A lithofacies identification method, device, equipment and storage medium
By acquiring core description data and logging curves from cored wells, and combining them with deep learning models for lithofacies identification, the problems of reliance on experience and lack of geological concepts in existing technologies have been solved, achieving refined interpretation and high accuracy in lithofacies identification.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- CHINA UNIV OF PETROLEUM (BEIJING)
- Filing Date
- 2025-02-10
- Publication Date
- 2026-06-19
AI Technical Summary
Existing technologies rely heavily on the experience of interpreters in lithofacies identification, resulting in low accuracy. Furthermore, methods based on artificial intelligence algorithms lack geological concepts and internal connections, leading to inaccurate lithofacies interpretation.
By obtaining core description data from cored wells, we determine the lithofacies type and combination characteristics. Combined with well logging curve analysis, we use a deep learning model for lithofacies identification, including normalization processing, sensitivity analysis, and weight value calculation. We then establish model training data and iteratively update the model to improve identification accuracy.
It improves the interpretation accuracy of complex lithofacies, incorporates the development law of lithofacies in the vertical direction, lays the foundation for sedimentary geological interpretation, and realizes adaptive and refined interpretation of lithofacies.
Smart Images
Figure CN120028875B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of physical logging, and in particular to a method, apparatus, equipment and storage medium for lithofacies identification. Background Technology
[0002] Geophysical well logging interpretation primarily uses measurements from various logging instruments, such as natural gamma ray, sonic transit time, neutron density, and resistivity, to interpret formation characteristics and determine important parameters such as lithology, lithofacies, physical properties, and hydrocarbon potential. Currently, there are two main types of geophysical well logging lithofacies interpretation methods: one is conventional well logging interpretation, which relies on core observation and thin section analysis combined with qualitative and quantitative well logging methods under the guidance of sedimentary models. This method largely depends on the knowledge and experience of the interpreters, resulting in relatively slow interpretation progress and lower accuracy. The second type is well logging lithofacies interpretation based on artificial intelligence algorithms, which mainly uses supervised, unsupervised, or semi-supervised methods for intelligent lithofacies interpretation. Although this method has been widely used in well logging interpretation, most methods only consider the relationship between sample labels and well logging response characteristics, lacking constraints from geological concepts or laws, and failing to explore the intrinsic connection between geological laws and artificial intelligence algorithms. This means that machine learning methods are merely mathematical algorithms, lacking mechanistic understanding.
[0003] In conclusion, how to combine intelligent algorithms to identify complex lithofacies is a problem that urgently needs to be solved. Summary of the Invention
[0004] In view of this, the purpose of this invention is to provide a method, apparatus, device, and storage medium for lithofacies identification, capable of identifying complex lithofacies by combining intelligent algorithms. The specific solution is as follows:
[0005] Firstly, this application provides a method for identifying petrographic features, including:
[0006] Based on the obtained core description data of the core well, the lithofacies type corresponding to the core section is obtained. Based on the obtained lithofacies type corresponding to the core section, the lithofacies combination characteristics corresponding to the core section are determined. Based on the lithofacies type corresponding to the core section, the target probability distribution characteristics of the vertical lithofacies are determined. The heterogeneous characteristics of the target lithofacies structure are analyzed using the lithofacies combination characteristics.
[0007] Obtain the target layer logging curve, normalize the target layer logging curve to obtain the processed target layer logging curve, and then determine the target logging curve based on the processed target layer logging curve and the lithofacies type.
[0008] The first depth value of the target peak point and the second depth value of the target valley point corresponding to the target logging curve are determined. Based on the first depth value and the second depth value, the third depth value of the target half-amplitude point is obtained. The corresponding target weight value is determined using the third depth value of the target half-amplitude point. The target layer data between the target half-amplitude points is obtained according to the target weight value.
[0009] Based on the target layer data between the obtained target half-width points, model training data is established. The initial lithofacies identification model is trained using the model training data to obtain model prediction results. The predicted probability distribution features in the obtained model prediction results are compared with the target probability distribution features. The predicted lithofacies structural heterogeneity features in the obtained model prediction results are compared with the target lithofacies structural heterogeneity features. The target lithofacies identification model is determined based on the comparison results so that lithofacies identification can be completed using the target lithofacies identification model.
[0010] Optionally, obtaining the lithofacies type corresponding to the core segment based on the acquired core description data includes:
[0011] Based on the obtained core description data, the lithology and bedding type of the core section are determined, and the corresponding lithofacies type of the core section is obtained according to the obtained lithology and bedding type of the core section.
[0012] Optionally, determining the target probability distribution characteristics of the vertical lithofacies based on the lithofacies type corresponding to the core segment includes:
[0013] A preset number of cored segments are obtained, and the corresponding lithofacies type and the number of lithofacies types are determined based on the corresponding cored segments. The target probability distribution characteristics of the lithofacies in the vertical direction are determined based on the preset number and the number of lithofacies types.
[0014] Optionally, the step of analyzing the heterogeneous characteristics of the target lithofacies structure using the lithofacies assemblage features includes:
[0015] The lithofacies density and lithofacies frequency in the lithofacies assemblage characteristics corresponding to the core segment are statistically analyzed, and the heterogeneous characteristics of the target lithofacies structure are analyzed based on the lithofacies density and lithofacies frequency.
[0016] Optionally, the step of normalizing the logging curves of the target formation to obtain processed logging curves of the target formation includes:
[0017] Obtain the values in the target layer logging curve that satisfy the preset maximum value condition and the preset minimum value condition. Based on the values that satisfy the preset maximum value condition and the preset minimum value condition, normalize the target layer logging curve to obtain the processed target layer logging curve.
[0018] Optionally, determining the target logging curve based on the processed target layer logging curve and lithofacies type includes:
[0019] Sensitivity analysis was performed using Pearson correlation coefficient analysis between the core logging curve of the processed target layer and the lithofacies type, and between the logging curves of each processed target layer, to obtain the target logging curve.
