Core-logging double-constraint lacustrine facies beach and dam distinguishing method

Abstract

The invention relates to the technical field of oil-gas exploration, in particular to a core-logging double-constraint lacustrine facies beach-bar distinguishing method, which comprises the following steps of S1, beach-bar core characteristic parameter selection; s2, determining characteristic parameters of a beach and bar logging curve and establishing a plate; s3, data extraction and preprocessing; s4, constructing a core-logging comprehensive judgment model sample database; and S5, constructing a beach and bar rock core-logging comprehensive judgment model based on a standard mapping clustering method. According to the method, rock core and logging information is combined, and the delta transformation type beach bar is accurately identified on the basis of beach bar cause classification, further on the basis of delta transformation type beach bar rock core analysis and logging analysis and on the basis of theories such as clustering analysis method analysis and standard mapping clustering method. The problem that at present, no delta transformation type beach bar core and logging information comprehensive distinguishing exists is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of oil and gas exploration, and in particular to a core-logging dual-constraint lacustrine beach-bar differentiation method. Background Art

[0002] Beach-bars primarily form in coastal zones, influenced by the combined effects of waves and longshore currents. They are reconstructed sedimentary bodies. Beaches are typically sheet-like, while bars are typically ribbon-like, often developing in symbiosis. Early beach-bar research focused on basic theoretical concepts, such as defining the concepts of beaches and bars. Current research has expanded beyond the classification and source-sink systems of beaches and bars to encompass a comprehensive framework encompassing wind fields, provenance, and basins.

[0003] The classification of beach-bar types is a core topic in beach-bar research. Classification criteria are primarily based on factors such as sandbody composition, spatial distribution, provenance characteristics, and hydrodynamic conditions. Among the various classification systems, the genetic classification method holds particular practical value. This method subdivides beach-bar systems into frontal, lateral, bedrock, submerged, and storm types. It is noteworthy that the accurate identification and differentiation of frontal and lateral beach bars, as crucial components of lacustrine sedimentary systems, holds important practical significance for improving the exploration efficiency of lithologic oil and gas reservoirs in my country.

[0004] Currently, beach-bar systems are primarily classified into five types based on their genesis: frontal, lateral, bedrock, submerged, and storm. Specifically, researchers categorize lacustrine sedimentary environments into two types: frontal and lateral, based on their spatial distribution. While this location-based spatial classification method has achieved preliminary classification of lacustrine beach-bars, it still has significant limitations, primarily due to a lack of in-depth analysis and systematic differentiation of key geological parameters such as core characteristics, well logging responses, and mud logging data.

[0005] Existing technologies for sedimentary facies identification primarily rely on Bayesian algorithms, Fisher algorithms, and machine learning. These methods construct sedimentary facies discrimination models by analyzing the characteristics of well logging parameters. However, this modeling approach, which relies solely on well logging data, has significant drawbacks. The inability to integrate multiple sources of geological information, such as core data, limits the accuracy of the resulting identification models. This technical limitation has, to a certain extent, hampered the exploration and development of lacustrine beach-bar sandstone reservoirs, particularly in the detailed characterization of complex sedimentary facies. There is still room for improvement.

[0006] One existing technology, Fisher-based quantitative sedimentary facies identification technology, primarily utilizes well logging data and uses statistical analysis methods to establish a quantitative sedimentary facies identification model. The main process includes data collection and preprocessing, feature extraction and selection, statistical analysis model construction, model training and validation, and sedimentary facies identification and prediction. First, well logging data is collected and organized, and standardized, including but not limited to gamma ray (GR), acoustic transit time (AC), neutron porosity (CNL), and density (DEN). Second, key characteristic parameters are extracted and selected, and statistical models are constructed using cluster analysis and principal component analysis, such as the average GR value and the changing trend of the GR curve. The model is then trained and validated using training samples, optimizing model parameters. Finally, the trained model is applied to the target interval for sedimentary facies identification, and the results are interpreted in conjunction with the geological context.

