A method, device and storage medium for identifying "sweet spots" in shale oil and gas reservoirs

Abstract

The present invention relates to a method, device and storage medium for identifying "sweet spots" in shale oil and gas reservoirs, the method comprising: obtaining main controlling factors; determining classification categories according to the main controlling factors; constructing a sweet spot evaluation model based on random forests according to the main controlling factors and the classification categories, and determining sweet spot evaluation factors; determining favorable shale gas exploration areas according to the sweet spot evaluation factors. The present invention establishes a sweet spot evaluation model using a machine learning algorithm, which can complete the comprehensive evaluation of shale gas sweet spots in complex structural areas, optimize the enriched high-yield zones, and help efficient exploration and development.

Description

Technical Field

[0001] The present invention relates to the field of exploration and production technology, and in particular to a method, device and storage medium for identifying "sweet spots" in shale oil and gas reservoirs. Background Art

[0002] Shale gas is produced from organic-rich shales, and shale gas reservoirs are self-generating and self-storing continuous enriched gas reservoirs. Shale gas reservoirs have the characteristics of fine mineral particles and bedding structure, extremely low permeability, multi-scale pores, and unique gas storage and flow characteristics, which increase the complexity of the study of shale rock physical properties. Shale gas has the characteristics of self-generation, self-storage and self-protection, and has the characteristics of low production, no natural production capacity, and no water-gas boundary. At present, shale gas resources have been discovered in many places in China, and the development prospects are good. However, compared with North America, the formation and development of shale gas in China are obviously particular, and many exploration and development problems need to be solved urgently. The development of shale gas is an important resource guarantee, and the evaluation of the "sweet spots" of shale oil and gas reservoirs is an effective method to weigh whether shale formations have mining value.

[0003] In the existing technology, one is to evaluate the parameters of shale sweet spots based on geophysical prediction. This method has also been studied in depth. The geological sweet spot prediction technology includes total organic carbon content (TOC), pressure coefficient, gas content prediction, etc. The engineering sweet spot prediction technology includes brittleness index prediction, etc. The sweet spot evaluation system combining geology and engineering is still imperfect. It mainly adopts the prediction of a single sensitive parameter factor. Because a single factor sometimes produces some unpredictable problems, it is not conducive to judging the shale gas sweet spot; and another traditional prediction method is mainly to use rock physics, logging data, omnidirectional or wide-azimuth three-dimensional seismic data to comprehensively evaluate shale gas reservoirs and find sweet spots for shale gas exploration and development. Or based on the organic characteristics of the shale layer, the on-site gas content and the rock mechanics characteristics of the shale layer, the sweet spots at different levels of the shale gas well are comprehensively evaluated. However, due to the complex process of studying and identifying the shale sweet spot layer in the existing technology and the low accuracy, it is not conducive to achieving high shale gas production.

[0004] In summary, the existing evaluation methods for the "sweet spots" of shale oil and gas reservoirs have the problems of inaccurate evaluation and complicated processes. At present, there is still a lack of accurate and effective methods for identifying the "sweet spots" of shale oil and gas reservoirs. Therefore, how to efficiently identify and predict the "sweet spots" of shale oil and gas reservoirs is an urgent problem to be solved. Summary of the invention

[0005] In view of this, it is necessary to provide a method, device and storage medium for identifying the "sweet spots" of shale oil and gas reservoirs, so as to overcome the problem in the prior art that it is difficult to efficiently identify and predict the "sweet spots" of shale oil and gas reservoirs.

[0006] In order to solve the above technical problems, the present invention provides a method for identifying "sweet spots" in shale oil and gas reservoirs, comprising:

[0007] Get the master control factors;

[0008] Determine the classification category based on the main controlling factors;

[0009] Based on random forest, according to the main control factors and the classification categories, a dessert evaluation model is constructed, and dessert evaluation factors are determined;

[0010] Based on the sweet spot evaluation factor, favorable exploration areas for shale gas are determined.

[0011] Furthermore, the obtaining of the main control factors includes:

[0012] Acquire core data, seismic data, and well logging data;

[0013] Performing Pearson correlation analysis on all the data in the core data, the seismic data and the well logging data to calculate the correlation coefficient;

[0014] Data that meets the preset range of the correlation coefficient is selected as the main control factor.

[0015] Furthermore, determining the classification category according to the main control factor includes:

[0016] According to the main controlling factor, the classification category corresponding to the main controlling factor is determined, wherein the classification category includes an organic matter quality characteristic category, a reservoir quality characteristic category and a completion quality characteristic category, the organic matter quality characteristic category includes at least one of total organic carbon content, maturity, and storage capacity, the reservoir quality characteristic category includes at least one of porosity, density, and thickness, and the completion quality category includes at least one of clay content, Poisson's ratio, Young's modulus, brittleness, and pressure.

[0017] Furthermore, the sweet spot evaluation model includes an organic matter quality model, the sweet spot evaluation factor includes an organic matter influencing factor, and the sweet spot evaluation model is constructed based on the random forest according to the main controlling factors and the classification categories, including:

[0018] According to the main controlling factors belonging to the organic matter quality characteristic category, the organic matter quality training set is constructed;

[0019] K are extracted from the organic matter quality training set 1 organic matter sample sets, wherein the capacity of each organic matter sample set is a preset capacity;

[0020] For K 1 organic matter sample sets, and establish K 1 decision tree models, corresponding to K 1 decision results;

[0021] According to K 1 The decision results are voted on to determine the final organic matter impact factor.