[0020] Optionally, obtaining the target layer data between the target half-amplitude points based on the target weight value includes:
[0021] Based on the target weight value, the curve data between the target half-amplitude points is determined, and the target layer data between the target half-amplitude points is determined based on the curve data between the target half-amplitude points.
[0022] Optionally, the step of building model training data based on the target layer data between the obtained target half-width points includes:
[0023] The target layer data between the target half-width points is used to determine the data that meets the preset minimum sample layer unit. The depth slice unit data is obtained by splicing the data that meets the preset minimum sample layer unit. The average depth value of the depth slice unit data is determined to obtain model training data containing vertical information.
[0024] Optionally, training the initial lithofacies identification model using the model training data to obtain the model prediction result includes:
[0025] Based on the model training data, the current training set and the current validation set are determined. The target probability distribution features and the current training set are input into the initial lithofacies identification model for iterative updates to obtain the iterated lithofacies identification model. The depth slice data to be predicted is input into the iterated lithofacies identification model to obtain the model prediction result.
[0026] Secondly, this application provides a petrographic identification device, comprising:
[0027] The feature analysis module is used to obtain the lithofacies type corresponding to the core segment based on the obtained core description data of the core well, determine the lithofacies combination characteristics corresponding to the core segment according to the obtained lithofacies type, determine the target probability distribution characteristics of the vertical lithofacies according to the lithofacies type, and analyze the heterogeneous characteristics of the target lithofacies structure using the lithofacies combination characteristics.
[0028] The well logging curve determination module is used to acquire the well logging curve of the target layer, normalize the well logging curve of the target layer to obtain the processed well logging curve of the target layer, and then determine the target well logging curve based on the processed well logging curve of the target layer and the lithofacies type.
[0029] The data acquisition module is used to determine the first depth value of the target peak point and the second depth value of the target valley point corresponding to the target logging curve, obtain the third depth value of the target half-amplitude point based on the first depth value and the second depth value, determine the corresponding target weight value using the third depth value of the target half-amplitude point, and obtain the target layer data between the target half-amplitude points according to the target weight value.
[0030] The lithofacies identification module is used to establish model training data based on the target layer data between the obtained target half-width points, train the initial lithofacies identification model using the model training data to obtain model prediction results, compare the predicted probability distribution features in the obtained model prediction results with the target probability distribution features, compare the predicted lithofacies structural heterogeneity features in the obtained model prediction results with the target lithofacies structural heterogeneity features, and determine the target lithofacies identification model based on the comparison results so as to complete lithofacies identification using the target lithofacies identification model.
[0031] Thirdly, this application provides an electronic device, comprising:
[0032] Memory, used to store computer programs;
[0033] A processor is used to execute the computer program to implement the lithofacies identification method as described above.
[0034] Fourthly, this application provides a computer-readable storage medium for storing a computer program; wherein, when the computer program is executed by a processor, it implements the aforementioned petrographic identification method.
[0035] In summary, this application first obtains the lithofacies type corresponding to the core section based on the obtained core description data, determines the lithofacies assemblage characteristics corresponding to the core section based on the obtained lithofacies type, determines the target probability distribution characteristics of the vertical lithofacies based on the lithofacies type corresponding to the core section, and analyzes the heterogeneous characteristics of the target lithofacies structure using the lithofacies assemblage characteristics; obtains the target interval logging curve, normalizes the target interval logging curve to obtain the processed target interval logging curve, and then determines the target logging curve based on the processed target interval logging curve and the lithofacies type; determines the first depth value of the target peak point and the second depth value of the target valley point corresponding to the target logging curve, and determines the target logging curve based on the first depth value and the second depth value. The process involves obtaining the third depth value of a target half-width point, determining the corresponding target weight value using this value, and obtaining target layer data between the target half-width points based on the target weight value. Model training data is then established based on this target layer data, and the initial lithofacies identification model is trained using this data to obtain model prediction results. The predicted probability distribution features in the model prediction results are compared with the target probability distribution features, and the predicted lithofacies structural heterogeneity features in the model prediction results are compared with the target lithofacies structural heterogeneity features. Based on the comparison results, a target lithofacies identification model is determined to complete lithofacies identification. As can be seen from the above, this application first obtains core description data from the core well to determine the lithofacies type corresponding to the core section. Based on this lithofacies type, the lithofacies assemblage characteristics corresponding to the core section are further clarified, and the target probability distribution characteristics of the vertical lithofacies are determined. Then, the lithofacies assemblage characteristics are used to analyze the heterogeneity characteristics of the target lithofacies structure. Subsequently, the logging curves of the target formation are acquired and normalized to obtain the processed logging curves of the target formation. Based on the processed logging curves and the lithofacies type, the target logging curve is determined. The first depth value of the target peak and the second depth value of the target valley in the target logging curve are determined. Based on these two depth values, the third depth value of the target half-amplitude point is obtained. The corresponding target weight value is determined using the third depth value, and then the target layer data between the target half-amplitude points is obtained based on the target weight value. Next, model training data is established based on the obtained target layer data. The initial lithofacies identification model is trained using the model training data to obtain the model prediction results. The predicted probability distribution characteristics in the model prediction results are compared with the previously determined target probability distribution characteristics. Simultaneously, the predicted lithofacies structural heterogeneity characteristics are also compared with the target lithofacies structural heterogeneity characteristics. Finally, based on these comparison results, the target lithofacies identification model is determined so that it can be used to complete the lithofacies identification work subsequently.In this way, based on the vertical development probability characteristics of different lithofacies within a single lithofacies assemblage, the heterogeneous characteristics of lithofacies structure, and the corresponding well logging response characteristics, lithofacies slices from vertical depth sequences can be used in conjunction with deep learning models for adaptive and refined interpretation of lithofacies. This can improve the interpretation accuracy of complex lithofacies and incorporate the development patterns of vertical lithofacies, laying a solid foundation for further sedimentary geological interpretation. Attached Figure Description
[0036] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.