[0007] There are many problems with this approach:

[0008] (1) Fisher has poor ability to handle nonlinear relationships and is unable to accurately depict the complex relationship between sedimentary facies and logging responses, resulting in limited recognition accuracy;

[0009] (2) This method lacks consideration of geological laws. The selected parameters are mainly based on well logging data, and there is no effective integration of multi-source geological information, which leads to deviations between the identification results and geological understanding.

[0010] The second existing technology is the quantitative identification method of sedimentary phases based on machine learning. The main processes include data collection and preprocessing, feature engineering, model selection and training, model verification and optimization, sedimentary phase identification and prediction, and result interpretation. First, collect the logging data of the study area, including natural gamma, acoustic time difference, neutron porosity, density, etc., and perform preprocessing such as cleaning, denoising, and standardization on the data; secondly, extract characteristic parameters sensitive to sedimentary phases through feature engineering, such as statistical features such as mean, variance, and trend, and use methods such as principal component analysis (PCA) and linear discriminant analysis (LDA) to perform feature dimensionality reduction and selection to remove redundant information and improve model efficiency; then, select appropriate machine learning algorithms according to the characteristics of the data, such as support vector machines (SVM), random forests (RF), gradient boosting trees (GBDT) or deep learning models (such as convolutional neural networks). The model is trained using labeled training samples and optimized through cross-validation, grid search and other methods to improve the generalization ability and recognition accuracy of the model. Then, an independent validation dataset is used to evaluate the model performance, and indicators such as accuracy, recall rate, and F1 score are calculated. The error types of the model are analyzed in combination with the confusion matrix, and the feature selection or model structure is further adjusted. Finally, the trained model is applied to the target layer to identify the sedimentary phases of the well logging curves, and the identification results are interpreted in combination with the geological background and sedimentary environment to ensure their rationality and reliability.

[0011] This method has the following technical disadvantages:

[0012] (1) High data dependence. Machine learning models are very sensitive to the quality of input data. Noise, missing values or outliers in the data will significantly affect the training effect and recognition accuracy of the model. In addition, the annotation of geological data requires expert experience, which is time-consuming and labor-intensive.

[0013] (2) High computing resource requirements, especially deep learning models, which require a lot of computing resources and time for training, and have high requirements for hardware equipment; in addition, large-scale logging data requires a large storage space, which increases the difficulty of data management and processing. Summary of the Invention

[0014] The present invention provides a lacustrine beach-bar differentiation method with dual core and well logging constraints. By combining core and well logging data, a standard mapping clustering method and self-developed software programming are adopted to establish a more accurate lacustrine beach-bar differentiation model, providing support for oil and gas exploration and development of lacustrine beach-bar sand bodies.

[0015] The technical problems to be solved by the present invention are:

[0016] (1) The standard mapping clustering method used in this paper projects high-dimensional data into a low-dimensional space by maximizing the ratio of inter-class divergence to intra-class divergence, while preserving the information that distinguishes between classes. This dimensionality reduction method can effectively reduce the data dimension and computational complexity, while improving classification performance and better preserving classification-related features.

[0017] (2) In addition to using well logging data, the present invention also introduces core data, including skewness and peak values in particle size, and realizes the quantification of sedimentary structural parameters for the first time based on hydrodynamic conditions, thus achieving quantitative identification of sedimentary facies constrained by multi-source information.

[0018] (3) The present invention is the first to software-based quantitative identification process, realizing the integration of the entire process from data processing to sedimentary phase identification, which can effectively reduce labor costs and time consumption, avoid errors and omissions that may occur in manual operations, and improve work efficiency and identification accuracy.

[0019] The purpose of the present invention is to provide a core-logging dual-constrained lacustrine beach-bar differentiation method, aiming to solve the problem of distinguishing front beach bars from side beach bars.