[0022] Furthermore, the sweet spot evaluation model includes a reservoir quality model, the sweet spot evaluation factors include reservoir quality influencing factors, and the sweet spot evaluation model is constructed based on the random forest according to the main controlling factors and the classification categories, including:

[0023] Constructing a reservoir quality training set based on the main controlling factors belonging to the reservoir quality characteristic category;

[0024] K is extracted from the reservoir quality training set 2 reservoir quality sample sets, wherein the capacity of each reservoir quality sample set is a preset capacity;

[0025] For K 2 reservoir quality sample sets, and establish K 2 decision tree models, corresponding to K 2 decision results;

[0026] According to K 2 The decision results are voted on to determine the final reservoir quality influencing factors.

[0027] Further, the sweet spot evaluation model includes a completion quality model, the sweet spot evaluation factors include completion quality influencing factors, and the sweet spot evaluation model is constructed based on the random forest according to the main control factors and the classification categories, including:

[0028] According to the main control factors belonging to the completion quality feature category, a completion quality training set is constructed;

[0029] K is extracted from the completion quality training set 3 organic matter sample sets, where the capacity of each completion quality sample set is a preset capacity;

[0030] For K 3 completion quality sample sets, and establish K 3 decision tree models, corresponding to K 3 decision results;

[0031] According to K 3 The decision results are voted on to determine the final completion quality influencing factors.

[0032] Furthermore, the sweet spot evaluation factors include organic matter influencing factors, reservoir quality influencing factors and completion quality influencing factors. The determination of favorable shale gas exploration areas based on the sweet spot evaluation factors includes:

[0033] Sorting the organic matter influencing factors screened out by the organic matter quality model according to sensitivity to determine a first sorting result;

[0034] Sorting the reservoir quality influencing factors screened out by the reservoir quality model according to sensitivity to determine a second sorting result;

[0035] Sorting the completion quality influencing factors screened out by the completion quality model according to sensitivity to determine a third sorting result;

[0036] Based on the first sorting result, the second sorting result and the third sorting result, favorable shale gas exploration areas are determined.

[0037] Furthermore, the process of establishing the dessert evaluation model includes:

[0038] According to the main control factors belonging to the classification category, a sample set is formed, wherein the label corresponding to the main control factor is a dessert type;

[0039] For a single decision tree, samples are extracted from the sample set to form at least one training set and a test set;

[0040] The node is split from the root node. The two child nodes formed after the split are measured by Gini coefficient or information gain. The number of M features selected during the split is randomly selected from the M features of the training sample. According to the calculated information gain or Gini coefficient, the most appropriate feature is selected from the M features according to the splitting principle for splitting.

[0041] Each node is split according to the splitting method of the root node until it cannot be split anymore, forming a leaf node;

[0042] A random forest is formed for a plurality of decision trees that have been constructed, and the dessert evaluation model is established.

[0043] The present invention also provides a device for identifying "sweet spots" in shale oil and gas reservoirs, comprising:

[0044] An acquisition unit, used for acquiring the main control factor;

[0045] A processing unit, configured to determine a classification category according to the main control factor; and to construct a sweet spot evaluation model based on a random forest and according to the main control factor and the classification category, and to determine a sweet spot evaluation factor;

[0046] The prediction unit is used to determine the favorable exploration area of ​​shale gas according to the sweet spot evaluation factor.

[0047] The present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method for identifying "sweet spots" in shale oil and gas reservoirs as described above.

[0048] Compared with the prior art, the beneficial effects of the present invention include: first, effectively obtaining the main controlling factors; then, effectively classifying the main controlling factors to determine which category the main controlling factors belong to; then, using random forests, modeling the main controlling factors of different classification categories, determining the sweet spot evaluation model, and thus determining the sweet spot evaluation factor in the main controlling factors; finally, based on the sweet spot evaluation factor, effectively exploring the favorable exploration area of ​​shale gas. In summary, the present invention is a new method improved on the basis of using a single sensitive parameter to predict the key parameters of shale gas sweet spots. It considers factors more comprehensively and comprehensively considers various sensitive parameters, avoiding the problem of similar results when a single sensitive parameter predicts different sweet spot key parameters, making the prediction more accurate and generalizable. By using a machine learning algorithm to establish a sweet spot evaluation model, the comprehensive evaluation of shale gas sweet spots in complex structural areas can be completed. The process is simple, and the complex areas can be predicted through model establishment. It has strong practicality and a wide range of application scenarios. It selects enriched high-yield zones and helps efficient exploration and development. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 A schematic flow chart of an embodiment of a method for identifying "sweet spots" in shale oil and gas reservoirs provided by the present invention;

[0050] Figure 2 The present invention provides Figure 1 A schematic diagram of a flow chart of an embodiment of step S101;

[0051] Figure 3 The present invention provides Figure 1 A schematic diagram of a flow chart of an embodiment of step S103;

[0052] Figure 4 The present invention provides Figure 1 A flow chart of another embodiment of step S103;

[0053] Figure 5 The present invention provides Figure 1 A flowchart of another embodiment of step S103;

[0054] Figure 6 The present invention provides Figure 1 A schematic diagram of a flow chart of an embodiment of step S104;

[0055] Figure 7 A schematic diagram of a flow chart of an embodiment of establishing a dessert evaluation model provided by the present invention;

[0056] Figure 8 A schematic diagram of establishing a model according to an embodiment of the random forest algorithm provided by the present invention;

[0057] Fig. 9 A schematic diagram of the reasoning chain of an embodiment of the normalization of three-quality data provided by the present invention;

[0058] Fig.10 A schematic diagram of the structure of an embodiment of a device for identifying "sweet spots" in shale oil and gas reservoirs provided by the present invention;

[0059] Fig.11 A schematic structural diagram of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0060] The preferred embodiments of the present invention are described in detail below in conjunction with the accompanying drawings, wherein the accompanying drawings constitute a part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, but are not used to limit the scope of the present invention.