[0037] Figure 1 This is a flowchart of a petrographic identification method disclosed in this application;
[0038] Figure 2 This application discloses a statistical diagram showing the target probability distribution characteristics of the previous lithofacies of the current lithofacies.
[0039] Figure 3 This application discloses a statistical chart showing the target probability distribution characteristics of the next lithofacies of the current lithofacies.
[0040] Figure 4 This application discloses a statistical diagram of the heterogeneous characteristics of lithofacies structure related to lithofacies density.
[0041] Figure 5 This application discloses a statistical chart of heterogeneous characteristics of lithofacies structure based on lithofacies frequency.
[0042] Figure 6 This is a schematic diagram illustrating the sensitivity analysis between core section logging curves and lithofacies types disclosed in this application.
[0043] Figure 7 This is a schematic diagram illustrating the sensitivity analysis between coring section logging curves and logging curves disclosed in this application.
[0044] Figure 8 This is a schematic diagram illustrating the establishment of a sample layer database based on well logging curves disclosed in this application;
[0045] Figure 9 This is a schematic diagram illustrating the establishment of a petrographic tag library based on depth slices as disclosed in this application;
[0046] Figure 10 This is a schematic diagram illustrating the training of a bidirectional long short-term memory neural network model disclosed in this application.
[0047] Figure 11 This is a schematic diagram of a single-well lithofacies identification result disclosed in this application;
[0048] Figure 12 This is a flowchart of a specific petrographic identification method disclosed in this application;
[0049] Figure 13 This is a schematic diagram of the structure of a petrographic identification device disclosed in this application;
[0050] Figure 14 This is a structural diagram of an electronic device disclosed in this application. Detailed Implementation
[0051] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0052] Currently, there are two main categories of methods for lithofacies interpretation in geophysical logging: one is conventional logging interpretation, which relies on core observation and thin section analysis combined with qualitative and quantitative logging methods under the guidance of sedimentary models. This method largely depends on the knowledge and experience of the interpreters, resulting in relatively slow interpretation progress and low accuracy. The second category is logging lithofacies interpretation based on artificial intelligence algorithms, which mainly uses supervised, unsupervised, or semi-supervised methods for intelligent lithofacies interpretation. Although this method has been widely used in logging interpretation, most methods only consider the relationship between sample labels and logging response characteristics, lacking constraints from geological concepts or laws, and failing to explore the intrinsic connection between geological laws and artificial intelligence algorithms. Consequently, machine learning methods are merely mathematical algorithms, lacking mechanistic understanding. To address the above technical problems, this application discloses a lithofacies identification method, device, equipment, and storage medium, capable of identifying complex lithofacies by combining intelligent algorithms.
[0053] See Figure 1 As shown, an embodiment of the present invention discloses a method for identifying petrographic features, including:
[0054] Step S11: Based on the obtained core description data of the core well, obtain the lithofacies type corresponding to the core section, determine the lithofacies combination characteristics corresponding to the core section according to the obtained lithofacies type, determine the target probability distribution characteristics of the vertical lithofacies according to the lithofacies type, and analyze the heterogeneous characteristics of the target lithofacies structure using the lithofacies combination characteristics.
[0055] In this embodiment, the lithology and bedding type of the cored sections are first determined based on the obtained core description data from the core wells. Then, the corresponding lithofacies type of the cored sections is obtained based on the obtained lithology and bedding type. Specifically, the lithology and bedding type of the cored sections are determined based on the core description data from the core wells in the study area. Based on the determined lithology and bedding type, the lithofacies type of the cored sections is further determined as shown in Table 1. For example, a specific lithofacies type code is obtained based on the lithology and bedding type. Next, the lithofacies assemblage characteristics are summarized based on the different lithofacies types in the vertical direction within a single continuous cored section.
[0056] Table 1
[0057] Furthermore, it is necessary to obtain a predetermined number of cored segments, determine the corresponding lithofacies type and the number of lithofacies types based on the corresponding cored segments, and determine the target probability distribution characteristics of the lithofacies in the vertical direction based on the predetermined number and the number of lithofacies types. Specifically, a heat map is used to display the target probability distribution characteristics of a certain lithofacies development above and below the current lithofacies in the vertical direction, and the analysis yields results such as... Figure 2 The target probability distribution characteristics of the previous lithofacies shown are as follows: Figure 3 The probability distribution characteristics of the next target lithofacies shown are as follows:
[0058] In one specific implementation, assuming there are M single-layer cored segments, a specific lithofacies is selected as the research object, and the target probability distribution characteristics of lithofacies development are statistically analyzed. The vertical lithofacies distribution of a single layer can be represented as follows: , where n is the total number of lithofacies in a single core section, and the lithofacies type can be represented as Assuming any lithofacies type The number of occurrences of a given lithofacies assemblages is N, where T represents any lithofacies. The number of adjacent facies of a specific lithofacies, then any lithofacies The probability P of the occurrence of this specific lithofacies in the upper and lower parts is:
[0059] ;
[0060] Where P is the probability of a specific lithofacies; M is the number of cored data segments; N is the number of times any lithofacies type appears in M lithofacies combinations; and T is the number of adjacent specific lithofacies above and below any lithofacies.
[0061] Next, it is necessary to statistically analyze the lithofacies density and frequency within the lithofacies assemblage corresponding to the cored segment, and then analyze the heterogeneous characteristics of the target lithofacies structure based on these density and frequency. Specifically, this involves statistically analyzing the differences within the lithofacies assemblage of a complete single-layer cored segment, such as... Figure 4 The lithological density shown, such as Figure 5The lithofacies frequencies shown are used to quantitatively analyze the heterogeneous characteristics of the target lithofacies structure. Here, lithofacies density is the thickness of a single lithofacies layer or a single layer; lithofacies frequency is the number of occurrences of a certain type of lithofacies within a single lithofacies assemblage.