[0020] To achieve the above object, the present invention adopts the following technical solutions:

[0021] The present invention provides a core-logging dual-constraint lacustrine beach-bar differentiation method, which comprises the following steps:

[0022] S1. Selection of characteristic parameters of beach-bar cores;

[0023] S2. Determine the characteristic parameters of beach bar logging curve and establish the chart;

[0024] S3, data extraction and preprocessing;

[0025] S4. Construction of sample database for core-logging integrated judgment model;

[0026] S5. Construction of a comprehensive beach-bar core-logging judgment model based on the standard mapping clustering method.

[0027] According to the above scheme, in S1, when selecting the core characteristic parameters of the beach-bar, it is necessary to collect core samples with statistical significance to determine the core characteristic indicators of the beach-bar. Through systematic core description, the development intervals of the front beach bar and the side beach bar are divided, and the core parameters that can effectively characterize the sedimentary characteristics of the beach bar are screened out as the basis for distinguishing the two types of beach bars. This includes:

[0028] ① Average particle size (Mz): reflects the average size of rock particles and reveals the hydrodynamic conditions;

[0029] ② Standard deviation (σ1): reflects the degree of sediment sorting. The larger the standard deviation, the less ideal the sorting and the more unstable the hydrodynamics.

[0030] ③ Skewness (SK): reflects the symmetry of the particle size distribution. If the skewness value is greater than 0, the overall particle size is fine, otherwise the overall particle size is coarse.

[0031] ④ Peak shape (KG): reflects the shape of the particle size frequency curve. The wider the peak, the more unstable the hydrodynamics.

[0032] ⑤ Hydrodynamic conditions: Different sedimentary structures are formed under different hydrodynamic conditions. Based on the principles of sedimentology, the hydrodynamic strength of different sedimentary structures is analyzed. As the hydrodynamic strength increases, the sedimentary structure gradually transitions from massive bedding (mudstone), wavy bedding, deformed bedding, parallel bedding, massive bedding (sandstone), grain-sequenced bedding, and large-scale cross-bedding. Different types of sedimentary structures are then assigned points.

[0033] According to the above scheme, in S2, the determination of characteristic parameters of beach-bar logging curve and establishment of chart include the following processes:

[0034] First, the logging data corresponding to the cored wells selected through core observation were obtained. Then, through the analysis of the logging curve characteristics, the front beach bar and lateral beach bar development intervals were selected, and the logging curve morphology, amplitude, and envelope combination characteristics of the front beach bar and lateral beach bar were analyzed.

[0035] Then, the logging parameters that best reflect the sedimentary characteristics of the beach-bar are selected as the core discrimination parameters for the front beach-bar and the side beach-bar, including:

[0036] ① Specific amplitude (α): the ratio of the difference between the maximum or minimum value of the curve and the mudstone baseline value (ΔA) to the thickness of the curve element (Δd);

[0037] ② Root of variance (GS): reflects the serration of the curve. The smaller the GS, the fewer serrations, the smoother the curve, the more stable the sedimentary environment, and the better the sorting and roundness of the sediments. S 2 is the variance, γ* is the average fluctuation amplitude of the curve);

[0038] ③ Average median (A): reflects the changes in sedimentary environment;

[0039] ④Average amplitude (K): reflects the changes in the logging curve and can be used to describe the sedimentary environment and hydrodynamic trends Represents the logging curve value of the mth sampling point in the nth segment of the curve).

[0040] In addition, a logging identification chart for the front beach bar and the side beach bar was established to provide a basis for the subsequent automatic classification of sedimentary facies in non-coring wells.

[0041] According to the above scheme, S3 specifically includes the following sub-steps:

[0042] Step 3.1: Core-log data extraction based on manual screening;

[0043] Step 3.2: Normalization of core-log data.

[0044] In step 3.1, based on the manually screened core-logging data extraction, and based on the front and side beach-bar development intervals divided by the previous core description, no less than 10 representative beach-bar development wells are selected to extract quantitative indicators including sedimentary structure and grain size parameters (average grain size, standard deviation, skewness, and kurtosis). For the beach-bar development wells, the GR value range of sandstone and mudstone is counted and effectively distinguished. The mudstone interval is eliminated, and the front beach bar and side beach bar sandstone intervals with complete sedimentary characteristics are retained, and the corresponding logging curves are extracted. The specific implementation steps are as follows: first, the core grain size parameters of the selected 10 transformed beach-bar wells are counted; second, the GR curve is used to distinguish the mudstone and sandstone intervals, and the corresponding logging curves of the sandstone are extracted; finally, the integrity and rationality of the extracted parameters are determined by combining the logging line chart.

