Reservoir perforation interval determination method and device

Abstract

The invention provides a reservoir perforation interval determination method and device. Obtaining logging response data and core actual measurement data of a target well in the target area; determining a plurality of reservoir physical property parameters of the target well along the depth direction according to the logging response data and the core actual measurement data; determining a multi-dimensional feature set according to the reservoir physical property parameters and the logging response data; determining the effective reservoir probability of each depth point of the target well according to the multi-dimensional feature set by using a preset reservoir effectiveness prediction model; wherein the preset reservoir effectiveness prediction model comprises a feature fitting module, a constraint rule execution module and a feature interaction calculation module; according to the effective reservoir probability, an effective reservoir interval meeting the continuity requirement is determined, and the effective reservoir interval serves as a target perforation interval of the target well; wherein the target perforation interval is used for guiding perforation position determination of the target well. Therefore, a more scientific and reasonable basis is provided for perforation position determination.

Description

Technical Field

[0001] This manual belongs to the field of oil and gas exploration and development technology, and in particular relates to a method and apparatus for determining reservoir perforation intervals. Background Technology

[0002] Currently, the industry generally uses a model that calculates physical properties based on a single physical formula and selects layers based on human experience to predict perforation locations. This model has low prediction accuracy in reservoirs with high clay content, well-developed fractures, or strong heterogeneity, and is prone to inaccurate identification of perforation layers.

[0003] There is currently no effective solution to the above problems. Summary of the Invention

[0004] This specification provides a method and apparatus for determining reservoir perforation intervals, which solves the problems of poor adaptability and strong subjectivity in the effectiveness judgment of existing well logging prediction perforation methods in heterogeneous reservoirs.

[0005] This specification provides a method for determining reservoir perforation intervals, including:

[0006] Acquire logging response data and core measurement data of the target well in the target area;

[0007] Based on the logging response data and the core measurement data, multiple reservoir physical parameters along the depth direction of the target well are determined;

[0008] Based on the reservoir physical properties and the logging response data, a multidimensional feature set is determined; wherein, the multidimensional feature set is used to determine the effectiveness of the reservoir.

[0009] Using a pre-defined reservoir effectiveness prediction model, the effective reservoir probability at each depth point of the target well is determined based on the multi-dimensional feature set; wherein, the pre-defined reservoir effectiveness prediction model includes a feature fitting module, a constraint rule execution module, and a feature interaction calculation module;

[0010] Based on the effective reservoir probability, an effective reservoir interval that meets the continuity requirement is determined, and the effective reservoir interval is used as the target perforation section of the target well; wherein, the target perforation section is used to guide the determination of the perforation location of the target well.

[0011] In one embodiment, determining the effective reservoir interval that meets the continuity requirement based on the effective reservoir probability, and using the effective reservoir interval as the target perforated section of the target well, includes:

[0012] Based on the effective reservoir probability, identify continuous effective reservoir segments in the target well and calculate the length of each continuous effective segment.

[0013] The effective reservoir interval that meets the continuity requirement is determined by the depth interval where the effective reservoir probability is not less than a first preset threshold and the corresponding continuous effective segment length is not less than a second preset threshold, and the effective reservoir interval is used as the target perforation segment of the target well.

[0014] In one embodiment, determining the multidimensional feature set based on the reservoir physical parameters and the logging response data includes:

[0015] Based on the reservoir physical properties such as porosity, permeability, and oil saturation, as well as the resistivity and natural gamma in the logging response data, a response characteristic reflecting the reservoir response properties is constructed.

[0016] Based on the combined relationships between porosity and permeability, resistivity and natural gamma, and porosity and oil saturation, a composite characteristic reflecting the synergistic relationship between reservoir capacity and oil content is constructed.

[0017] Based on the permeability and oil saturation, a capacity correlation feature reflecting the reservoir's capacity potential is constructed.

[0018] The response features, composite features, and capacity-related features are combined to determine a multidimensional feature set.

[0019] In one embodiment, determining the effective reservoir probability at each depth point of the target well using a preset reservoir effectiveness prediction model based on the multidimensional feature set includes:

[0020] Using the feature fitting module, based on the relationship between the multidimensional feature set and the historically labeled reservoir effectiveness tags, the initial effective reservoir probability corresponding to each depth point is determined by fitting the multidimensional feature set.

[0021] Using the constraint rule execution module, the intermediate effective reservoir probability at each depth point of the target well is determined based on the initial effective reservoir probability;

[0022] Using the feature interaction calculation module, the effective reservoir probability at each depth point of the target well is determined based on the intermediate effective reservoir probability.

[0023] In one embodiment, determining the intermediate effective reservoir probability at each depth point of the target well using the constraint rule execution module based on the initial effective reservoir probability includes:

[0024] Based on the contribution value of each feature in the initial effective reservoir probability to the prediction result, the corresponding feature importance coefficient is determined;

[0025] If the initial effective reservoir probability does not meet the preset engineering constraint rules, the feature weights of key features in the multidimensional feature set are adjusted according to the feature importance coefficient; wherein, the preset engineering constraint rules include the minimum continuous segment length threshold, the maximum upper and lower layer difference amplitude threshold, and the fluctuation stability threshold of the effective reservoir distribution.

[0026] Based on the adjusted feature weights, the initial effective reservoir probability is corrected to obtain the intermediate effective reservoir probability at each depth point of the target well.

[0027] In one embodiment, determining the effective reservoir probability at each depth point of the target well using the feature interaction calculation module based on the intermediate effective reservoir probability includes:

[0028] Calculate the corresponding feature interaction weight coefficients based on the vector relationships between the features in the multidimensional feature set;

[0029] Based on the feature interaction weight coefficient, the intermediate effective reservoir probability is dynamically corrected to obtain the effective reservoir probability at each depth point of the target well.

[0030] In one embodiment, the method further includes:

[0031] Based on the target perforated section, a perforation parameter configuration scheme for the target well is determined; wherein, the perforation parameter configuration scheme includes perforation density, perforation diameter, perforation phase, and perforation depth.

[0032] This specification provides a reservoir perforation interval determination apparatus, comprising:

[0033] The data acquisition module is used to acquire logging response data and core measurement data of the target well in the target area;

[0034] The parameter determination module is used to determine multiple reservoir physical property parameters along the depth direction of the target well based on the logging response data and the core measured data.

[0035] The feature determination module is used to determine a multi-dimensional feature set based on the reservoir physical property parameters and the logging response data; wherein the multi-dimensional feature set is used for reservoir effectiveness determination.

[0036] The probability determination module is used to determine the effective reservoir probability at each depth point of the target well based on the multidimensional feature set using a preset reservoir effectiveness prediction model; wherein, the preset reservoir effectiveness prediction model includes a feature fitting module, a constraint rule execution module, and a feature interaction calculation module.

[0037] The layer segment determination module is used to determine the effective reservoir interval that meets the continuity requirement based on the effective reservoir probability, and to use the effective reservoir interval as the target perforation segment of the target well; wherein, the target perforation segment is used to guide the determination of the perforation location of the target well.

[0038] This specification also provides a computer-readable storage medium having computer instructions stored thereon, which, when executed, implement a method for determining reservoir perforation intervals.