[0062] Step S12: Obtain the target layer logging curve, normalize the target layer logging curve to obtain the processed target layer logging curve, and then determine the target logging curve based on the processed target layer logging curve and the lithofacies type.
[0063] In this embodiment, it is first necessary to obtain the values of the target segment logging curves that meet the preset maximum value condition and the preset minimum value condition. Based on the values that meet the preset maximum value condition and the preset minimum value condition, the target segment logging curves are normalized to obtain the processed target segment logging curves. Specifically, the target segment logging curves can first be extracted, and the target segment logging curves of each well can be standardized to unify the scale. Then, the maximum and minimum values of the logging curves of different target segments are obtained respectively. Based on the maximum and minimum values of the target segment logging curves, the min-max normalization method is used to normalize the different curves, thereby mapping the response values contained in each target segment logging curve to the range [0-1], thus obtaining the processed target segment logging curves.
[0064] Then, Pearson correlation coefficient analysis was used to perform sensitivity analysis between the core logging curves and lithofacies types in the processed target layer logging curves, and sensitivity analysis between the logging curves of each processed target layer, to obtain the target logging curves. Specifically, Pearson correlation coefficient analysis was used to perform the following... Figure 6 Sensitivity analysis of the correlation between the core logging curves and lithofacies type, as shown in the figure. Figure 7 The sensitivity analysis between the cored section logging curves and other logging curves shows that logging curves with high correlation coefficients with lithofacies and low correlation coefficients among themselves are preferred as target logging curves for model training. It is important to understand that GR (Gamma Ray), AC (Acoustic Time), LLD (Laterolog-Deep), and PE (Photoelectric Absorption Index) can be selected as logging sensitivity parameters, and the logging curves corresponding to these parameters can be obtained as target logging curves.
[0065] Step S13: Determine the first depth value of the target peak point and the second depth value of the target valley point corresponding to the target logging curve; obtain the third depth value of the target half-amplitude point based on the first depth value and the second depth value; determine the corresponding target weight value using the third depth value of the target half-amplitude point; and obtain the target layer data between the target half-amplitude points based on the target weight value.
[0066] In this embodiment, after determining the target logging curve, the value of the logging curve is first differentiated using the natural logarithm function to obtain the slope of the curve at a certain point, i.e. The slope of the curve is negative, that is... If the curve changes from rising to falling, and the trend of its derivative changes from positive to negative, this is identified as the target peak point; if the curve changes from falling to rising, and the trend of its derivative changes from negative to positive, this is identified as the target trough point. Figure 8 As shown, local peaks and valleys are determined by comparing the sizes of adjacent data points. Next, the first depth value of the target peak and the second depth value of the target valley are calculated, and the average value between the target peak and the target valley is used as the depth value of the target half-amplitude point.
[0067] Furthermore, after obtaining the depth values of the target half-width points, the curve data between the target half-width points is determined based on the target weight values, and the target layer data between the target half-width points is determined based on the curve data between the target half-width points. Specifically, the distance between the peak point within the range of two target half-width points and any depth point within that range is used as a weight parameter, and the weight value w is calculated using the inverse distance weighting method. i :
[0068] ;
[0069] in, This represents the weight value of any depth point i between half-width points; H represents the depth value of any depth point i between half-width points; H represents the depth value of the peak point between half-width points.
[0070] Adjust the weights for different depths, and use the following formula to calculate the weighted average of the curve data between the target half-width points, thus obtaining the corresponding weighted average, which is the target layer data between the target half-width points:
[0071] ;
[0072] in This represents the weighted average curve value of the data layers between half-amplitude points; This represents the weight value of any depth point i between half-width points; This represents the curve value at any depth point i between half-width points; n is the number of depth points.
[0073] Step S14: Based on the target layer data between the obtained target half-width points, establish model training data, use the model training data to train the initial lithofacies identification model to obtain model prediction results, compare the predicted probability distribution features in the obtained model prediction results with the target probability distribution features, and compare the predicted lithofacies structural heterogeneity features in the obtained model prediction results with the target lithofacies structural heterogeneity features, determine the target lithofacies identification model according to the comparison results, so as to complete lithofacies identification using the target lithofacies identification model.
[0074] In this embodiment, after determining the target layer data between each target half-frame point, the target layer data between the target half-frame points is used to determine the data that meets the preset minimum sample layer unit. Depth slice unit data is obtained by splicing the data that meets the preset minimum sample layer unit, and the average depth value of the depth slice unit data is determined to obtain model training data containing vertical information. Specifically, the target layer data determined using the target half-frame points is used as the minimum sample layer unit. Depth slice unit data is constructed by splicing these minimum sample layer unit data. During the depth slicing operation, I minimum sample layer unit data (I=1, 2, 3, ..., 10) are sequentially spliced along the vertical depth direction. Different depth slice unit data can be obtained as the values change. The mean curve value corresponding to each depth slice unit data can be calculated using the thickness-weighted average method. The final value depends on the accuracy required for model training. Simultaneously, by statistically analyzing the average thickness of the minimum sample layer unit, the thickness of a single lithofacies combination, and the lithofacies density, a reasonable range for the minimum sample layer unit can be determined, which is instructive for training the initial lithofacies identification model. After calculating the average depth value of the depth slice data, the following was obtained: Figure 9 The depth slice label library containing vertical information shown serves as the model training data. Furthermore, for the task of identifying lithofacies units with probabilistic distribution characteristics in the vertical direction, BiLSTM (Bidirectional Long Short-Term Memory) was selected as the initial lithofacies identification model. This neural network performs excellently in processing time-series data, effectively maintaining long-term memory. Moreover, its model incorporates a gating mechanism, which addresses the vanishing and exploding gradient problems, significantly reducing the time and difficulty required for model training and providing strong support for model training.