[0045] In step 3.2, the core-logging data is normalized to eliminate dimensionless effects, and the selected core parameters and logging curve values are normalized to provide a basis for subsequent parameter calculations.

[0046] According to the above scheme, step 4 specifically includes the following sub-steps:

[0047] Step 4.1: Establishment of core database;

[0048] Step 4.2: Establishment of logging database.

[0049] In step 4.1, a core database is established to organize core identification data such as sedimentary structure, average grain size, standard deviation, skewness, and peak state of the delta-reformed beach-bar development intervals of 10 wells determined through core observation;

[0050] In step 4.2, a well logging database is established, and after determining the beach-bar deposits through cluster analysis, well logging discriminant data such as amplitude ratio, root mean square error, mean median, and mean amplitude of 10 delta-reformed beach-bar development wells are obtained.

[0051] According to the above scheme, step 5 specifically includes the following sub-steps:

[0052] Step 5.1: Establishment of a comprehensive judgment model for delta-reconstructed beach-bars based on the standard mapping clustering method;

[0053] Step 5.2: Error analysis of the comprehensive judgment model for delta-reconstructed beach-bars based on the standard mapping clustering method.

[0054] In step 5.1, the establishment of a comprehensive determination model for delta-reconstructed beach-bars based on the standard mapping clustering method includes the following core processes: feature mapping of the preprocessed sample data set to identify effective segmentation boundary points; establishing a classification discriminant function based on the determined one-dimensional feature vector, and normalizing the function coefficients. The detailed formula is as follows:

[0055] First, establish the discriminant function. Suppose there are k populations G1, G2, ..., G k , the sample points are n1, n2, ..., n k , let n=n1+n2+…+n k , is the observation vector of the αth sample point of the i-th population.

[0056] Assume that the discriminant function established is:

[0057]

[0058] Where c is the coefficient vector;

[0059] y(x) in the Gth i The overall mean of the items is:

[0060]

[0061] in, is the sample mean vector of x in G;

[0062] y(x) in G i The sample variance on is:

[0063]

[0064] Among them, s (i) is the sample covariance matrix of x in G.

[0065] To select the coefficient vector c, we need to use Maximum, where q is a positive weighting coefficient that depends on the prior probability;

[0066] Will Substitution Available models:

[0067]

[0068] Among them, E is the within-group variance matrix, A is the sample covariance matrix between the populations, so Reaching the maximum, Ac=λEc, it can be seen that λ and c are the generalized eigenvectors of A and E;

[0069] Thus, m discriminant functions can be constructed:

[0070] y t (x) = C (i) x,i=1,2,...,m

[0071] With λ1, λ2, ..., λ m ,(λ1≥λ2≥,...,λ m ≥0) represents all non-zero characteristic roots, l1, l2, ..., l m is the corresponding eigenvector. When a=l1, Δa reaches its maximum;

[0072] Then, suppose the population is divided into p classes, if Then y P Belongs to Category 1;

[0073] On the basis of the above mathematical formulas and the analyzed core and logging parameters, a mathematical discriminant function was constructed and coefficients were normalized, ultimately achieving the construction of a comprehensive identification model for delta-reformed beach bars based on the standard mapping clustering algorithm.

[0074] In the step 5.2, the back-judgment division of the delta-reformed beach-bar comprehensive judgment model using the standard mapping clustering method is to back-judgment and divide the identified delta-reformed beach-bar on the basis of establishing the delta-reformed beach-bar discrimination model.