[0039] Based on the reservoir perforation interval determination method provided in this specification, the following steps are taken: Well logging response data and core measurement data of a target well in a target area are acquired; multiple reservoir physical property parameters along the depth direction of the target well are determined based on the well logging response data and the core measurement data; a multidimensional feature set is determined based on the reservoir physical property parameters and the well logging response data; wherein the multidimensional feature set is used for reservoir effectiveness determination; using a preset reservoir effectiveness prediction model, the effective reservoir probability at each depth point of the target well is determined based on the multidimensional feature set; wherein the preset reservoir effectiveness prediction model includes a feature fitting module, a constraint rule execution module, and a feature interaction calculation module; based on the effective reservoir probability, an effective reservoir interval that meets the continuity requirement is determined, and the effective reservoir interval is used as the target perforation interval of the target well; wherein the target perforation interval is used to guide the determination of the perforation location of the target well. In this way, by integrating well logging response data and core measurement data, reservoir physical parameters along the depth direction are comprehensively determined, thereby constructing a multi-dimensional feature set reflecting reservoir response characteristics, storage capacity, and production potential, which can more comprehensively characterize the reservoir state. Based on this feature set, a reservoir effectiveness prediction model consisting of three modules—feature fitting, constraint execution, and feature interaction—is used to achieve precise prediction of the probability of effective reservoirs at different depths, improving the accuracy and adaptability of reservoir evaluation. Finally, stable effective reservoir intervals are identified through continuity discrimination rules, ensuring that the target perforated interval has a good physical property foundation and production potential, thus providing a more scientific and reasonable basis for perforation location determination. Attached Figure Description

[0040] To more clearly illustrate the embodiments of this specification, the accompanying drawings used in the embodiments will be briefly introduced below. The drawings described below are only some embodiments recorded in this specification. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0041] Figure 1 This is a schematic flowchart of a method for determining reservoir perforation intervals provided in one embodiment of this specification;

[0042] Figure 2This is a schematic diagram of the electronic device structure provided in one embodiment of this specification;

[0043] Figure 3 This is a schematic diagram of the structural composition of a reservoir perforation section determination device provided in one embodiment of this specification;

[0044] Figure 4 This is a schematic diagram of the predicted perforation location of a well in a block, provided as an embodiment of this specification. Detailed Implementation

[0045] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this specification, and not all embodiments. Based on the embodiments in this specification, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this specification.

[0046] Currently, the mainstream logging prediction methods in the industry mostly adopt a model of calculating physical properties using a single physical formula and manually screening perforated layers based on experience. In the reservoir property prediction stage, porosity calculation relies on the Willy formula, permeability relies on porosity-permeability index regression, and oil saturation relies on the Archie formula. Although these methods have clear physical meanings, they are only applicable to ideal reservoirs with homogeneous lithology. They have poor adaptability to reservoirs containing clay, fractures, or with strong heterogeneity. For example, in sandstone reservoirs with clay content exceeding 15%, the Willy formula can have a porosity prediction error of more than 5%, and the Archie formula is prone to misclassifying oil layers as water layers in low-resistivity oil layers. In the effective reservoir screening stage, traditional methods mostly judge reservoir effectiveness by manually drawing logging curve cross plots (such as GR-AC cross plots and RT-φ cross plots). This is greatly affected by the differences in the experience of operators, and there is a problem that the interpretation results of the same logging curve by different people can vary by more than 10%. Moreover, it is difficult to simultaneously take into account multiple dimensions of indicators such as reservoir quality and oil content.

[0047] As oil and gas resource development gradually extends to low-permeability and unconventional reservoirs, the limitations of traditional logging prediction perforation methods are becoming increasingly apparent: the increased proportion of heterogeneous formations further amplifies the error in predicting physical properties using a single physical formula, the efficiency and accuracy of manual experience-based interpretation are insufficient to meet the needs of large-scale development, and the imbalance between "production capacity and risk" in perforation target selection is more likely to trigger development conflicts.

[0048] To address the root causes of the aforementioned problems, this specification integrates well logging response data and core measurement data to comprehensively determine reservoir physical parameters along the depth direction. This allows for the construction of a multi-dimensional feature set reflecting reservoir response characteristics, storage capacity, and production potential, providing a more comprehensive characterization of reservoir conditions. Based on this feature set, a reservoir effectiveness prediction model, comprised of feature fitting, constraint execution, and feature interaction modules, enables precise prediction of the probability of effective reservoirs at different depths, improving the accuracy and adaptability of reservoir evaluation. Finally, a continuity discrimination rule identifies stable effective reservoir intervals, ensuring that the target perforated interval possesses a sound physical property foundation and production potential, thus providing a more scientific and reasonable basis for perforation location determination.

[0049] See Figure 1 As shown in the embodiments of this specification, a method for determining reservoir perforation intervals is provided, wherein the method is specifically applied to the server side. In specific implementation, the method may include the following:

[0050] S101: Obtain the logging response data and core measurement data of the target well in the target area;

[0051] S102: Based on the logging response data and the core measured data, determine multiple reservoir physical property parameters of the target well along the depth direction;

[0052] S103: Determine a multidimensional feature set based on the reservoir physical property parameters and the well logging response data; wherein, the multidimensional feature set is used for reservoir effectiveness determination;

[0053] S104: Using a preset reservoir effectiveness prediction model, determine the effective reservoir probability at each depth point of the target well based on the multidimensional feature set; wherein, the preset reservoir effectiveness prediction model includes a feature fitting module, a constraint rule execution module, and a feature interaction calculation module.

[0054] S105: Based on the effective reservoir probability, determine the effective reservoir interval that meets the continuity requirement, and use the effective reservoir interval as the target perforation section of the target well; wherein, the target perforation section is used to guide the determination of the perforation location of the target well.

[0055] The aforementioned logging response data can be logging curve data such as natural gamma (GR), sonic transit time (AC), density (DEN), resistivity (RT), and neutron porosity (CNL) obtained during the drilling or completion of the target well. The logging curves are continuously collected based on depth and are used to reflect the physical response characteristics of the formation.

[0056] The aforementioned core measurement data can be experimental measurement results at the depth points corresponding to the core samples obtained during drilling, including parameters that characterize the actual reservoir performance of the formation, such as porosity, permeability, and oil saturation, which are used for quantitative calibration of logging data and training of inversion models.

[0057] The aforementioned reservoir physical properties parameters can be porosity, permeability, and oil saturation obtained by combining well logging response data and core measurement data, and are used to quantitatively characterize the formation reservoir performance at different depths of the target well.

[0058] The aforementioned multidimensional feature set can be a combination of features, including response features, composite features, and production capacity correlation features, constructed based on the reservoir physical parameters and well logging response data. It is used to input into the reservoir effectiveness prediction model to achieve a comprehensive judgment on reservoir effectiveness.

[0059] In some embodiments, determining multiple reservoir physical parameters along the depth direction of the target well based on the logging response data and the core measurement data may specifically include:

[0060] S1: Based on the logging curve parameters such as natural gamma, resistivity, density and neutron porosity in the logging response data, establish a characteristic correspondence with the core measured porosity, permeability and oil saturation data at the corresponding depth points;

[0061] S2: Based on the aforementioned feature correspondence, the mapping coefficient between the well logging response data and reservoir physical parameters is determined using a fitting calculation model;

[0062] S3: Apply the mapping coefficient to the logging response data of the entire well section of the target well, calculate the porosity, permeability and oil saturation parameters along the depth direction, and determine multiple reservoir physical property parameters of the target well.

[0063] In some embodiments, determining the multidimensional feature set based on the reservoir physical parameters and the logging response data may specifically include:

[0064] Based on the consistency of the changing trends and the amplitude coupling relationship among different logging curves, dynamic morphological features reflecting lithological transformation and pore structure changes are extracted; the morphological features include the synergistic gradient between natural gamma and density curves, the range of the difference between sonic transit time and neutron porosity, and the statistical skewness and kurtosis within their sliding windows.