[0075] Furthermore, after obtaining the model training data, the current training set and the current validation set can be determined based on the model training data. The target probability distribution features and the current training set are input into the initial lithofacies identification model for iterative updates to obtain the iterative lithofacies identification model. The depth slice data to be predicted is then input into the iterative lithofacies identification model to obtain the model prediction result. Specifically, as shown... Figure 10 As shown, firstly, a depth slice sample label library with different I values is used as the current training set and the current validation set. For example, 75% of the data in the depth slice label library can be randomly selected as the current training set, and 25% as the current validation set. Then, the target lithofacies probability distribution characteristics and the current training set data are input into the initial lithofacies recognition model. The model is trained by adjusting the parameters of the initial lithofacies recognition model, and then validated using the current validation set. By continuously updating the density and frequency of the predicted lithofacies within the lithofacies assemblage and calculating the error of the frequency and density of similar lithofacies in the core segment, the final result and model accuracy are obtained. Simultaneously, I in the depth slices is adjusted to obtain new model training data. The model is repeatedly trained using this new training data to obtain the iterative lithofacies recognition model. Furthermore, by continuously updating the label library and adjusting the model parameters, the optimal lithofacies label library can be obtained, and the number of combinations of the smallest sample layer unit I can be determined. Next, the depth slice data to be predicted is input into the iterative lithofacies recognition model. The model framework is used to process the data of the depth slice data to be predicted, ultimately yielding the result shown below. Figure 11 The model prediction results are shown.
[0076] Finally, the predicted probability distribution characteristics of vertical lithofacies in the statistical model prediction results are used to verify whether they match the probability distribution characteristics of the target lithofacies in the core section. At the same time, based on the predicted lithofacies heterogeneity characteristics in the prediction results, the model is used to verify whether it matches the heterogeneity of the target lithofacies structure, and corresponding comparison results are obtained. Based on the obtained lithofacies probability distribution characteristics and the comparison results of the internal structural heterogeneity of lithofacies, the target lithofacies identification model is finally determined so that different lithofacies can be identified using the target lithofacies identification model.
[0077] As described above, this embodiment first obtains core description data from the cored well to determine the lithofacies type corresponding to the cored section. Based on this lithofacies type, the lithofacies assemblage characteristics corresponding to the cored section are further clarified, and the target probability distribution characteristics of the lithofacies in the vertical direction are determined. Then, the lithofacies assemblage characteristics are used to analyze the heterogeneous characteristics of the target lithofacies structure. Subsequently, the logging curves of the target layer are obtained and normalized to obtain the processed logging curves of the target layer. The target logging curve is determined based on the processed logging curves and the lithofacies type. The first depth value of the target peak point and the second depth value of the target valley point in the target logging curve are determined. Based on these two depth values, the third depth value of the target half-amplitude point is obtained. The corresponding target weight value is determined using the third depth value, and then the target layer data between the target half-amplitude points is obtained based on the target weight value. Afterward, model training data is established based on the obtained target layer data, and the initial lithofacies identification model is trained using the model training data to obtain the model prediction results. The predicted probability distribution features in the model's prediction results are compared with the previously determined target probability distribution features. Simultaneously, the predicted heterogeneous features of lithofacies structure are also compared with the heterogeneous features of the target lithofacies structure. Based on these comparisons, a target lithofacies identification model is determined, which is then used to complete subsequent lithofacies identification work. In this way, based on the vertical development probability features of different lithofacies within a single lithofacies assemblage, the heterogeneous features of lithofacies structure, and the corresponding well logging response features, and using lithofacies slices from vertical depth sequences combined with a deep learning model for adaptive and refined lithofacies interpretation, the interpretation accuracy of complex lithofacies can be improved. Furthermore, the development patterns of vertical lithofacies are incorporated, laying a solid foundation for further sedimentary geological interpretation.
[0078] As can be seen from the previous embodiment, this application discloses a lithofacies identification method that can identify complex lithofacies by combining intelligent algorithms. Next, we will address methods such as... Figure 12 As shown, the specific methods for identifying lithofacies are explained in detail.
[0079] This application first determines the lithology and bedding type of the cored sections based on core description data from the core wells in the study area. Then, based on the determined lithology and bedding type, the lithofacies type of the cored sections is further determined. Next, a heat map is used to display the target probability distribution characteristics of a certain lithofacies developing vertically above and below the current lithofacies, and the target probability distribution characteristics of the previous lithofacies and the next target lithofacies are analyzed. Simultaneously, the different lithofacies densities and lithofacies frequencies within the lithofacies assemblage of a complete single-layer cored section are statistically analyzed to quantitatively determine the heterogeneous characteristics of the target lithofacies structure.
[0080] Furthermore, after determining the target probability distribution characteristics and the heterogeneous characteristics of the target lithofacies structure, the target interval logging curves are extracted. The target interval logging curves for each well are standardized to unify the scale. Then, the maximum and minimum values of the logging curves for different target intervals are obtained. Based on these maximum and minimum values, the min-max normalization method is used to normalize the different curves, thus mapping the response values contained in each target interval logging curve to the range [0-1], resulting in the processed target interval logging curves. Subsequently, Pearson correlation coefficient analysis is used to perform sensitivity analyses between the cored section logging curves and lithofacies types, and between logging curves themselves. The target logging curves are determined based on the analysis results.
[0081] Next, after determining the target logging curve, the natural logarithm function is used to differentiate the logging curve values to determine the target peak and valley points, and the depth values of the target peak and valley points are obtained. Based on the depth values of the target peak and valley points, the depth values of the target half-amplitude points are obtained. After obtaining the depth values of the target half-amplitude points, the curve data between the target half-amplitude points is determined according to the target weight value, and the target layer data between the target half-amplitude points is determined based on the curve data between the target half-amplitude points.
