer corresponding to the lithology combination based on the basic primitive element layer, specifically including the following steps, Step 3.1, traverse all lithology combinations from bottom to top in the sedimentation order; Step 3.2, for each lithology combination, find all basic primitive elements that have an intersection with it in depth; Step 3.3, construct a composite layer from the set of basic primitive elements found in Step 3.

2.

6. A method for automatically identifying single-well sedimentary microfacies based on an optimization model according to claim 1, characterized in that, Step 4: Select a graphical model, and solve the fitting result and fitting degree of the composite layer, specifically including the following steps, Step 4.1, traverse all composite layers from bottom to top in the sedimentation order; Step 4.2, select the function model y = kx + b with an image of a straight line as the graphical model; Step 4.3, for each composite layer L, obtain k(L) corresponding to the fitting result of its y = kx + b, and calculate the variance δ(L) of the composite layer when k = k(L); Take the reciprocal of the variance δ(L) as the fitting degree of the composite layer L with respect to the straight-line graphical model y = kx + b.

7. A method for automatically identifying single-well sedimentary microfacies based on an optimization model according to claim 1, characterized in that, select characteristic points and cut the composite layer, specifically including the following steps, Step 6.1, traverse the boundaries between all adjacent basic primitives in the composite layer, and obtain the depth point p(max) with the highest curvature at all boundaries; Step 6.2, at the depth point p(max), split the composite layer into two new composite layers. For each new composite layer, iterate repeatedly until all composite layers meet the requirements, and then continue to evaluate the matching degree between the composite layer and the fitting result.

8. A method for automatically identifying single-well sedimentary microfacies based on an optimization model according to claim 1, characterized in that, merging adjacent layers that meet the complex graph according to the graph fitting result of the composite layer, specifically including the following steps, Step 7.1, traverse the fitting slope k(L) of the function model y = kx + b corresponding to all composite layers from bottom to top; Step 7.2, for the fitting results of continuously adjacent composite layers of the same lithology from bottom to top, if they form a combination of k(L) < -1.5, -1.5 < k(L) < 1.5, k(L) > 1.5, then merge them into a box-shaped composite layer; if they form a combination of k(L) < -1.5, k(L) > 1.5, then merge them into a finger-shaped composite layer; Step 7.3, do not merge in other cases. If k(L) < -1.5, the composite layer is a funnel-shaped composite layer; if k(L) > 1.5, the composite layer is a bell-shaped composite layer.

9. A method for automatically identifying single-well sedimentary microfacies based on an optimization model according to claim 1, characterized in that, constructing sedimentary microfacies based on the depth range of the composite layer and determining the type of sedimentary microfacies, specifically including the following steps, Step 8.1, determine the depth range of the sedimentary microfacies according to the depth range of the composite layer; Step 8.2, determine the type of sedimentary microfacies according to the graph type of the composite layer and in combination with the type of sedimentary subfacies corresponding to the depth range.

Images