[0065] Furthermore, for the permeability and oil saturation data in the reservoir physical properties, an interval clustering model is constructed to identify the characteristic distribution patterns of different reservoir microfacies or productive zones, and the clustering results are added to the multidimensional feature set as high-order category feature codes.

[0066] In addition, by using feature dimensionality reduction and screening methods such as principal component analysis (PCA) or maximum information coefficient (MIC), key feature variables with high main information content and low redundancy are extracted from the constructed high-dimensional combined features to form a multi-dimensional feature set for reservoir effectiveness determination. This feature set has stronger expressive power and discriminative power, which can improve the accuracy of subsequent models in identifying complex reservoir features.

[0067] Based on the above embodiments, by fusing well logging response data and core measurement data, reservoir physical parameters along the depth direction are comprehensively determined, thereby constructing a multi-dimensional feature set reflecting reservoir response characteristics, storage capacity, and production potential, which can more comprehensively characterize the reservoir state. Based on the feature set, a reservoir effectiveness prediction model composed of three modules—feature fitting, constraint execution, and feature interaction—is used to achieve precise prediction of the probability of effective reservoirs at different depths, improving the accuracy and adaptability of reservoir evaluation. Finally, stable effective reservoir intervals are identified through continuity discrimination rules, ensuring that the target perforated interval has a good physical property foundation and production potential, thus providing a more scientific and reasonable basis for perforation location determination.

[0068] In some embodiments, the method of determining an effective reservoir interval that meets the continuity requirement based on the effective reservoir probability and using the effective reservoir interval as the target perforation interval of the target well may further include the following:

[0069] S1: Based on the effective reservoir probability, identify continuous effective reservoir segments in the target well and calculate the length of each continuous effective segment;

[0070] S2: Determine the effective reservoir interval that meets the continuity requirement by the depth interval where the effective reservoir probability is not less than the first preset threshold and the corresponding continuous effective segment length is not less than the second preset threshold, and use the effective reservoir interval as the target perforation segment of the target well.

[0071] Specifically, the effective reservoir probability sequence is traversed through a sliding window according to depth, and it is determined whether there is a continuous depth segment within the window of each depth point that is greater than a preset effective probability threshold; wherein, the effective probability threshold can be selected empirically as a value between 0.6 and 0.8. If the length of the effective depth segment within the window exceeds a preset minimum effective layer thickness threshold (e.g., 2 meters), then the segment is initially marked as an effective reservoir segment.

[0072] After initial marking, multiple adjacent effective reservoir segments are merged: if the invalid interval between two effective reservoir segments is less than a preset interval threshold (e.g., 0.5 meters), they are merged into a continuous effective segment, thereby avoiding the false cutting of effective layers due to short-interval interference caused by formation heterogeneity.

[0073] Finally, all valid reservoir sections that meet the above continuity discrimination rules are designated as target perforation sections for the target well, and the top depth, bottom depth, and thickness information of each perforation section are output for subsequent perforation parameter configuration and perforation operation plan formulation.

[0074] In some embodiments, the method for determining a multidimensional feature set based on the reservoir physical parameters and the logging response data may further include the following:

[0075] S1: Based on the reservoir physical properties parameters such as porosity, permeability, and oil saturation, as well as the resistivity and natural gamma in the logging response data, construct response characteristics that reflect the reservoir response properties.

[0076] S2: Based on the combined relationship between porosity and permeability, resistivity and natural gamma, and porosity and oil saturation, a composite feature reflecting the synergistic relationship between reservoir capacity and oil content is constructed.

[0077] S3: Based on the permeability and the oil saturation, construct a capacity correlation feature that reflects the reservoir's capacity potential;

[0078] S4: Combine the response features, composite features, and capacity-related features to determine a multidimensional feature set.

[0079] Specifically, to construct a multi-dimensional feature set for reservoir effectiveness assessment, reservoir physical parameters and logging response data along the depth direction of the target well are first acquired. The reservoir physical parameters include porosity, permeability, and oil saturation, while the logging response data includes conventional logging curves such as natural gamma ray, resistivity, density, and neutron porosity. Based on this data, an information set containing three types of features is constructed: first, response features, which characterize the basic properties of the reservoir's response to logging by extracting raw logging or physical dimension data such as porosity, natural gamma ray, and resistivity; second, composite features, which reflect the synergistic coupling characteristics between reservoir capacity and oil content by constructing combined relationships such as porosity and permeability, porosity and oil saturation, and resistivity and natural gamma ray; and third, production capacity correlation features, which reveal the reservoir's potential production capacity under current fluid conditions by calculating the product of permeability and oil saturation. The three types of features mentioned above are concatenated into feature vectors at depth points and combined according to well interval sequences to form a multi-dimensional feature set covering the entire well interval. This provides a unified input basis for subsequent reservoir effectiveness probability calculations and perforation interval identification. Through this process, not only is a systematic mapping from physical property data and logging curves to the feature space achieved, but geological attributes and production potential are also effectively integrated, which helps to improve the accuracy of reservoir evaluation and the precision of perforation design.

[0080] Through the aforementioned feature construction process, a multi-dimensional fusion of well logging response information and reservoir physical parameters is achieved. This enables the model to simultaneously consider the reservoir's physical characteristics, response patterns, and production potential when determining reservoir effectiveness, thereby significantly improving the accuracy of identifying heterogeneous, low-permeability, or argillaceous reservoirs. Compared to traditional methods that rely on a single physical formula or manual interpretation, this scheme can automatically extract key feature correlations, reduce the influence of human subjectivity, and improve the stability of reservoir classification and the precision of perforation interval determination.

[0081] In some embodiments, the method of determining the effective reservoir probability at each depth point of the target well using a preset reservoir effectiveness prediction model based on the multidimensional feature set may further include the following:

[0082] S1: Using the feature fitting module, based on the relationship between the multidimensional feature set and the historically labeled reservoir effectiveness labels, the initial effective reservoir probability corresponding to each depth point is determined by fitting the multidimensional feature set.

[0083] S2: Using the constraint rule execution module, determine the intermediate effective reservoir probability at each depth point of the target well based on the initial effective reservoir probability;

[0084] S3: Using the feature interaction calculation module, determine the effective reservoir probability at each depth point of the target well based on the intermediate effective reservoir probability.

[0085] Specifically, firstly, a supervised learning model is constructed, taking well logging feature parameters as input and historical reservoir effectiveness labels as output. The well logging feature parameters include a multi-dimensional feature set consisting of porosity, permeability, oil saturation, resistivity, spontaneous potential, natural gamma, and sonic transit time. The reservoir effectiveness labels characterize whether a corresponding depth point is a valid reservoir. During model construction, the feature fitting module selects a historical sample dataset containing well logging features and corresponding labels, and performs standardization, missing value processing, and feature filtering operations on this dataset as the input basis for fitting calculations. Further, the feature fitting module uses any one or a combination of algorithms such as Gradient Boosting Tree, Support Vector Machine, Random Forest, or Multilayer Perceptron Neural Network (MLP) to train a mapping function model, which characterizes the nonlinear correspondence between the multi-dimensional feature set and the reservoir effectiveness labels. After model training is completed, the multi-dimensional features corresponding to each depth point of the target well are input using the trained model, and the initial effective reservoir probability value of that depth point is output to characterize the degree of its possibility of becoming an effective reservoir. In a preferred embodiment, to improve model accuracy, the feature fitting module also iteratively optimizes the algorithm parameters based on cross-validation and adjusts the feature combination in combination with the feature importance ranking results to achieve high-precision prediction of reservoir effectiveness probability.