[0082] Finally, after determining the target layer data between each target half-frame point, the target layer data determined using the target half-frame points is used as the smallest sample layer unit. By stitching together these smallest sample layer unit data, depth slice unit data is constructed. After statistically analyzing the average depth value of the depth slice unit data, a depth slice label library containing vertical information is obtained, which is the model training data. After obtaining the model training data, the depth slice sample label library with different target weight values is used as the current training set and the current validation set. The target lithofacies probability distribution characteristics and the current training set data are input into the initial lithofacies recognition model. The model training is completed by adjusting the parameters of the initial lithofacies recognition model, resulting in the iterative lithofacies recognition model. Then, the depth slice data to be predicted is input into the iterative lithofacies recognition model, and the model framework is used to process the data of the depth slice data to be predicted, finally obtaining the corresponding model prediction results. The predicted probability distribution characteristics of vertical lithofacies in the statistical model prediction results are used to verify whether they match the probability distribution characteristics of the target lithofacies in the cored segment. Simultaneously, based on the predicted lithofacies heterogeneity characteristics in the prediction results, the model is used to verify whether it matches the structural heterogeneity of the target lithofacies, yielding corresponding comparison results. Based on the obtained lithofacies probability distribution characteristics and the comparison results of the internal structural heterogeneity of the lithofacies, the target lithofacies identification model is finally determined.
[0083] This application comprehensively considers the development characteristics of different lithofacies in the vertical direction and the logging response under the influence of these characteristics. It utilizes the learning advantages of neural networks for vertical sequences and their long-term memory characteristics to perform adaptive intelligent identification of complex lithofacies, so as to achieve a detailed interpretation of lithofacies.
[0084] See Figure 13 As shown, an embodiment of the present invention discloses a petrographic identification device, which may include:
[0085] Feature analysis module 11 is used to obtain the lithofacies type corresponding to the core segment based on the obtained core description data of the core well, determine the lithofacies combination characteristics corresponding to the core segment according to the obtained lithofacies type, determine the target probability distribution characteristics of the vertical lithofacies according to the lithofacies type, and analyze the heterogeneous characteristics of the target lithofacies structure using the lithofacies combination characteristics.
[0086] The logging curve determination module 12 is used to acquire the logging curve of the target layer, normalize the logging curve of the target layer to obtain the processed logging curve of the target layer, and then determine the target logging curve based on the processed logging curve of the target layer and the lithofacies type.
[0087] The data acquisition module 13 is used to determine the first depth value of the target peak point and the second depth value of the target valley point corresponding to the target logging curve, obtain the third depth value of the target half-amplitude point based on the first depth value and the second depth value, determine the corresponding target weight value using the third depth value of the target half-amplitude point, and obtain the target layer data between the target half-amplitude points according to the target weight value.
[0088] The lithofacies identification module 14 is used to establish model training data based on the target layer data between the obtained target half-width points, train the initial lithofacies identification model using the model training data to obtain model prediction results, compare the predicted probability distribution features in the obtained model prediction results with the target probability distribution features, compare the predicted lithofacies structural heterogeneity features in the obtained model prediction results with the target lithofacies structural heterogeneity features, and determine the target lithofacies identification model based on the comparison results so as to complete lithofacies identification using the target lithofacies identification model.
[0089] As described above, this application first obtains core description data from the cored well to determine the lithofacies type corresponding to the cored section. Based on this lithofacies type, the lithofacies assemblage characteristics corresponding to the cored section are further clarified, and the target probability distribution characteristics of the lithofacies in the vertical direction are determined. Then, the lithofacies assemblage characteristics are used to analyze the heterogeneous characteristics of the target lithofacies structure. Subsequently, the logging curves of the target interval are obtained and normalized to obtain the processed logging curves of the target interval. The target logging curves are determined based on the processed logging curves and the lithofacies type. The first depth value of the target peak point and the second depth value of the target valley point in the target logging curve are determined. Based on these two depth values, the third depth value of the target half-amplitude point is obtained. The corresponding target weight value is determined using the third depth value, and then the target layer data between the target half-amplitude points is obtained based on the target weight value. Afterwards, model training data is established based on the obtained target layer data, and the initial lithofacies identification model is trained using the model training data to obtain the model prediction results. The predicted probability distribution features in the model's prediction results are compared with the previously determined target probability distribution features. Simultaneously, the predicted heterogeneous features of lithofacies structure are also compared with the heterogeneous features of the target lithofacies structure. Based on these comparisons, a target lithofacies identification model is determined, which is then used to complete subsequent lithofacies identification work. In this way, based on the vertical development probability features of different lithofacies within a single lithofacies assemblage, the heterogeneous features of lithofacies structure, and the corresponding well logging response features, and using lithofacies slices from vertical depth sequences combined with a deep learning model for adaptive and refined lithofacies interpretation, the interpretation accuracy of complex lithofacies can be improved. Furthermore, the development patterns of vertical lithofacies are incorporated, laying a solid foundation for further sedimentary geological interpretation.
[0090] In some specific implementations, the feature analysis module 11 includes:
[0091] The lithofacies type acquisition unit is used to determine the lithology and bedding type of the cored section based on the acquired core description data, and to obtain the lithofacies type corresponding to the cored section based on the obtained lithology and bedding type of the cored section.
[0092] In some specific implementations, the feature analysis module 11 includes:
[0093] The target probability distribution feature determination unit is used to acquire a preset number of core segment data, determine the corresponding lithofacies type and the number of lithofacies types according to the corresponding core segment, and determine the target probability distribution feature of the lithofacies in the vertical direction based on the preset number and the number of lithofacies types.
[0094] In some specific implementations, the feature analysis module 11 includes:
[0095] The target lithofacies structure heterogeneous feature determination unit is used to statistically analyze the lithofacies density and lithofacies frequency in the lithofacies assemblage features corresponding to the core segment, and to analyze the target lithofacies structure heterogeneous features based on the lithofacies density and lithofacies frequency.