[0075] This method, combined with core and well logging data, goes a step further in the classification of beach-bar genesis. By analyzing delta-reformed beach-bar core and well logging data, and applying cluster analysis and standard mapping clustering, it accurately identifies delta-reformed beach-bars. This overcomes the current lack of comprehensive differentiation of delta-reformed beach-bars using core and well logging data. This method provides guidance for oil and gas exploration and development of lacustrine beach-bar sandstones in my country. BRIEF DESCRIPTION OF THE DRAWINGS

[0076] The accompanying drawings, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0077] Figure 1 It is a technical flow chart of the present invention;

[0078] Figure 2 It is the overall characteristic of the core of a specific embodiment of the present invention;

[0079] Figure 3This is a diagram showing the effect of the median filtering method for the drilling GR curve according to a specific embodiment of the present invention;

[0080] Figure 4 This is a well logging chart for distinguishing between the leading edge beach bar and the side edge beach bar in a specific embodiment of the present invention;

[0081] Figure 5 This is a diagram showing the effect of distinguishing lacustrine beach and bar using the core-logging integrated discrimination model according to a specific embodiment of the present invention. DETAILED DESCRIPTION

[0082] In order to enable those skilled in the art to more clearly understand the technical solution of the present invention, the technical solution of the present invention will be described in detail below with reference to specific embodiments.

[0083] The Paleogene research area of the Liaodong Bay Depression covers an area of 10,000 km 2 The tectonic units include the Liaodong Sag, Liaodong Uplift, Liaozhong Sag, Liaoxi Uplift, and Liaoxi Sag. Traditionally, it has been believed that the Paleogene in this area primarily developed braided river delta front deposits, lacking large-scale beach-bar depositional systems. However, recent research indicates that the Paleogene on both sides of the Liaoxi Low Uplift and in the northern Liaoxi Depression has significant potential for beach-bar development. This study, for the first time, applied the "core-logging dual-constrained lacustrine beach-bar identification technology" to successfully achieve comprehensive identification of front and lateral beach bars in this area.

[0084] like Figure 1 As shown, a core-logging dual-constraint lacustrine beach-bar differentiation technology includes the following steps:

[0085] (1) Selection of beach bar core characteristic parameters

[0086] First, a statistically representative core sample of the Paleogene in the Liaodong Bay Depression was systematically collected, and the sedimentary development well locations of the frontal beach-bar and lateral beach-bar were identified through core description. Figure 2 Based on the grain size test results, core indicators that can effectively characterize sedimentary facies characteristics, such as sedimentary structure (Table 1) and grain size parameters (including average grain size, standard deviation, skewness coefficient, and kurtosis coefficient), were selected as the basis for distinguishing the two types of beach bars.

[0087] Table 1 Diagram for distinguishing hydrodynamic strength of different sedimentary structures

[0088]

[0089] (2) Determination of characteristic parameters of beach bar logging curve

[0090] A representative Paleogene logging dataset from the Liaodong Bay Depression was systematically collected. By analyzing logging response characteristics, the development intervals of frontal and lateral beach bars were identified. The logging curve morphology, response amplitude, and envelope combination patterns of the two types of beach bars were compared and analyzed. Log parameters such as relative amplitude, coefficient of variation, median amplitude, and average amplitude that effectively characterize sedimentary facies were selected as logging indicators to distinguish between frontal and lateral beach bars.

[0091] (3) Data extraction and preprocessing based on manual screening

[0092] First, core data preprocessing based on core screening was performed. Through systematic core characterization, 10 wells representing leading and lateral beach-bar development were selected, including Wells J1, J2, …, and J10. After removing mudstone intervals, a lithofacies sequence of different beach-bar types was established, ensuring that each well possessed complete sedimentary structural characteristics and core discriminant indicators such as grain size parameters (including standard deviation, mean grain size, kurtosis, and skewness). Ultimately, a dataset of 37 valid core samples was obtained from these 10 wells.

[0093] Then, the logging data were preprocessed based on manual screening. In view of the limited number of coring wells in Liaodong Bay, the GR curves of the 10 wells (J1-J10) with potential development of frontal beach bars and lateral beach bars were first processed by median filtering ( Figure 3 ), and used the GR curve to effectively distinguish sandstone and mudstone intervals, extracted the logging curve segments corresponding to beach-bar development, and finally obtained a total of 37 valid logging sample data sets from 10 wells, which were uniformly standardized.