[0086] Next, using the constraint rule execution module, engineering constraint rules for determining reservoir effectiveness are constructed based on the geological characteristic parameters of the target well and the development parameters of the regional well network, and the initial effective reservoir probability is corrected according to the rules. These rules may include, but are not limited to: an upper limit for effectiveness corresponding to low-porosity, low-permeability sections; a reservoir penalty mechanism corresponding to abnormally high natural gamma sections; and a comparison of connectivity constraints between adjacent wells. During implementation, it is first assessed whether the current initial effective reservoir probability significantly deviates from these rules (e.g., consecutive high-probability sections correspond to significantly low-permeability areas). If it deviates, the weights of key features (such as permeability) that significantly affect the prediction results are adjusted according to a pre-determined feature importance coefficient, thereby recalculating the corrected reservoir probability to obtain the intermediate effective reservoir probability.

[0087] Finally, utilizing the aforementioned feature interaction calculation module, the correlation and contextual coupling information between features are further combined to enhance the model's understanding of nonlinear feature synergistic relationships. Specifically, the feature embedding method can be used to represent each feature as a vector and calculate the cosine similarity or distance matrix to uncover coupling patterns between resistivity and oil saturation, natural gamma and porosity, etc. Based on the interaction weight matrix, the intermediate effective reservoir probability is locally corrected to improve the discrimination accuracy for ambiguous boundary intervals or weakly responsive reservoirs, and finally, the effective reservoir probability at each depth point of the target well is output.

[0088] Through the above steps, this embodiment can achieve integrated modeling of three types of information: reservoir properties, well logging response, and engineering experience. This not only improves the accuracy of predictions but also provides stronger interpretability and adaptability.

[0089] In some embodiments, the method utilizes the constraint rule execution module to determine the intermediate effective reservoir probability at each depth point of the target well based on the initial effective reservoir probability. In specific implementations, the method may further include the following:

[0090] S1: Determine the corresponding feature importance coefficient based on the contribution value of each feature in the initial effective reservoir probability to the prediction result;

[0091] S2: If the initial effective reservoir probability does not meet the preset engineering constraint rules, adjust the feature weights of key features in the multidimensional feature set according to the feature importance coefficient; wherein, the preset engineering constraint rules include the minimum continuous segment length threshold of the effective reservoir probability, the maximum upper and lower layer difference amplitude threshold, and the fluctuation stability threshold of the effective reservoir distribution.

[0092] S3: Based on the adjusted feature weights, correct the initial effective reservoir probability to obtain the intermediate effective reservoir probability at each depth point of the target well.

[0093] In some embodiments, determining the corresponding feature importance coefficient based on the contribution value of each feature in the initial effective reservoir probability to the prediction result may specifically include:

[0094] Based on the initial prediction results of the target well using the pre-defined reservoir effectiveness prediction model, sensitivity indices of each input feature at the corresponding depth point to changes in the output probability value are extracted. Then, using gradient-based rate-of-change analysis methods (e.g., feature perturbation method or integrated gradient method), the influence of each feature on the effective reservoir probability is evaluated. All features are normalized and ranked according to their influence level to determine the feature importance coefficient corresponding to each input feature. This feature importance coefficient measures the degree to which each feature dominates the current effective reservoir probability prediction.

[0095] In some embodiments, when the initial effective reservoir probability does not meet the preset engineering constraint rules, adjusting the feature weights of key features in the multidimensional feature set according to the feature importance coefficient may specifically include:

[0096] The initial effective reservoir probability sequence predicted along the depth direction of the target well is matched and judged with preset engineering constraint rules, which include: the minimum continuous segment length threshold of the effective reservoir interval (e.g., not less than 3m), the maximum change amplitude threshold of the effective reservoir probability between adjacent layers (e.g., not more than 0.4), and the fluctuation stability threshold of the reservoir effectiveness distribution along the longitudinal direction (e.g., the standard deviation of 3 consecutive points is not greater than 0.15), etc.

[0097] If the target well prediction results exhibit abnormalities such as excessively short effective segment length, drastic probability changes, or non-stationary sequences, they are determined to be non-compliant with engineering constraints. Subsequently, based on the aforementioned feature importance coefficients, key features dominating prediction bias are identified, and their contribution weights are adjusted using a weighted adjustment method. For example, the weights of features influencing porosity and suppressing spurious noise features (such as interfering high-frequency gamma responses) are increased, resulting in an updated feature weight set.

[0098] In some embodiments, the step of correcting the initial effective reservoir probability based on the adjusted feature weights to obtain the intermediate effective reservoir probability at each depth point of the target well may specifically include:

[0099] The adjusted feature weights are reapplied to the multidimensional feature set of the target well, and the effective reservoir probability calculation process is performed again to obtain the corrected probability distribution sequence, which serves as the intermediate effective reservoir probability of the target well for subsequent feature interaction calculations and final reservoir effectiveness confirmation.

[0100] In some embodiments, the method of determining the effective reservoir probability at each depth point of the target well using the feature interaction calculation module based on the intermediate effective reservoir probability may further include the following:

[0101] S1: Calculate the corresponding feature interaction weight coefficients based on the vector relationships between the features in the multidimensional feature set;

[0102] S2: Based on the feature interaction weight coefficient, the intermediate effective reservoir probability is dynamically corrected to obtain the effective reservoir probability at each depth point of the target well.

[0103] Specifically, the multidimensional feature set includes, but is not limited to, the following three categories: The first category is response features, which reflect the lithology, pore structure, and electrical response of the formation, including logging curve features such as natural gamma, sonic transit time, lateral resistivity, and spontaneous potential; the second category is composite features, which characterize the local morphological features and trends of logging signals, including curve range, sliding window statistics, and local curvature; the third category is production capacity-related features, which reflect the physical properties and development potential of the reservoir, including parameters such as porosity, permeability, and oil saturation.

[0104] In the feature interaction calculation module, for any two different feature combinations in the aforementioned multidimensional feature set, they are converted into vector form, and the correlation coefficients between different feature pairs are obtained based on a preset vector similarity calculation strategy. Specifically, the vector dot product method can be used to measure the cooperative directionality of different features at the same depth point; or the cosine similarity method can be used to analyze the directional consistency of features in high-dimensional space to determine the coupling strength and interrelationships between them. This type of similarity calculation can reveal the enhancing or inhibiting effect of each feature combination on reservoir effectiveness prediction from a spatial distribution perspective, thereby uncovering the implicit dependencies between features.

[0105] To ensure consistency and comparability of the interaction effects under different feature combinations, the correlation coefficients mentioned above are uniformly normalized. Specifically, the original similarity scores between features are standardized using the Softmax normalization function to form a feature interaction weight matrix. Each element in this matrix represents the weight assigned to the importance of one feature to another, dynamically describing the degree of mutual reinforcement or inhibition formed by each feature in reservoir effectiveness assessment.

[0106] In actual geological reservoirs, various characteristics are often not independent but rather exhibit complex coupling relationships. For example, in fractured tight reservoirs, a simultaneous increase in high sonic transit time and lateral resistivity may simultaneously indicate tight zones or areas with naturally developed fractures. Similarly, in areas with interbedded sand and mud, although porosity and permeability belong to different physical property dimensions, their numerical trends often show a positive correlation, constituting an important characteristic for synergistic productivity development. These interactive relationships are difficult to fully represent through univariate analysis and require identification and quantification through interactive modeling strategies.