[0096] In some specific embodiments, the well logging curve determination module 12 includes:
[0097] The target segment logging curve acquisition unit is used to acquire the values of the target segment logging curve that meet the preset maximum value condition and the preset minimum value condition, and to normalize the target segment logging curve based on the values that meet the preset maximum value condition and the preset minimum value condition to obtain the processed target segment logging curve.
[0098] In some specific embodiments, the well logging curve determination module 12 includes:
[0099] The target logging curve acquisition unit is used to perform sensitivity analysis between the core logging curve in the processed target layer logging curve and the lithofacies type, and sensitivity analysis between the logging curves of each processed target layer, using Pearson correlation coefficient analysis, so as to obtain the target logging curve.
[0100] In some specific embodiments, the data acquisition module 13 includes:
[0101] The target layer data determination unit is used to determine the curve data between the target half-amplitude points according to the target weight value, and to determine the target layer data between the target half-amplitude points based on the curve data between the target half-amplitude points.
[0102] In some specific embodiments, the lithofacies identification module 14 includes:
[0103] The model training data acquisition unit is used to determine the data that satisfies the preset minimum sample layer unit by using the target layer data between the target half-frame points, obtain depth slice unit data by splicing the data that satisfies the preset minimum sample layer unit, and determine the average depth value of the depth slice unit data to obtain model training data containing vertical information.
[0104] In some specific embodiments, the lithofacies identification module 14 includes:
[0105] The model prediction result acquisition unit is used to determine the current training set and the current validation set based on the model training data, input the target probability distribution features and the current training set into the initial lithofacies identification model for iterative update to obtain the iterated lithofacies identification model, and input the depth slice data to be predicted into the iterated lithofacies identification model to obtain the model prediction result.
[0106] Furthermore, embodiments of this application also disclose an electronic device, Figure 14 This is a structural diagram of an electronic device 20 according to an exemplary embodiment. The content of the diagram should not be construed as limiting the scope of this application.
[0107] Figure 14 This is a schematic diagram of the structure of an electronic device 20 provided in an embodiment of this application. Specifically, the electronic device 20 may include: at least one processor 21, at least one memory 22, a power supply 23, a communication interface 24, an input / output interface 25, and a communication bus 26. The memory 22 stores a computer program, which is loaded and executed by the processor 21 to implement the relevant steps in the petrographic identification method disclosed in any of the foregoing embodiments. Alternatively, the electronic device 20 in this embodiment may specifically be an electronic computer.
[0108] In this embodiment, the power supply 23 is used to provide operating voltage for each hardware device on the electronic device 20; the communication interface 24 can create a data transmission channel between the electronic device 20 and external devices, and the communication protocol it follows can be any communication protocol applicable to the technical solution of this application, and is not specifically limited here; the input / output interface 25 is used to acquire external input data or output data to the outside world, and its specific interface type can be selected according to specific application needs, and is not specifically limited here.
[0109] In addition, the memory 22, as a carrier for resource storage, can be a read-only memory, random access memory, disk or optical disk, etc. The resources stored thereon can include operating system 221, computer program 222, etc., and the storage method can be temporary storage or permanent storage.
[0110] The operating system 221 is used to manage and control the various hardware devices on the electronic device 20 and the computer program 222, which may be Windows Server, Netware, Unix, Linux, etc. In addition to including a computer program capable of performing the petrographic identification method executed by the electronic device 20 as disclosed in any of the foregoing embodiments, the computer program 222 may further include a computer program capable of performing other specific tasks.
[0111] Furthermore, this application also discloses a computer-readable storage medium for storing a computer program; wherein, when the computer program is executed by a processor, it implements the aforementioned petrographic identification method. Specific steps of this method can be found in the corresponding content disclosed in the foregoing embodiments, and will not be repeated here.
[0112] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to in the method section.
[0113] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.
[0114] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly by hardware, a software module executed by a processor, or a combination of both. The software module can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.
[0115] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.
[0116] The technical solutions provided in this application have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.
Claims
1. A method of lithofacies identification, characterized by, include: Based on the obtained core description data of the core well, the lithofacies type corresponding to the core section is obtained. Based on the obtained lithofacies type corresponding to the core section, the lithofacies combination characteristics corresponding to the core section are determined. Based on the lithofacies type corresponding to the core section, the target probability distribution characteristics of the vertical lithofacies are determined. The heterogeneous characteristics of the target lithofacies structure are analyzed using the lithofacies combination characteristics. Obtain the target layer logging curve, normalize the target layer logging curve to obtain the processed target layer logging curve, and then determine the target logging curve based on the processed target layer logging curve and the lithofacies type. The first depth value of the target peak point and the second depth value of the target valley point corresponding to the target logging curve are determined. Based on the first depth value and the second depth value, the third depth value of the target half-amplitude point is obtained. The corresponding target weight value is determined using the third depth value of the target half-amplitude point. The target layer data between the target half-amplitude points is obtained according to the target weight value. Based on the target layer data between the obtained target half-width points, model training data is established. The initial lithofacies identification model is trained using the model training data to obtain model prediction results. The predicted probability distribution features in the obtained model prediction results are compared with the target probability distribution features. The predicted lithofacies structural heterogeneity features in the obtained model prediction results are compared with the target lithofacies structural heterogeneity features. The target lithofacies identification model is determined based on the comparison results so that lithofacies identification can be completed using the target lithofacies identification model. The step of analyzing the heterogeneous characteristics of the target lithofacies structure using the lithofacies assemblage features includes: The lithofacies density and lithofacies frequency in the lithofacies assemblage characteristics corresponding to the core segment are statistically analyzed, and the heterogeneous characteristics of the target lithofacies structure are analyzed based on the lithofacies density and lithofacies frequency. The step of establishing model training data based on the target layer data between the obtained target half-width points includes: The target layer data between the target half-width points is used as the minimum sample layer unit. The target number of minimum sample layer units are sequentially spliced and combined along the vertical direction to construct the depth slice unit data. The average depth value of the depth slice unit data is determined, thereby obtaining a depth slice label library containing vertical information as model training data. The lithofacies identification method includes: Based on the deep slice label library, a training set and a validation set are determined. The target probability distribution features and the training set are input into the initial lithofacies identification model for iterative training. During the training process, the number of targets in the smallest sample layer unit is continuously adjusted to obtain new model training data. The model is then repeatedly trained using the new model training data until the probability distribution features of the model prediction results and the heterogeneous features of the lithofacies structure match the target lithofacies probability distribution features of the cored segment, thus determining the target lithofacies identification model.