[0061] In the description of the present invention, the terms "first" and "second" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In addition, the meaning of "plurality" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0062] In the description of the present invention, reference to an "embodiment" means that a particular feature, structure, or characteristic described in conjunction with the embodiment may be included in at least one embodiment of the present invention. The appearance of the phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive of other embodiments. It is explicitly and implicitly understood by those skilled in the art that the described embodiments may be combined with other embodiments.

[0063] The present invention provides a method, device and storage medium for identifying the "sweet spots" of shale oil and gas reservoirs, explores the prediction of different key parameters of sweet spots by multiple sensitive parameters, and comprehensively evaluates shale gas sweet spots in structurally complex areas, providing new ideas for further improving the accuracy and efficiency of identifying the "sweet spots" of shale oil and gas reservoirs.

[0064] Before describing the embodiments, the following definitions are given for the relevant terms:

[0065] Shale gas sweet spot: refers to the area with relatively high commercial production rate in a shale gas reservoir exploration area;

[0066] Organic matter quality: refers to the parameters related to the various animal and plant residues and microorganisms and their decomposition and synthesis of organic matter contained in a unit volume of soil, including total organic carbon content (TOC), maturity, kerogen type, reservoir capacity, etc.;

[0067] Reservoir quality: timely evaluation of the possibility, characteristics, reserves, quality, etc. of oil and gas reservoir formation, including porosity, permeability, saturation, density, thickness, etc.;

[0068] Completion quality: After the open hole well reaches the designed depth, the bottom of the well and the oil layer are connected in a certain structure, and the relevant parameters of the generated structure are evaluated, including clay content, Poisson's ratio, Young's modulus, brittleness, pressure, etc.

[0069] Based on the description of the above technical terms, in the prior art, predictions are often made using a single sensitive parameter factor, or a comprehensive evaluation is performed in combination with rock physics, logging data, and three-dimensional seismic data. The above methods have the problems of inaccurate evaluation and complicated processes, and it is difficult to effectively predict shale gas sweet spots. Therefore, the present invention aims to propose a method for efficiently and accurately identifying the "sweet spots" of shale oil and gas reservoirs.

[0070] The specific embodiments are described in detail below:

[0071] The embodiment of the present invention provides a method for identifying the "sweet spots" of shale oil and gas reservoirs. Figure 1 Come and see, Figure 1 A flow chart of an embodiment of a method for identifying "sweet spots" in shale oil and gas reservoirs provided by the present invention includes steps S101 to S104, wherein:

[0072] In step S101, the main control factor is obtained;

[0073] In step S102, a classification category is determined according to the main control factor;

[0074] In step S103, based on random forest, a dessert evaluation model is constructed according to the main control factors and the classification categories, and a dessert evaluation factor is determined;

[0075] In step S104, a favorable exploration area for shale gas is determined according to the sweet spot evaluation factor.

[0076] In an embodiment of the present invention, first, the main controlling factors are effectively acquired; then, the main controlling factors are effectively classified to determine to which category the main controlling factors belong; then, random forests are used to model the main controlling factors of different classification categories, and the sweet spot evaluation model is determined, thereby determining the sweet spot evaluation factors in the main controlling factors; finally, based on the sweet spot evaluation factors, the favorable exploration areas for shale gas are effectively explored.

[0077] It should be noted that the technical idea of ​​the present invention is to use correlation analysis to screen out the main controlling factors of shale gas enrichment and accumulation from core data, logging data and seismic data; based on the main controlling factors selected by correlation analysis, the indicators affecting the "sweet spot" are divided into three categories: organic quality, reservoir quality and completion quality, and the classification results are obtained; the classification results obtained by the three qualities are used to establish a shale sweet spot evaluation model M using the random forest algorithm in machine learning; based on the evaluation model M obtained by the random forest algorithm, the shale sweet spot evaluation factor Ms is determined; and the favorable shale gas exploration area is determined according to the evaluation factor Ms. The present invention can complete the comprehensive evaluation of shale sweet spots in complex structural areas by using machine learning to establish a sweet spot evaluation model, optimize the enriched high-yield belt, and help efficient exploration and development.

[0078] As a preferred embodiment, Figure 2 Come and see, Figure 2 The present invention provides Figure 1 The flowchart of step S101 in an embodiment of the present invention is shown in FIG. 1 , wherein step S101 includes steps S201 to S203, wherein:

[0079] In step S201, core data, seismic data and well logging data are obtained;

[0080] In step S202, a Pearson correlation analysis is performed on all the data in the core data, the seismic data and the well logging data to calculate a correlation coefficient;

[0081] In step S203, data satisfying a preset range of correlation coefficients are selected as the main control factors.

[0082] In the embodiment of the present invention, the Pearson coefficient is used to perform Pearson correlation analysis on all data in the core data, seismic data and logging data, wherein the smaller the correlation, the easier it is to distinguish. Because we need to select features with little correlation for research, the features with a cutoff between [-0.5, 0.5] are finally selected based on the value calculated by Pearson.