[0094] (4) Construction of core and well logging training sample database

[0095] The core and well logging training sample database consists of two types of characteristic parameters: one is the core parameters that can accurately characterize the sedimentary facies characteristics, and the other is the well logging parameters that can effectively reflect the sedimentary facies characteristics. Specifically, they include: core characteristic data such as the particle size parameters (average particle size, standard deviation, skewness coefficient, and kurtosis coefficient) of wells J1-J10 obtained based on core descriptions; qualitative identification charts ( Figure 4 ), extracted GR curve values for wells J11-J20, and calculated logging response parameters (specific amplitude, root mean square error, mean median, and mean amplitude) and other logging characteristic data (Table 2). The standardized data were used to establish a core and logging training sample database.

[0096] Table 2. Training sample set of core-log integrated discrimination model

[0097]

[0098] (5) Establishment of a core-logging integrated discrimination model based on programming language

[0099] The data were imported into the compiled core-logging comprehensive discrimination program, where sedimentary structure, average grain size, standard deviation, skewness, kurtosis, amplitude ratio, root variance of variance, average median, and average amplitude data were imported into the independent variables, and sedimentary facies type was imported into the grouping variable. The grouping range was defined as 1-2 to obtain the discriminant function:

[0100] y1=1.400M z +1.282σ1+2.311α+2.147GS-1.282S k -0.945K G -0.675A-0.137K-5.477

[0101] y2=-1.375M z +1.850σ1+13.473α-2.507GS-2.536S k -0.833K G -0.785A-056K-1.354

[0102] Based on the establishment of the front beach bar and lateral beach bar discrimination model, the error analysis of the identified front beach bar and lateral beach bar was carried out. The results showed that the accuracy of the core-logging comprehensive discrimination model was 82.6% (Table 3). The front beach bar and lateral beach bar were distinguished in Well J1 using the beach bar core-logging comprehensive discrimination model. The discrimination results are as follows: Figure 5 shown.

[0103] Table 3 Error analysis results of leading edge beach bar and side edge beach bar

[0104]

[0105]

[0106] This invention provides a technical method for quantitatively distinguishing between leading and lateral beach bars. Based on extensive literature, core logging data, and optimized algorithms, a quantitative sedimentary facies identification model is constructed to conduct a comprehensive differentiation of lacustrine beach bars using a "core-logging dual constraint" approach, thereby enabling the identification of leading and lateral beach bars. This method addresses the subjectivity and accuracy issues inherent in qualitative identification, as well as the time-consuming and labor-intensive manual interpretation of well logs. It offers the advantages of simplicity, speed, and ease of implementation for lacustrine beach bar identification, while also maintaining a reasonable degree of accuracy.

[0107] This method was first applied in the project "Paleogene Lithologic Reservoir Formation, Sandbody Identification, Characterization, and Target Evaluation in the Liaodong Bay Region," with promising results. The Liaodong Bay is controlled by the right-lateral strike-slip action of the Tanlu Fault Zone, resulting in a northeast-trending dustpan-shaped fault structure with multiple fault terraces and slope breaks. Furthermore, the abundant provenance and favorable hydrodynamic conditions of the second member of the Shahejie Formation create favorable conditions for the vertical superposition and lateral migration of beach-bar sandbodies.

[0108] After analyzing the geological conditions, it was found that the area has two types of beach bars: frontal and lateral. Through the collection and preprocessing of core and well logging data, a training set was established, and a quantitative identification model for lacustrine beach bars was established using the standard mapping clustering method. Ultimately, the quantitative distinction between frontal and lateral beach bars was achieved, providing a certain foundation and basis for the selection of favorable areas for distributed beach-bar oil and gas reservoirs in the Liaodong Bay region and the evaluation of resource potential. In summary:

[0109] (1) The present invention optimizes the algorithm, realizes the quantification of sedimentary structural parameters for the first time, and realizes the quantitative identification of sedimentary phases constrained by multi-source information.