[0107] Therefore, in this embodiment, for the multidimensional feature vector of each depth point, the numerical distribution, direction of change, and degree of similarity among the feature components are comprehensively analyzed to identify reinforcing or weakening relationships between features. For example, if the natural gamma and sonic transit time at a certain depth point fluctuate drastically at the same time, and the predicted values ​​of porosity and permeability are significantly higher, it can be determined that the area possesses a composite feature combination of "high-quality reservoirs superimposed on geological interfaces," thereby increasing its reservoir effectiveness weight through the interaction matrix. Conversely, when some features perform well individually, but the combination relationship differs significantly from the distribution of effective reservoir features in historically labeled samples, the effectiveness prediction weight of the area will be automatically reduced to decrease the risk of misjudgment.

[0108] By introducing a feature interaction weight calculation and dynamic adjustment mechanism, the system can integrate combined interaction information while fully preserving the expressive power of various individual features, thereby enhancing its adaptability to complex reservoir types. This interaction weight matrix, as a feature adjustment factor, participates in the subsequent calculation of reservoir effectiveness probability, achieving more accurate reservoir boundary characterization and reservoir quality identification in both spatial and feature dimensions. This contributes to decision support in key areas such as perforation section selection, reservoir stimulation scheme formulation, and production forecasting.

[0109] In some embodiments, the dynamic correction of the intermediate effective reservoir probability based on the feature interaction weight coefficient to obtain the effective reservoir probability at each depth point of the target well may specifically include:

[0110] The various feature pairs in the multidimensional feature set are represented by vectors;

[0111] Based on a preset vector similarity calculation strategy, the correlation coefficient between different feature pairs is determined. The vector similarity calculation strategy includes dot product calculation or cosine similarity calculation.

[0112] The correlation coefficients are standardized using a preset normalization function to generate a feature interaction weight matrix, which reflects the relative influence of each feature on other features in reservoir effectiveness determination.

[0113] Based on the feature interaction weight matrix, the intermediate effective reservoir probability is dynamically corrected to obtain the effective reservoir probability at each depth point of the target well, thereby enhancing the adaptability of multi-feature combination to reservoir effectiveness identification.

[0114] By introducing a feature interaction weight matrix to dynamically correct the probability of intermediate effective reservoirs, the coupling relationship between response features, composite features, and production capacity-related features can be comprehensively considered. This fully explores the synergistic indicative role of multiple feature combinations on reservoir effectiveness under different geological conditions, improves the model's discrimination accuracy and generalization ability in complex reservoirs, and thus more accurately depicts the effective reservoir distribution boundary, providing a reliable basis for subsequent perforation optimization and reserve assessment.

[0115] In some embodiments, the method may further include the following:

[0116] Based on the target perforated section, a perforation parameter configuration scheme for the target well is determined; wherein, the perforation parameter configuration scheme includes perforation density, perforation diameter, perforation phase, and perforation depth.

[0117] Specifically, firstly, the depth range of the previously determined target perforated section and its corresponding reservoir attribute data are obtained, including characterizing parameters such as porosity, permeability, oil saturation, and effective reservoir probability. Based on these reservoir attributes, a set of input features for perforation parameter selection is constructed, including permeability parameters reflecting reservoir fluid flow capacity, porosity parameters reflecting reservoir storage capacity, and oil saturation parameters reflecting production potential.

[0118] Secondly, based on the differences in reservoir properties and productivity characteristics within the perforated sections, the corresponding perforation density, perforation diameter, and perforation phase distribution are determined. Specifically, for high-permeability, high-oil-saturation, high-perforation-density and large-perforation-diameter reservoir sections, a higher perforation density and larger perforation diameter are selected to enhance the connectivity of oil and gas flow channels; for reservoir sections with low permeability or high clay content, a medium perforation density is selected and the perforation phase angle is optimized to avoid seepage channel closure caused by stress interference. The perforation phase can be adjusted within a variable angle range of 45° to 90° to match the direction of the local stress field and achieve wellbore-fracture system coupling optimization.

[0119] Next, based on the burial depth, well inclination angle, and casing size of the target formation, the corresponding perforation depth parameters are determined. The perforation depth is used to control the penetration depth of the perforating projectile after penetrating the casing and cement sheath, ensuring that the penetration depth reaches the effective reservoir interior. Specifically, the optimal perforation depth range can be calculated based on the average permeability of the perforated section and the formation pressure gradient, for example, between 10 and 25 cm, to achieve optimal fluid connectivity while maintaining casing integrity.

[0120] Finally, the perforation density, perforation diameter, perforation phase, and perforation depth are combined and configured according to the characteristics of the formation to form a perforation parameter configuration scheme for the target well. This scheme can be used as well completion design input, providing quantitative parameter basis for field perforation operations.

[0121] Through the above embodiments, automated configuration of perforation parameters driven by reservoir characteristic data is realized, avoiding the subjective problem of relying on experience to select parameters. The perforation strategy can be adaptively adjusted for different types of reservoirs, significantly improving the completion quality and single-well productivity of oil and gas wells.

[0122] As can be seen from the above, the reservoir perforation interval determination method provided in this specification involves acquiring well logging response data and core measurement data of a target well in a target area; determining multiple reservoir physical property parameters along the depth direction of the target well based on the well logging response data and the core measurement data; determining a multidimensional feature set based on the reservoir physical property parameters and the well logging response data; wherein the multidimensional feature set is used for reservoir effectiveness determination; using a preset reservoir effectiveness prediction model, determining the effective reservoir probability at each depth point of the target well based on the multidimensional feature set; wherein the preset reservoir effectiveness prediction model includes a feature fitting module, a constraint rule execution module, and a feature interaction calculation module; determining an effective reservoir interval that meets the continuity requirement based on the effective reservoir probability, and using the effective reservoir interval as the target perforation interval of the target well; wherein the target perforation interval is used to guide the determination of the perforation location of the target well. In this way, by integrating well logging response data and core measurement data, reservoir physical parameters along the depth direction are comprehensively determined, thereby constructing a multi-dimensional feature set reflecting reservoir response characteristics, storage capacity, and production potential, which can more comprehensively characterize the reservoir state. Based on this feature set, a reservoir effectiveness prediction model consisting of three modules—feature fitting, constraint execution, and feature interaction—is used to achieve precise prediction of the probability of effective reservoirs at different depths, improving the accuracy and adaptability of reservoir evaluation. Finally, stable effective reservoir intervals are identified through continuity discrimination rules, ensuring that the target perforated interval has a good physical property foundation and production potential, thus providing a more scientific and reasonable basis for perforation location determination.

[0123] See Figure 2 As shown in the embodiments of this specification, a specific electronic device is also provided, wherein the electronic device includes a network communication port 201, a processor 202 and a memory 203, and the above structures are connected by internal cables so that the various structures can perform specific data interaction.

[0124] Specifically, the network communication port 201 can be used to acquire well logging response data and core measurement data of the target well in the target area.

[0125] The processor 202 can specifically be used to determine multiple reservoir physical property parameters along the depth direction of the target well based on the logging response data and the core measurement data; determine a multi-dimensional feature set based on the reservoir physical property parameters and the logging response data; wherein the multi-dimensional feature set is used for reservoir effectiveness determination; determine the effective reservoir probability at each depth point of the target well based on the multi-dimensional feature set using a preset reservoir effectiveness prediction model; wherein the preset reservoir effectiveness prediction model includes a feature fitting module, a constraint rule execution module, and a feature interaction calculation module; determine an effective reservoir interval that meets the continuity requirement based on the effective reservoir probability, and use the effective reservoir interval as the target perforation segment of the target well; wherein the target perforation segment is used to guide the determination of the perforation location of the target well.