[0083] As a preferred embodiment, the above step S102 includes:

[0084] According to the main controlling factor, the classification category corresponding to the main controlling factor is determined, wherein the classification category includes an organic matter quality characteristic category, a reservoir quality characteristic category and a completion quality characteristic category, the organic matter quality characteristic category includes at least one of total organic carbon content, maturity, and storage capacity, the reservoir quality characteristic category includes at least one of porosity, density, and thickness, and the completion quality category includes at least one of clay content, Poisson's ratio, Young's modulus, brittleness, and pressure.

[0085] In the embodiment of the present invention, the main controlling factors are effectively classified, and the indicators affecting the "sweet spot" are divided into three categories: organic matter quality includes: total organic carbon content (TOC), maturity, kerogen type, and reservoir capacity; reservoir quality includes: porosity, permeability, saturation, density, and thickness; completion quality includes: clay content, Poisson's ratio, Young's modulus, brittleness, compressive strength, stress field, and pressure. We classified all the selected features and determined 11 features, among which organic matter quality includes: total organic carbon content, maturity, and storage capacity; reservoir quality includes: porosity, density, and thickness; completion quality includes: clay content, Poisson's ratio, Young's modulus, brittleness, and pressure.

[0086] As a preferred embodiment, the sweet spot evaluation model includes an organic matter quality model, and the sweet spot evaluation factor includes an organic matter impact factor. Figure 3 Come and see, Figure 3 The present invention provides Figure 1 The flowchart of step S103 in an embodiment of the present invention is shown in FIG. 1 , wherein step S103 includes steps S301 to S304, wherein:

[0087] In step S301, an organic matter quality training set is constructed according to the main controlling factors belonging to the organic matter quality feature category;

[0088] In step S302, K is extracted from the organic matter quality training set. 1 organic matter sample sets, wherein the capacity of each organic matter sample set is a preset capacity;

[0089] In step S303, for K 1 organic matter sample sets, and establish K 1 decision tree models, corresponding to K 1 decision results;

[0090] In step S304, according to K 1 The decision results are voted on to determine the final organic matter impact factor.

[0091] In the embodiment of the present invention, for the main controlling factors selected belonging to the organic matter quality characteristic category, the random forest algorithm in machine learning is used to establish an organic matter quality model.

[0092] As a preferred embodiment, the sweet spot evaluation model includes a reservoir quality model, and the sweet spot evaluation factor includes a reservoir quality influencing factor. Figure 4 Come and see, Figure 4 The present invention provides Figure 1 The flowchart of another embodiment of step S103 is shown in FIG. 1 , wherein step S103 includes steps S401 to S404, wherein:

[0093] In step S401, a reservoir quality training set is constructed according to the main controlling factors belonging to the reservoir quality characteristic category;

[0094] In step S402, K is extracted from the reservoir quality training set. 2 reservoir quality sample sets, wherein the capacity of each reservoir quality sample set is a preset capacity;

[0095] In step S403, for K 2 reservoir quality sample sets, and establish K 2 decision tree models, corresponding to K 2 decision results;

[0096] In step S404, according to K 2 The decision results are voted on to determine the final reservoir quality influencing factors.

[0097] In the embodiment of the present invention, for the main controlling factors selected belonging to the reservoir quality characteristic category, a reservoir quality model is established using the random forest algorithm in machine learning.

[0098] As a preferred embodiment, the sweet spot evaluation model includes a completion quality model, and the sweet spot evaluation factor includes a completion quality influencing factor. Figure 5 Come and see, Figure 5 The present invention provides Figure 1 The flowchart of another embodiment of step S103 is shown in FIG. 1 , wherein step S103 includes steps S501 to S504, wherein:

[0099] In step S501, a completion quality training set is constructed according to the main control factors belonging to the completion quality feature category;

[0100] In step S502, K are extracted from the completion quality training set. 3 organic matter sample sets, where the capacity of each completion quality sample set is a preset capacity;

[0101] In step S503, for K 3 completion quality sample sets, and establish K 3 decision tree models, corresponding to K3 decision results;

[0102] In step S504, according to K 3 The decision results are voted on to determine the final completion quality influencing factors.

[0103] In the embodiment of the present invention, for the main control factors selected belonging to the category of completion quality characteristics, a completion quality model is established using the random forest algorithm in machine learning.

[0104] In a specific embodiment of the present invention, the selected features are modeled using the random forest algorithm of machine learning according to organic matter quality, reservoir quality and completion quality. The basic process of random forest modeling of organic matter quality is as follows: First, K is extracted from the organic matter quality training set using bootstrap sampling. 1 samples, each of which has the same sample size as the original training set; then, 1 For each sample, k1 decision tree models are established to obtain K 1 Results; Finally, according to K 1 The results are voted for each record to determine its final important influencing factors. The basic process of random forest modeling of reservoir quality: First, bootstrap sampling is used to extract K from the organic matter quality training set. 2 samples, each of which has the same sample size as the original training set; then, 2 Each sample establishes K 2 decision tree model, and get K 2 Results; Finally, according to K 2 The basic process of random forest modeling of completion quality is as follows: First, bootstrap sampling is used to extract K from the organic matter quality training set. 3 samples, each of which has the same sample size as the original training set; then, 3 Each sample establishes K 3 decision tree model, and get K 3 Results; Finally, according to K 3 The results are voted on for each record to determine its final important influencing factors. Input: three-quality data, output: shale gas sweet spot prediction results. The training sample is the selected formation data, and the label is whether the formation is type I, type II, or type III sweet spot.