[0110] (2) The present invention is the first to software-based quantitative identification process, realizing the integration of the entire process from data processing to sedimentary phase identification, avoiding errors and omissions that may occur in manual operation, and improving work efficiency and identification accuracy.

[0111] The above embodiments are preferred implementation modes of the present invention, but the implementation modes of the present invention are not limited to the above embodiments. Any other changes, modifications, substitutions, combinations, and simplifications that do not deviate from the spirit and principles of the present invention should be considered as equivalent replacement methods and are included in the scope of protection of the present invention.

Claims

1. A core-logging dual-constraint method for distinguishing lake facies beaches and bars, characterized by: The following steps are involved: S1. Selection of characteristic parameters of beach-bar cores; S2. Determine the characteristic parameters of beach bar logging curve and establish the chart; S3, data extraction and preprocessing; S4. Construction of sample database for core-logging integrated judgment model; S5. Construction of a comprehensive beach-bar core-logging judgment model based on the standard mapping clustering method.

2. The core-logging dual-constrained lacustrine beach-bar differentiation method according to claim 1, characterized in that: In S1, when determining the core characteristic indicators of beach-bars, statistically significant core samples are collected. Through systematic core description, the development intervals of frontal beach-bars and lateral beach-bars are divided, and core parameters that can effectively characterize the sedimentary characteristics of beach-bars are selected as the basis for distinguishing the two types of beach-bars, including: ① Average particle size Mz: reflects the average size of rock particles and reveals the hydrodynamic conditions; ② Standard deviation σ1: reflects the degree of sediment sorting. The larger the standard deviation, the less ideal the sorting and the more unstable the hydrodynamics. ③ Skewness SK: reflects the symmetry of particle size distribution. If the skewness value is greater than 0, the overall particle size is fine, otherwise the overall particle size is coarse. ④ Peak state KG: reflects the shape of the particle size frequency curve. The wider the peak, the more unstable the hydrodynamics. ⑤ Hydrodynamic conditions: Different sedimentary structures are formed under different hydrodynamic conditions. Based on the principles of sedimentology, the hydrodynamic strength of different sedimentary structures is analyzed. With the increase of hydrodynamic intensity, the sedimentary structure gradually transitions from massive bedding, wavy bedding, deformed bedding, parallel bedding, massive bedding, and grain-sequence bedding to large-scale cross-bedding. Different types of sedimentary structures are scored.

3. The core-logging dual-constrained lacustrine beach-bar differentiation method according to claim 1, characterized in that: S2 includes the following processes: First, the logging data corresponding to the cored wells selected through core observation were obtained. Then, through the analysis of the logging curve characteristics, the front beach bar and lateral beach bar development intervals were selected, and the logging curve morphology, amplitude, and envelope combination characteristics of the front beach bar and lateral beach bar were analyzed. Then, the logging parameters that best reflect the sedimentary characteristics of the beach bar are selected as the logging discrimination parameters for the front beach bar and the side beach bar, including: ① Ratio α: the ratio of the difference ΔA between the maximum or minimum value of the curve and the mudstone baseline value to the thickness of the curve element Δd; ② Root of variance GS: reflects the serration of the curve. The smaller the GS, the fewer serrations, the smoother the curve, the more stable the sedimentary environment, and the better the sorting and roundness of the sediments. S 2 is the variance, γ* is the average fluctuation amplitude of the curve; ③ Average median A: reflects the changes in sedimentary environment; ④Average amplitude K: reflects the changes in the logging curve and is used to describe the sedimentary environment and hydrodynamic trends. X n (m) represents the well logging curve value of the mth sampling point in the nth segment of the curve; In addition, a qualitative identification chart for front beach bars and side beach bars was established based on well logging, providing a theoretical basis for the subsequent automatic classification of sedimentary facies in non-coring wells.