[0126] The memory 203 can be used to store the corresponding instruction program.

[0127] Based on the above method, the relevant structural performance of electronic equipment can be effectively utilized to improve the data processing speed of electronic equipment and efficiently realize a method for determining reservoir perforation sections.

[0128] In this embodiment, the network communication port 201 can be a virtual port bound to different communication protocols, thereby enabling the sending or receiving of different data. For example, the network communication port can be a port responsible for web data communication, a port responsible for FTP data communication, or a port responsible for email data communication. Furthermore, the network communication port can also be a physical communication interface or communication chip. For example, it can be a wireless mobile network communication chip, such as GSM or CDMA; it can also be a Wi-Fi chip; or it can be a Bluetooth chip.

[0129] In this embodiment, the processor 202 can be implemented in any suitable manner. For example, the processor can take the form of a microprocessor or processor and a computer-readable medium storing computer-readable program code (e.g., software or firmware) executable by the (micro)processor, logic gates, switches, application-specific integrated circuits (ASICs), programmable logic controllers, and embedded microcontrollers, etc. This specification is not limiting.

[0130] In this embodiment, the memory 203 may include a hierarchy. In a digital system, anything that can store binary data can be a memory. In an integrated circuit, a circuit with storage function but no physical form is also called a memory, such as RAM, FIFO, etc. In a system, a storage device with a physical form is also called a memory, such as a memory stick, TF card, etc.

[0131] This specification also provides a computer-readable storage medium based on the above-described method for determining reservoir perforation intervals. The method acquires logging response data and core measurement data of a target well in a target area; determines multiple reservoir physical property parameters along the depth direction of the target well based on the logging response data and the core measurement data; determines a multidimensional feature set based on the reservoir physical property parameters and the logging response data; wherein the multidimensional feature set is used for reservoir effectiveness determination; using a preset reservoir effectiveness prediction model, the effective reservoir probability at each depth point of the target well is determined based on the multidimensional feature set; wherein the preset reservoir effectiveness prediction model includes a feature fitting module, a constraint rule execution module, and a feature interaction calculation module; based on the effective reservoir probability, an effective reservoir interval meeting continuity requirements is determined, and the effective reservoir interval is used as the target perforation interval of the target well; wherein the target perforation interval is used to guide the determination of the perforation location of the target well.

[0132] In this embodiment, the storage medium includes, but is not limited to, Random Access Memory (RAM), Read-Only Memory (ROM), Cache, Hard Disk Drive (HDD), or Memory Card. The memory can be used to store computer program instructions. The network communication unit can be an interface configured according to standards specified in the communication protocol for network connection communication.

[0133] In this embodiment, the specific functions and effects implemented by the program instructions stored in the computer-readable storage medium can be explained in comparison with other embodiments, and will not be repeated here.

[0134] See Figure 3 At the software level, embodiments of this specification also provide a reservoir perforation section determination device, which may specifically include the following structural modules:

[0135] The data acquisition module 301 is used to acquire the logging response data and core measurement data of the target well in the target area;

[0136] The parameter determination module 302 is used to determine multiple reservoir physical property parameters along the depth direction of the target well based on the logging response data and the core measured data.

[0137] The feature determination module 303 is used to determine a multi-dimensional feature set based on the reservoir physical property parameters and the logging response data; wherein the multi-dimensional feature set is used for reservoir effectiveness determination.

[0138] The probability determination module 304 is used to determine the effective reservoir probability at each depth point of the target well based on the multidimensional feature set using a preset reservoir effectiveness prediction model; wherein, the preset reservoir effectiveness prediction model includes a feature fitting module, a constraint rule execution module, and a feature interaction calculation module.

[0139] The segment determination module 305 is used to determine an effective reservoir interval that meets the continuity requirement based on the effective reservoir probability, and to use the effective reservoir interval as the target perforation segment of the target well; wherein, the target perforation segment is used to guide the determination of the perforation location of the target well.

[0140] In some embodiments, the above-mentioned segment determination module 305, in specific implementation, identifies continuous effective reservoir segments in the target well based on the effective reservoir probability, and calculates the continuous effective segment length of each effective reservoir segment; determines the effective reservoir interval that meets the continuity requirement by the depth interval where the effective reservoir probability is not less than a first preset threshold and the corresponding continuous effective segment length is not less than a second preset threshold, and uses the effective reservoir interval as the target perforated segment of the target well.

[0141] In some embodiments, the feature determination module 303, in specific implementation, constructs response features reflecting reservoir response characteristics based on the porosity, permeability, and oil saturation in reservoir physical properties, and the resistivity and natural gamma in well logging response data; constructs composite features reflecting the synergistic relationship between reservoir capacity and oil content based on the combined relationships between porosity and permeability, resistivity and natural gamma, and porosity and oil saturation; constructs production capacity correlation features reflecting reservoir production potential based on the permeability and oil saturation; and combines the response features, composite features, and production capacity correlation features to determine a multidimensional feature set.

[0142] In some embodiments, the probability determination module 304, in its specific implementation, utilizes the feature fitting module to determine the initial effective reservoir probability corresponding to each depth point by fitting the multidimensional feature set according to the relationship between the multidimensional feature set and the historically labeled reservoir effectiveness tags; the constraint rule execution module is used to determine the intermediate effective reservoir probability of each depth point of the target well based on the initial effective reservoir probability; and the feature interaction module is used to determine the effective reservoir probability of each depth point of the target well based on the intermediate effective reservoir probability using the feature interaction calculation module.

[0143] In some embodiments, the constraint rule execution module, in its specific implementation, determines the corresponding feature importance coefficient based on the contribution value of each feature in the initial effective reservoir probability to the prediction result; if the initial effective reservoir probability does not meet the preset engineering constraint rules, it adjusts the feature weights of key features in the multidimensional feature set according to the feature importance coefficients; wherein, the preset engineering constraint rules include the minimum continuous segment length threshold, the maximum upper and lower layer difference amplitude threshold, and the fluctuation stability threshold of the effective reservoir distribution; based on the adjusted feature weights, the initial effective reservoir probability is corrected to obtain the intermediate effective reservoir probability at each depth point of the target well.

[0144] In some embodiments, the feature interaction module, when specifically implemented, calculates the corresponding feature interaction weight coefficient based on the vector relationship between each feature in the multidimensional feature set; and dynamically corrects the intermediate effective reservoir probability based on the feature interaction weight coefficient to obtain the effective reservoir probability at each depth point of the target well.

[0145] In some embodiments, the device further includes: determining a perforation parameter configuration scheme for the target well based on the target perforation zone; wherein the perforation parameter configuration scheme includes perforation density, perforation diameter, perforation phase, and perforation depth.

[0146] It should be noted that the units, devices, or modules described in the above embodiments can be implemented by computer chips or physical entities, or by products with certain functions. For ease of description, the above devices are described by dividing them into various modules according to their functions. Of course, in implementing this specification, the functions of each module can be implemented in the same software and / or hardware, or modules that implement the same function can be implemented by a combination of sub-modules or sub-units, etc. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and there may be other division methods in actual implementation. For example, units or components can be combined or integrated into another system, or some features can be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection between the devices or units shown or discussed can be through some interfaces, and the indirect coupling or communication connection between devices or units can be electrical, mechanical, or other forms.