[0105] It should be noted that Bootstrap, also known as self-bootstrapping, bootstrap, self-help, and bootstrap, is a resampling technique in statistical learning, used to estimate standard error, confidence interval, and deviation. Bootstrap is a popular statistical method in modern statistics, and it works well when the sample size is small. Methods such as Bagging and AdaBoost in machine learning actually contain the idea of ​​Boostrap. In the field of ensemble learning, Bootstrap directly derives the Bagging model.

[0106] As a preferred embodiment, the sweet spot evaluation factors include organic matter influencing factors, reservoir quality influencing factors and completion quality influencing factors. Figure 6 Come and see, Figure 6 The present invention provides Figure 1 The flowchart of step S104 in an embodiment of the present invention is shown in FIG. 1 , wherein step S104 includes steps S601 to S604, wherein:

[0107] In step S601, the organic matter influencing factors screened out by the organic matter quality model are sorted according to sensitivity to determine a first sorting result;

[0108] In step S602, the reservoir quality influencing factors selected by the reservoir quality model are sorted according to sensitivity to determine a second sorting result;

[0109] In step S603, the completion quality influencing factors screened out by the completion quality model are sorted according to sensitivity to determine a third sorting result;

[0110] In step S604, favorable shale gas exploration areas are determined based on the first sorting result, the second sorting result and the third sorting result.

[0111] In the embodiments of the present invention, a method for selecting sensitive parameters of shale gas sweet spot key parameters TOC, gas content, and porosity is introduced, which is highly targeted. In the study, the selected sensitive parameters are sorted according to sensitivity, and the relationship between the key parameters of shale gas sweet spot and the sensitive parameters is fitted by multivariate statistical analysis method, with high similarity.

[0112] In a specific embodiment of the present invention, the selected evaluation factors Ms are ranked according to sensitivity to help determine favorable shale gas exploration areas. For example, the final ranking of the evaluation factors of several wells in a certain area, organic matter quality: σ 总有机碳含量 >σ 储存容量 >σ 成熟度 ; Reservoir quality:

[0113] σ 孔隙度 >σ厚度 >σ 密度 ; Completion quality: σ 压力 >σ 脆度 >σ 泊松比 >σ 杨氏模量 >σ 粘土含量 .

[0114] As a preferred embodiment, Figure 7 Come and see, Figure 7 This is a flow chart of an embodiment of establishing a dessert evaluation model provided by the present invention. The process of establishing the dessert evaluation model includes steps S701 to S704, wherein:

[0115] In step S701, a sample set is formed according to the main control factors belonging to the classification category, wherein the label corresponding to the main control factors is the dessert type;

[0116] In step S702, for a single decision tree, samples are extracted from the sample set to form at least one training set and a test set;

[0117] In step S703, node splitting is performed starting from the root node, wherein the two child nodes formed after the splitting are measured by the Gini coefficient or information gain, and the number of M features selected during the splitting is randomly selected from the M features of the training sample, and the most appropriate feature is selected from the M features according to the splitting principle according to the calculated information gain or Gini coefficient for splitting;

[0118] In step S704, each node is split according to the splitting method of the root node until it cannot be split anymore, forming a leaf node;

[0119] In step S705, a random forest is formed for the multiple decision trees that have been constructed, and the dessert evaluation model is established.

[0120] In an embodiment of the present invention, the data of the three qualities are divided into a training set and a test set, and then a machine learning model (random forest model M) is used for training, and finally the optimal parameter Ms is obtained by voting by multiple decision trees. The evaluation factor Ms helps to determine favorable shale gas exploration areas.

[0121] In a specific embodiment of the present invention, Figure 8 Come and see, Figure 8 This is a schematic diagram of an embodiment of establishing a model of the random forest algorithm provided by the present invention. The random forest modeling process is specifically as follows:

[0122] The first step is to build a single decision tree as follows:

[0123] (1) Select N samples from the sample set through the Bagging ensemble method to form a training set.

[0124] (2) Starting from the root node, the node is split. The principle of splitting is that the purity of the two child nodes after the split can be improved as much as possible. Generally, the Gini coefficient or information gain is used for measurement. The number of features m selected during the split is randomly selected from the M features of the training sample. According to the calculated information gain or Gini coefficient, the most appropriate feature is selected from these m features according to the splitting principle for splitting.

[0125] (3) Each node is split according to step (2) until it cannot be split any more, thus forming a leaf node.

[0126] In the second step, each decision tree follows the above steps to eventually form a forest. The method in the final prediction stage is the decision model of the random forest model.

[0127] The following is a specific application scenario to better illustrate the technical solution of the present invention. The process of identifying the "sweet spots" of shale oil and gas reservoirs is as follows:

[0128] Step 1: Use correlation analysis to screen out the main controlling factors of shale gas enrichment and accumulation from core data, well logging data and seismic data;

[0129] Step 2: Based on the main controlling factors selected by correlation analysis, the indicators affecting the "sweet spot" are divided into three categories: organic quality, reservoir quality and completion quality, and the classification results are obtained;

[0130] Step 3: Using the classification results obtained by the three qualities, the random forest algorithm in machine learning is used to establish a shale gas sweet spot evaluation model M;

[0131] Step 4: Based on the evaluation model M obtained by the random forest algorithm, determine the shale gas sweet spot evaluation factor Ms;

[0132] Step 5: Use the evaluation factor Ms to help determine favorable shale gas exploration areas.