4. The core-logging dual-constrained lacustrine beach-bar differentiation method according to claim 1, characterized in that: S3 specifically includes the following sub-steps: Step 3.1: Core-log data extraction based on manual screening; The specific method is as follows: Based on the front and side beach-bar development intervals delineated by the previous core description, no less than 10 representative beach-bar development wells are selected and relevant grain size parameters are calculated. For the beach-bar development wells, the GR value range of sandstone and mudstone is calculated and effectively distinguished. The mudstone intervals are then eliminated, and the front and side beach-bar sandstone intervals with complete sedimentary characteristics are retained. The corresponding logging curves are then extracted. The specific implementation steps are as follows: Firstly, the core grain size parameters of the 10 selected reformed beach-bar wells were counted; Secondly, the GR curve is used to distinguish between mudstone and sandstone intervals, and the corresponding logging curve of sandstone is extracted; Finally, the integrity and rationality of the extracted parameters are determined by combining the well logging line chart; Step 3.2: Standardization of core-logging data. After filtering the logging curves, the selected core parameters and logging curve values are standardized to eliminate the dimensionless effect, providing a basis for subsequent parameter calculations.

5. The core-logging dual-constrained lacustrine beach-bar differentiation method according to claim 1, characterized in that: The step 4 specifically includes the following sub-steps: Step 4.1: Establishment of core database; The sedimentary structure, average grain size, standard deviation, skewness, and peak-to-peak core discrimination data of 10 wells' delta-reformed beach-bar development intervals determined through core observation were collated; Step 4.2: Establishment of logging database; After confirming the beach-bar deposits, the logging discriminant data of the specific amplitude, root variance of variance, mean median, and mean amplitude of 10 wells with delta-reformation beach-bar development were obtained.

6. The core-logging dual-constrained lacustrine beach-bar differentiation method according to claim 1, characterized in that: The step 5 specifically includes the following sub-steps: Step 5.1: Establishment of a comprehensive judgment model for delta-reconstructed beach-bars based on the standard mapping clustering method; The specific process includes: performing feature mapping on the preprocessed sample data set to identify effective segmentation boundary points; establishing a classification discriminant function based on the determined one-dimensional feature vector and normalizing the function coefficients; Based on the analyzed core and logging parameters, a mathematical discriminant function was constructed and the coefficients were normalized, ultimately achieving the construction of a comprehensive identification model for delta-reformed beach bars based on a standard mapping clustering algorithm. Step 5.2: Error analysis of the comprehensive judgment model for delta-reconstruction beach-bar based on the standard mapping clustering method; The back-judgment division of the delta-reformed beach-bar comprehensive judgment model based on the standard mapping clustering method is based on the establishment of the delta-reformed beach-bar discrimination model, and the identified delta-reformed beach-bar is back-judged and divided.

7. The core-logging dual-constrained lacustrine beach-bar differentiation method according to claim 6, characterized in that: The formula in step 5.1 is as follows: First, establish the discriminant function; suppose there are k populations G1, G2, ..., G k , the sample points are n1, n2, ..., n k ,make is the observation vector of the αth sample point of the i-th population; Let the established discriminant function be: Where c is the coefficient vector; y(x) in the Gth i The overall mean of the items is: in, is the sample mean vector of x in G; y(x) in G i The sample variance on is: Among them, s (i) is the sample covariance matrix of x in G; To select the coefficient vector c, we need to use Maximum, where q is a positive weighting coefficient that depends on the prior probability; Will Substitution Get the model: Among them, E is the within-group variance matrix, A is the sample covariance matrix between the populations, so Reaching the maximum, Ac=λEc, it can be seen that λ and c are the generalized eigenvectors of A and E; Thus, construct m discriminant functions: y t (x)=C (i) x,i=1,2,...,m With λ1, λ2, ..., λ m , represents all non-zero characteristic roots, λ1≥λ2≥,...,λ m ≥0, l1, l2, …, l m is the corresponding eigenvector. When a=l1, Δa reaches its maximum; Then, divide the population into p classes, if p,q=1,2,…,m,q≠1, then y P Belongs to Category 1.

Images