2. The lithofacies identification method according to claim 1, characterized in that, The process of obtaining the lithofacies type corresponding to the core section based on the acquired core description data includes: Based on the obtained core description data, the lithology and bedding type of the core section are determined, and the corresponding lithofacies type of the core section is obtained according to the obtained lithology and bedding type of the core section.
3. The lithofacies identification method of claim 1, wherein, The determination of the target probability distribution characteristics of vertical lithofacies based on the lithofacies type corresponding to the core section includes: A preset number of cored segments are obtained, and the corresponding lithofacies type and the number of lithofacies types are determined based on the corresponding cored segments. The target probability distribution characteristics of the lithofacies in the vertical direction are determined based on the preset number and the number of lithofacies types.
4. The lithofacies identification method of claim 1, wherein, The step of normalizing the logging curves of the target formation to obtain the processed logging curves of the target formation includes: Obtain the values in the target layer logging curve that satisfy the preset maximum value condition and the preset minimum value condition. Based on the values that satisfy the preset maximum value condition and the preset minimum value condition, normalize the target layer logging curve to obtain the processed target layer logging curve.
5. The lithofacies identification method of claim 1, wherein, The determination of the target logging curve based on the processed target layer logging curve and lithofacies type includes: Sensitivity analysis was performed using Pearson correlation coefficient analysis between the core logging curve of the processed target layer and the lithofacies type, and between the logging curves of each processed target layer, to obtain the target logging curve.
6. The lithofacies identification method of claim 1, wherein, The step of obtaining the target layer data between the target half-amplitude points based on the target weight value includes: Based on the target weight value, the curve data between the target half-amplitude points is determined, and the target layer data between the target half-amplitude points is determined based on the curve data between the target half-amplitude points.
7. The lithofacies identification method according to any one of claims 1 to 6, characterized in that, The model training data built based on the target layer data between the obtained target half-width points includes: The target layer data between the target half-width points is used to determine the data that meets the preset minimum sample layer unit. The depth slice unit data is obtained by splicing the data that meets the preset minimum sample layer unit. The average depth value of the depth slice unit data is determined to obtain model training data containing vertical information.
8. The lithofacies identification method according to claim 7, characterized in that, The step of training the initial lithofacies identification model using the model training data to obtain the model prediction results includes: Based on the model training data, the current training set and the current validation set are determined. The target probability distribution features and the current training set are input into the initial lithofacies identification model for iterative updates to obtain the iterated lithofacies identification model. The depth slice data to be predicted is input into the iterated lithofacies identification model to obtain the model prediction result.
9. A petrographic identification device, characterized in that, include: The feature analysis module is used to obtain the lithofacies type corresponding to the core segment based on the obtained core description data of the core well, determine the lithofacies combination characteristics corresponding to the core segment according to the obtained lithofacies type, determine the target probability distribution characteristics of the vertical lithofacies according to the lithofacies type, and analyze the heterogeneous characteristics of the target lithofacies structure using the lithofacies combination characteristics. The well logging curve determination module is used to acquire the well logging curve of the target layer, normalize the well logging curve of the target layer to obtain the processed well logging curve of the target layer, and then determine the target well logging curve based on the processed well logging curve of the target layer and the lithofacies type. The data acquisition module is used to determine the first depth value of the target peak point and the second depth value of the target valley point corresponding to the target logging curve, obtain the third depth value of the target half-amplitude point based on the first depth value and the second depth value, determine the corresponding target weight value using the third depth value of the target half-amplitude point, and obtain the target layer data between the target half-amplitude points according to the target weight value. The lithofacies identification module is used to establish model training data based on the target layer data between the obtained target half-width points, train the initial lithofacies identification model using the model training data to obtain model prediction results, compare the predicted probability distribution features in the obtained model prediction results with the target probability distribution features, compare the predicted lithofacies structural heterogeneity features in the obtained model prediction results with the target lithofacies structural heterogeneity features, and determine the target lithofacies identification model based on the comparison results so as to complete lithofacies identification using the target lithofacies identification model; The feature analysis module includes: The target lithofacies structure heterogeneous feature determination unit is used to statistically analyze the lithofacies density and lithofacies frequency in the lithofacies assemblage features corresponding to the core segment, and analyze the target lithofacies structure heterogeneous features based on the lithofacies density and lithofacies frequency; Specifically, the lithofacies identification device is used to take the target layer data between the target half-width points as the minimum sample layer unit, and sequentially splice and combine the target number of minimum sample layer units along the vertical direction to construct depth slice unit data, and determine the average depth value of the depth slice unit data, thereby obtaining a depth slice label library containing vertical information as model training data. Specifically, the lithofacies identification device is used to determine the training set and validation set based on the depth slice label library, input the target probability distribution features and the training set into the initial lithofacies identification model for iterative training, continuously adjust the number of targets in the smallest sample layer unit to obtain new model training data, and use the new model training data to repeatedly train the model until the probability distribution features of the model prediction results and the heterogeneous features of the lithofacies structure match the target lithofacies probability distribution features of the core segment, thereby determining the target lithofacies identification model.
10. An electronic device, comprising: include: Memory, used to store computer programs; A processor for executing the computer program to implement the lithofacies identification method as described in any one of claims 1 to 8.
11. A computer readable storage medium, characterized in that, Used to store a computer program; wherein, when the computer program is executed by a processor, it implements the lithofacies identification method as described in any one of claims 1 to 8.