[0133] Among them, from step 1 to step 2, the input data is analyzed for correlation. When the correlation does not reach the specified threshold, the impurities are removed; when the correlation reaches the required threshold, it is retained and judged whether it is organic quality. If it is, it is classified as organic quality. If not, it is judged whether it is reservoir quality. If it is, it is classified as reservoir quality. If not, it is judged whether it is completion quality. If it is, it is retained. If not, it is deleted. (Here, the judgment is that we originally have our own specified corresponding features, which are organic quality, reservoir quality and completion quality).

[0134] Among them, combined Fig. 9 Come and see, Fig. 9 The schematic diagram of the reasoning chain of an embodiment of the normalization of three-quality data provided by the present invention. The main features used in this patent are as follows:

[0135] Organic quality: Total Organic Carbon, maturity, kerogen type, storage capacity;

[0136] Reservoir quality: porosity, permeability, saturation, density, thickness;

[0137] Completion quality: clay quality, Poisson's ratio, Young's modulus, brittleness, compressive strength, stress field, pressure. Normalize the features to get the inference chain as follows Figure 3 As shown, A to P represent what are shown in Table 1.

[0138] The physical meaning is described as follows:

[0139] Total organic carbon content TOC (A): It is the total amount of organic matter in water expressed as carbon content, and the result is expressed as mass concentration of carbon (C) (mg / L);

[0140] Maturity (B): Organic matter maturity can be used to measure the degree of evolution of organic matter, determine the thermal maturity stage of organic matter, and evaluate the hydrocarbon generation capacity and reservoir potential of shale;

[0141] Kerogen type (C): Different types of kerogen have different hydrocarbon generation capacities and amounts. Per unit volume, type I kerogen has greater hydrocarbon generation capacity and amount than type II and III kerogens.

[0142] Reservoir capacity (D): Capacity also refers to the number of unit objects that the shale gas in the area can accommodate;

[0143] Porosity (E): Porosity refers to the ratio of the sum of the volumes of all pore spaces in a rock sample to the volume of the rock sample, which is called the total porosity of the rock and is expressed as a percentage;

[0144] Permeability (F): Permeability refers to the ability of rock to allow fluid to pass through under a certain pressure difference. It is a parameter that characterizes the ability of soil or rock itself to conduct liquid;

[0145] Saturation (G): calculated using the Alzi formula;

[0146] Density (H): Density is a measure of mass within a specific volume. Density is equal to the mass of an object divided by its volume. It can be represented by the symbol ρ (pronounced róu). In the International System of Units and Chinese legal measurement units, the unit of density is kilograms per cubic meter, and the symbol is kg / m3.

[0147] Thickness (I): The scale of high-quality mudstone is the basis for obtaining high-yield shale gas. Thick mudstone is more conducive to shale gas enrichment and can improve the sealing ability of the shale layer and enhance its preservation capacity.

[0148] Clay quality (J): the quality of clay beneath the formation;

[0149] Poisson's ratio (K): Poisson's ratio refers to the ratio of lateral strain to longitudinal strain;

[0150] Young's modulus (L): Young's modulus is the ratio of longitudinal stress to longitudinal strain;

[0151] Brittleness (M): High brittle mineral content is the basis for natural fractures and subsequent development of hydraulic fractures;

[0152] Compressive strength (N): calculated by Kerricart;

[0153] Stress field (O): The prerequisite for shale gas exploration and development is mainly the geological conditions under stress;

[0154] Pressure (P): The main causes of pressure are undercompaction of mudstone, dehydration of clay minerals, hydrocarbon generation from organic matter, thermal expansion of fluids, and tectonic compression.

[0155] Specifically, step 4 is as follows: divide the three qualities of data into training set and test set, and then use the machine learning model (random forest model M) for training. Finally, multiple decision trees vote to obtain the optimal parameter Ms. The evaluation factor Ms helps determine favorable shale gas exploration areas.

[0156] The embodiment of the present invention also provides a device for identifying the "sweet spots" of shale oil and gas reservoirs. Fig.10 Come and see, Fig.10 The schematic diagram of the structure of an embodiment of a device for identifying a "sweet spot" in a shale oil and gas reservoir provided by the present invention is as follows. The device 1000 for identifying a "sweet spot" in a shale oil and gas reservoir comprises:

[0157] An acquisition unit 1001 is used to acquire a main control factor;

[0158] The processing unit 1002 is used to determine the classification category according to the main control factor; to construct a sweet spot evaluation model based on the random forest according to the main control factor and the classification category, and to determine the sweet spot evaluation factor;

[0159] The prediction unit 1003 is used to determine favorable exploration areas for shale gas according to the sweet spot evaluation factor.

[0160] The more specific implementation methods of each unit of the device for identifying the "sweet spots" of shale oil and gas reservoirs can be found in the description of the above-mentioned method for identifying the "sweet spots" of shale oil and gas reservoirs, and have similar beneficial effects, which will not be repeated here.

[0161] An embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon. When the program is executed by a processor, the method for identifying "sweet spots" in shale oil and gas reservoirs as described above is implemented.

[0162] Generally speaking, the computer instructions for implementing the method of the present invention can be carried by any combination of one or more computer-readable storage media. Non-transitory computer-readable storage media can include any computer-readable media except the signal itself that is temporarily propagating.

[0163] Computer-readable storage media can be, for example, but not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or devices, or any combination thereof. More specific examples of computer-readable storage media (a non-exhaustive list) include: an electrical connection with one or more conductors, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this document, a computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, device, or device.