[0147] As can be seen from the above, based on the reservoir perforation interval determination device provided in the embodiments of this specification, the following steps are taken: First, well logging response data and core measurement data of a target well in a target area are acquired. Then, multiple reservoir physical property parameters along the depth direction of the target well are determined based on the well logging response data and the core measurement data. Next, a multidimensional feature set is determined based on the reservoir physical property parameters and the well logging response data. This multidimensional feature set is used for reservoir effectiveness determination. Finally, using a preset reservoir effectiveness prediction model, the effective reservoir probability at each depth point of the target well is determined based on the multidimensional feature set. The preset reservoir effectiveness prediction model includes a feature fitting module, a constraint rule execution module, and a feature interaction calculation module. Based on the effective reservoir probability, an effective reservoir interval that meets the continuity requirement is determined, and this effective reservoir interval is used as the target perforation interval of the target well. The target perforation interval is used to guide the determination of the perforation location of the target well.

[0148] In a specific scenario example, the reservoir perforation interval determination method and apparatus provided in this specification can be applied, solving the problems of poor adaptability and strong subjectivity in effectiveness judgment of existing well logging prediction perforation methods in heterogeneous reservoirs. The specific implementation process may include the following:

[0149] S1: Perform data acquisition and preprocessing operations.

[0150] Specifically, it includes collecting three types of core data: First, acquiring logging response data of the target well, which includes curve parameters such as natural gamma (GR), sonic transit time (AC), shallow and deep lateral resistivity (RLLD / RLLS), microsphere resistivity (R05) and well diameter (CAL), forming a multi-parameter logging dataset indexed by depth.

[0151] Secondly, core measurement data of a comparative well in the same geological block as the target well and with similar lithology are obtained. The core measurement data includes porosity (φ), permeability (K) and oil saturation (So). The measured depth is corrected for depth error to ensure that the depth difference is less than 0.2m.

[0152] Third, collect production capacity verification data, including the perforation interval range of the perforated wells, the perforation parameter configuration, and oil testing or production data, which will serve as a standard for verifying the effectiveness of the perforation optimization method.

[0153] S2: Predict reservoir physical parameters based on well logging response data.

[0154] Specifically, this includes: calculating the porosity φ of the target well based on the acoustic transit time AC and the Willy formula, wherein the AC value is the pre-processed curve value, Δt maΔtf is set according to the sandstone and mudstone standards within the block where the target well is located; a fitting function is constructed based on the exponential relationship between the measured porosity and permeability data of the core, K=a×e^(b×φ), and the parameters a and b are calibrated by fitting at least 50 sets of core data; then, the Archie formula Sw=((Rw×φ) is used. m The oil saturation So is calculated using the formula φ / (Rt)^(1 / n), where Rw is the mean formation water resistivity, and m and n are parameters obtained from core experiments. The obtained φ, K, and So curves are aligned by depth, and abnormal well diameter segments (CAL greater than 1.2 times the casing inner diameter) are removed to eliminate interference from abnormal data. See Table 1 for details: well logging data and property prediction data for a certain well in the block.

[0155] Table 1

[0156]

[0157] Table 1 shows the logging response data and corresponding predicted physical property parameters of target wells in a certain oil and gas block at specific depths (e.g., 395.10033 m and 395.19953 m). The logging curve data in the table include natural gamma (GR), sonic transit time (AC), microsphere resistivity (R05), spontaneous potential (SP), deep lateral resistivity (RLLD), and shallow lateral resistivity (RLLS), which can be used to reflect the lithological, electrical characteristics, and pore structure of the formation. Based on the above logging data, the reservoir physical property prediction indicators obtained by fitting or machine learning methods include porosity (SHJ0516 and PORJ0516), permeability (PERMJ0516), and oil saturation (SOY0516), providing key input parameters for further reservoir effectiveness evaluation and perforation interval selection.

[0158] S3: Conduct a comprehensive evaluation of reservoir effectiveness using multiple parameters and predict perforation.

[0159] First, in the training sample construction phase, typical segments from developed wells in the same block are selected as positive and negative samples. Positive samples are limited to effective segments that have been verified as high-yielding through well testing and simultaneously meet the physical property parameters and oil-bearing standards. Negative samples are invalid segments verified as dry or water-bearing layers, ensuring the representativeness and breadth of the samples. Regarding feature construction, a multi-dimensional feature set is constructed by combining well logging response data with the reservoir physical properties, including: basic features such as porosity φ, permeability K, deep lateral resistivity RT, oil saturation So, and natural gamma ray GR; composite features such as φ×K (reflecting reservoir-flow synergy), RT / GR (used to correct for shale interference), and (1-Sw)×φ (reflecting oil-bearing space volume); and production-related features such as K×So (representing the production potential under the coupling of flowability and oil-bearing properties). The feature set is normalized to the 0-1 range using Min-Max normalization.

[0160] Meanwhile, in terms of feature weighting, basic weights are set based on formation seepage mechanisms: φ×K weight is not less than 25%, K≥15%, φ≥12%, and So≥13%, with the total proportion of the four being not less than 65%. Further weight adjustments are made for different types of reservoirs. For example, for low-permeability reservoirs (K<10mD), the feature weights of φ×K and (1-Sw)×φ are increased by 5% each; for high-muddy reservoirs, the RT / GR weight is increased by 8%, achieving adaptation of feature weighting to geological types.

[0161] To ensure that the output effective reservoir intervals meet the continuity requirements of engineering construction, a continuity constraint module is introduced, embedding the engineering constraint rule of "effective section thickness not less than 2m" into the evaluation model logic. Specifically, this includes: predicting the effectiveness probability for each depth point in 0.1m increments, identifying continuous sections using a 50-point sliding window (corresponding to a 5m depth range), counting the number of depth points that continuously satisfy "effective reservoir probability not less than 0.75," and calculating the corresponding continuous section length. When the continuous length is less than 2m, a penalty mechanism is triggered and it is marked as a "discontinuous section"; when the continuous length is greater than or equal to 2m, it is marked as a "continuous effective section."

[0162] The evaluation model employs a three-layer structure: a basic learner, a constraint enhancement layer, and an interactive attention layer. The basic learner uses an improved Extremely Random Tree (ERT), introducing random feature sampling and split point selection mechanisms to reduce the risk of overfitting, resulting in a reduction of over 30% in the correlation between features between trees. During model training, the gradient descent algorithm is used to optimize the prediction error (i.e., the deviation between the predicted effective probability and the actual sample label), dynamically optimizing split nodes and parameter weights to improve the fitting accuracy of the relationship between reservoir properties and oil content.

[0163] The constraint enhancement layer includes hard and soft constraint mechanisms: Hard constraints require that core features (φ×K, K, φ, So) participate in feature splitting in each round of tree structure construction, with a splitting priority weight of no less than 60%; if the cumulative weight is less than 65%, a weight correction procedure is automatically triggered to increase its splitting contribution. The soft constraint mechanism introduces a "geological penalty factor." When the weight of the "φ×K" feature in a low-permeability reservoir is less than 25%, a penalty term in the loss function is automatically added to drive the model towards convergence in a direction consistent with geological laws.

[0164] The interactive attention layer employs a self-attention mechanism to quantify the interaction strength between features. By constructing query (Q), key (K), and value (V) vectors, it calculates the dot product attention values ​​between feature pairs, normalizes them, and then weights and aggregates them to model the collaborative relationships between features. For example, in high-permeability reservoirs, it enhances the interaction weights of K and So to reinforce the logic of "high permeability requires high oil content"; in high-muddy reservoirs, it enhances the interaction relationship between GR and RT to correct the influence of muddy interference on oil content judgment.