[0164] Computer program code for performing the operation of the present invention can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as "C" language or similar programming languages, in particular, Python language suitable for neural network computing and platform frameworks based on TensorFlow, PyTorch, etc. can be used. The program code can be executed entirely on the user's computer, partially on the user's computer, as an independent software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (for example, using an Internet service provider to connect through the Internet).

[0165] The embodiment of the present invention further provides an electronic device, Fig.11 Come and see, Fig.11 This is a schematic diagram of the structure of an embodiment of an electronic device provided by the present invention. The electronic device 1100 includes a processor 1101, a memory 1102, and a computer program stored in the memory 1102 and executable on the processor 1101. When the processor 1101 executes the program, the method for identifying "sweet spots" in shale oil and gas reservoirs as described above is implemented.

[0166] As a preferred embodiment, the electronic device 1100 further includes a display 1103 for displaying the method of identifying "sweet spots" of shale oil and gas reservoirs executed by the processor 1101 as described above.

[0167] Exemplarily, the computer program may be divided into one or more modules / units, one or more modules / units are stored in the memory 1102, and are executed by the processor 1101 to complete the present invention. One or more modules / units may be a series of computer program instruction segments that can complete specific functions, and the instruction segments are used to describe the execution process of the computer program in the electronic device 1100. For example, the computer program may be divided into the acquisition unit 1001, the processing unit 1002, and the prediction unit 1003 in the above embodiment, and the specific functions of each unit are as described above, and are not repeated here one by one.

[0168] The electronic device 1100 may be a desktop computer, a notebook, a PDA, a smart phone or the like with an adjustable camera module.

[0169] Among them, the processor 1101 may be an integrated circuit chip with signal processing capabilities. The above-mentioned processor 1101 may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The disclosed methods, steps and logic block diagrams in the embodiments of the present invention can be implemented or executed. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0170] The memory 1102 may be, but is not limited to, a random access memory (RAM), a read only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electric erasable programmable read-only memory (EEPROM), etc. The memory 1102 is used to store programs, and the processor 1101 executes the program after receiving the execution instruction. The process definition method disclosed in any of the embodiments of the present invention may be applied to the processor 1101 or implemented by the processor 1101.

[0171] The display 1103 may be an LCD display screen or an LED display screen, for example, a display screen on a mobile phone.

[0172] Understandably, Fig.11 The structure shown is only a schematic diagram of the structure of the electronic device 1100. The electronic device 1100 may also include Fig.11 More or fewer components as shown. Fig.11 Each component shown in the figure can be implemented by hardware, software or a combination thereof.

[0173] The computer-readable storage medium and electronic device provided according to the above embodiments of the present invention can be implemented with reference to the specific description of the method for identifying the "sweet spots" of shale oil and gas reservoirs as described above according to the present invention, and have similar beneficial effects as the method for identifying the "sweet spots" of shale oil and gas reservoirs as described above, which will not be repeated here.

[0174] The present invention discloses a method, device and storage medium for identifying "sweet spots" in shale oil and gas reservoirs. First, the main controlling factors are effectively acquired; then, the main controlling factors are effectively classified to determine which category the main controlling factors belong to; then, random forests are used to model the main controlling factors of different classification categories, and a sweet spot evaluation model is determined, thereby determining the sweet spot evaluation factor in the main controlling factors; finally, based on the sweet spot evaluation factor, the favorable exploration area for shale gas is effectively explored.

[0175] The technical solution of the present invention is a new method improved on the basis of using a single sensitive parameter to predict the key parameters of shale gas sweet spots. It takes more comprehensive factors into consideration and can avoid the problem of similar results when a single sensitive parameter is used to predict different key parameters of sweet spots. It is generalizable and can establish a sweet spot evaluation model by using a machine learning algorithm. It can complete the comprehensive evaluation of shale gas sweet spots in complex structural areas, optimize enriched and high-yield zones, and help efficient exploration and development.

[0176] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by any technician familiar with the technical field within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention.

Claims

1. A method for identifying "sweet spots" in shale oil and gas reservoirs, characterized in that: include: Get the master control factors; Determine the classification category based on the main controlling factors; Based on random forest, according to the main control factors and the classification categories, a dessert evaluation model is constructed, and dessert evaluation factors are determined; Determining favorable shale gas exploration areas according to the sweet spot evaluation factors; The sweet spot evaluation model includes an organic matter quality model, the sweet spot evaluation factor includes an organic matter influencing factor, and the sweet spot evaluation model is constructed based on the random forest according to the main controlling factors and classification categories, including: According to the main controlling factors belonging to the organic matter quality characteristic category, the organic matter quality training set is constructed; Extracting K1 organic matter sample sets from the organic matter quality training set, wherein the capacity of each organic matter sample set is a preset capacity; For K1 organic matter sample sets, K1 decision tree models are established respectively, and K1 decision results are obtained accordingly; Voting is performed based on K1 decision results to determine the final organic matter impact factor; The sweet spot evaluation model includes a reservoir quality model, the sweet spot evaluation factors include reservoir quality influencing factors, and the sweet spot evaluation model is constructed based on random forest according to the main control factors and the classification categories, including: Constructing a reservoir quality training set based on the main controlling factors belonging to the reservoir quality characteristic category; Extracting K2 reservoir quality sample sets from the reservoir quality training set, wherein the capacity of each reservoir quality sample set is a preset capacity; For K2 reservoir quality sample sets, K2 decision tree models are established respectively, and K2 decision results are obtained accordingly; Voting is performed based on K2 decision results to determine the final reservoir quality influencing factor; The sweet spot evaluation model includes a completion quality model, the sweet spot evaluation factors include completion quality influencing factors, and the sweet spot evaluation model is constructed based on random forest according to the main control factors and the classification categories, including: According to the main control factors belonging to the completion quality feature category, a completion quality training set is constructed; Extracting K3 organic matter sample sets from the completion quality training set, wherein the capacity of each completion quality sample set is a preset capacity; For K3 completion quality sample sets, K3 decision tree models are established respectively, and K3 decision results are obtained accordingly; Voting is performed based on the K3 decision results to determine the final completion quality influencing factor.

2. The method for identifying "sweet spots" in shale oil and gas reservoirs according to claim 1, characterized in that: The acquisition of the main control factors includes: Acquire core data, seismic data, and well logging data; Performing Pearson correlation analysis on all the data in the core data, the seismic data and the well logging data to calculate the correlation coefficient; Data that meets the preset range of the correlation coefficient is selected as the main control factor.

3. The method for identifying "sweet spots" in shale oil and gas reservoirs according to claim 1, characterized in that: Determining the classification category according to the main controlling factor includes: According to the main controlling factor, the classification category corresponding to the main controlling factor is determined, wherein the classification category includes an organic matter quality characteristic category, a reservoir quality characteristic category and a completion quality characteristic category, the organic matter quality characteristic category includes at least one of total organic carbon content, maturity, and storage capacity, the reservoir quality characteristic category includes at least one of porosity, density, and thickness, and the completion quality characteristic category includes at least one of clay content, Poisson's ratio, Young's modulus, brittleness, and pressure.

4. The method for identifying "sweet spots" in shale oil and gas reservoirs according to claim 1, characterized in that: The sweet spot evaluation factors include organic matter impact factors, reservoir quality impact factors and completion quality impact factors. The determination of favorable shale gas exploration areas based on the sweet spot evaluation factors includes: Sorting the organic matter influencing factors screened out by the organic matter quality model according to sensitivity to determine a first sorting result; Sorting the reservoir quality influencing factors screened out by the reservoir quality model according to sensitivity to determine a second sorting result; Sorting the completion quality influencing factors screened out by the completion quality model according to sensitivity to determine a third sorting result; Based on the first sorting result, the second sorting result and the third sorting result, favorable shale gas exploration areas are determined.

5. The method for identifying "sweet spots" in shale oil and gas reservoirs according to claim 1, characterized in that: The process of establishing the dessert evaluation model includes: According to the main control factors belonging to the classification category, a sample set is formed, wherein the label corresponding to the main control factor is a dessert type; For a single decision tree, samples are extracted from the sample set to form at least one training set and a test set; The node is split from the root node. The two child nodes formed after the split are measured by Gini coefficient or information gain. The number of M features selected during the split is randomly selected from the M features of the training sample. According to the calculated information gain or Gini coefficient, the features are selected from the M features according to the splitting principle for splitting. Each node is split according to the splitting method of the root node until it cannot be split anymore, forming a leaf node; A random forest is formed for a plurality of decision trees that have been constructed, and the dessert evaluation model is established.

6. A device for identifying "sweet spots" in shale oil and gas reservoirs, characterized in that: include: An acquisition unit, used for acquiring the main control factor; A processing unit, used to determine a classification category according to the main control factor; Used to construct a sweet spot evaluation model based on random forest and according to the main control factors and the classification categories, and determine the sweet spot evaluation factors; A prediction unit, used to determine a favorable exploration area for shale gas according to the sweet spot evaluation factor; The sweet spot evaluation model includes an organic matter quality model, the sweet spot evaluation factor includes an organic matter influencing factor, and the sweet spot evaluation model is constructed based on the random forest according to the main controlling factors and classification categories, including: According to the main controlling factors belonging to the organic matter quality characteristic category, the organic matter quality training set is constructed; Extracting K1 organic matter sample sets from the organic matter quality training set, wherein the capacity of each organic matter sample set is a preset capacity; For K1 organic matter sample sets, K1 decision tree models are established respectively, and K1 decision results are obtained accordingly; Voting is performed based on K1 decision results to determine the final organic matter impact factor; The sweet spot evaluation model includes a reservoir quality model, the sweet spot evaluation factors include reservoir quality influencing factors, and the sweet spot evaluation model is constructed based on random forest according to the main control factors and the classification categories, including: Constructing a reservoir quality training set based on the main controlling factors belonging to the reservoir quality characteristic category; Extracting K2 reservoir quality sample sets from the reservoir quality training set, wherein the capacity of each reservoir quality sample set is a preset capacity; For K2 reservoir quality sample sets, K2 decision tree models are established respectively, and K2 decision results are obtained accordingly; Voting is performed based on K2 decision results to determine the final reservoir quality influencing factor; The sweet spot evaluation model includes a completion quality model, the sweet spot evaluation factors include completion quality influencing factors, and the sweet spot evaluation model is constructed based on random forest according to the main control factors and the classification categories, including: According to the main control factors belonging to the completion quality feature category, a completion quality training set is constructed; Extracting K3 organic matter sample sets from the completion quality training set, wherein the capacity of each completion quality sample set is a preset capacity; For K3 completion quality sample sets, K3 decision tree models are established respectively, and K3 decision results are obtained accordingly; Voting is performed based on the K3 decision results to determine the final completion quality influencing factor.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method for identifying "sweet spots" in shale oil and gas reservoirs according to any one of claims 1 to 5 is implemented.

Images