[0165] The final model optimizes the output results through a dual-loop mechanism: the inner loop focuses on model performance, controlling the weight allocation ratio of core features (≥65%) and the contribution ratio of interaction weights (≥10%), and optimizes the model structure and parameters by combining 5-fold cross-validation; the outer loop focuses on engineering validation, using oil testing data from more than 10 development wells in the same block to evaluate the consistency between the model output and the oil testing results, ensuring that the prediction results are within the range of "effective reservoir identification accuracy ≥90%" and "continuous segment development success rate ≥95%".

[0166] If a well section has a deviation exceeding 15%, sample or weight reconstruction and retraining will be automatically triggered. The final output includes: the effective reservoir probability, continuous section length, and engineering effectiveness label for each depth point. Intervals with "effective probability ≥ 0.75 and continuous length ≥ 2m" are selected to form a "Target Perforation Interval Recommendation Report" to support refined perforation site selection and parameter configuration.

[0167] Upload the logging curves of the target new well, call the "data preprocessing script" to complete: depth calibration, missing value completion, and feature calculation (φ, K, So, and composite features consistent with the training phase), ensuring that the data format matches the training data. Load the trained model file and run the "prediction script"; the code calculates the effective reservoir probability and continuity length at each depth point of the new well, filters the intervals with "probability ≥ 0.75 and continuity ≥ 2m" according to the rules, and generates the final perforation location. See Table 2 for the perforation prediction data table of a certain well in the block.

[0168] Table 2

[0169]

[0170] Table 2 contains fields covering key logging parameters and property prediction results used for perforation interval identification and evaluation. DEPT indicates the well depth location of the logging point, and Perf1231 is the perforation marker (a value of 1 indicates that the interval is determined to be a valid interval that meets the perforation conditions). GR (natural gamma), AC (acoustic transit time), R05 (microsphere resistivity), SP (spontaneous potential), RLLD and RLLS (deep and shallow lateral resistivity) are typical logging response features used to characterize formation lithology, electrical properties and fluid characteristics. SHJ0516, PERMJ0516, SOY0516 and PORJ0516 are porosity, permeability, oil saturation and spare porosity values ​​obtained through model prediction, respectively, reflecting the reservoir's physical properties and development potential. The overall field system together constitutes the data foundation supporting perforation point location judgment and parameter optimization configuration.

[0171] In some embodiments, see Figure 4The figure shows the logging interpretation and reservoir effectiveness prediction results of well D11-5 within a depth range of 395m to 406m. From left to right, the figure includes the perforation determination label (Perf1231), microsphere resistivity curve (R05), predicted porosity (SHJ0516), alternative porosity (PORJ0516PH), predicted oil saturation (SOY0516PH, SOJ0516PH), predicted permeability (PERMJ0516PH), and lithological interpretation results. The "Open" and "Closed" areas indicate recommended and non-recommended perforation intervals determined based on multidimensional characteristics and rule constraints; they also show sandstone, mudstone, and non-reservoir intervals, respectively. It is clearly observed that in well intervals with high effective reservoir porosity, permeability, and oil saturation, and interpreted as sandstone, the corresponding perforation label is marked "Open," indicating that these intervals meet reservoir quality and development potential requirements and are suitable as preferred perforation target areas. The graphical results intuitively reflect the process of segment selection and perforation recommendation under the fusion of reservoir property prediction and decision rules.

[0172] While this specification provides the steps of operation for the methods described in the embodiments or flowcharts, more or fewer steps may be included based on conventional or non-inventive means. The order of steps listed in the embodiments is merely one possible order of execution among many steps and does not represent the only possible order. In actual device or client product execution, the methods shown in the embodiments or drawings may be executed sequentially or in parallel (e.g., in a parallel processor or multi-threaded processing environment, or even a distributed data processing environment). The terms "comprising," "including," or any other variations thereof are intended to cover a non-exclusive inclusion, such that a process, method, product, 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, product, or apparatus. Without further limitations, the presence of other identical or equivalent elements in a process, method, product, or apparatus that includes said elements is not excluded. The terms "first," "second," etc., are used to denote names and do not indicate any particular order.

[0173] Those skilled in the art will also know that, besides implementing the controller using purely computer-readable program code, the same functions can be achieved by logically programming the method steps, making the controller function as logic gates, switches, application-specific integrated circuits (ASICs), programmable logic controllers (PLCs), and embedded microcontrollers. Therefore, such a controller can be considered a hardware component, and the devices within it used to implement various functions can also be considered structures within that hardware component. Alternatively, the devices used to implement various functions can be considered as both software modules implementing the method and structures within a hardware component.

[0174] As can be seen from the above description of the embodiments, those skilled in the art can clearly understand that this specification can be implemented by means of software plus necessary general-purpose hardware platforms. Based on this understanding, the technical solutions of this specification can essentially be embodied in the form of a software product. This computer software product can be stored in a storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, mobile terminal, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments of this specification.

[0175] Although this specification has been described by way of examples, those skilled in the art will recognize that many variations and modifications are possible without departing from the spirit of this specification, and it is intended that the appended claims cover such variations and modifications without departing from the spirit of this specification.

Claims

1. A method for determining reservoir perforation intervals, characterized in that, include: Acquire logging response data and core measurement data of the target well in the target area; Based on the logging response data and the core measurement data, multiple reservoir physical parameters along the depth direction of the target well are determined; Based on the reservoir physical properties and the logging response data, a multidimensional feature set is determined; wherein, the multidimensional feature set is used to determine the effectiveness of the reservoir. Using a pre-defined reservoir effectiveness prediction model, the effective reservoir probability at each depth point of the target well is determined based on the multi-dimensional feature set; wherein, the pre-defined reservoir effectiveness prediction model includes a feature fitting module, a constraint rule execution module, and a feature interaction calculation module; Based on the effective reservoir probability, an effective reservoir interval that meets the continuity requirement is determined, and the effective reservoir interval is used as the target perforation section of the target well; wherein, the target perforation section is used to guide the determination of the perforation location of the target well.

2. The method according to claim 1, characterized in that, The step of determining the effective reservoir interval that meets the continuity requirement based on the effective reservoir probability, and using the effective reservoir interval as the target perforation interval of the target well, includes: Based on the effective reservoir probability, identify continuous effective reservoir segments in the target well and calculate the length of each continuous effective segment. The effective reservoir interval that meets the continuity requirement is determined by the depth interval where the effective reservoir probability is not less than a first preset threshold and the corresponding continuous effective segment length is not less than a second preset threshold, and the effective reservoir interval is used as the target perforation segment of the target well.

3. The method according to claim 1, characterized in that, The determination of a multidimensional feature set based on the reservoir physical parameters and the logging response data includes: Based on the reservoir physical properties such as porosity, permeability, and oil saturation, as well as the resistivity and natural gamma in the logging response data, a response characteristic reflecting the reservoir response properties is constructed. Based on the combined relationships between porosity and permeability, resistivity and natural gamma, and porosity and oil saturation, a composite characteristic reflecting the synergistic relationship between reservoir capacity and oil content is constructed. Based on the permeability and oil saturation, a capacity correlation feature reflecting the reservoir's capacity potential is constructed. The response features, composite features, and capacity-related features are combined to determine a multidimensional feature set.

4. The method according to claim 1, characterized in that, The step of using a preset reservoir effectiveness prediction model to determine the effective reservoir probability at each depth point of the target well based on the multidimensional feature set includes: Using the feature fitting module, based on the relationship between the multidimensional feature set and the historically labeled reservoir effectiveness tags, the initial effective reservoir probability corresponding to each depth point is determined by fitting the multidimensional feature set. Using the constraint rule execution module, the intermediate effective reservoir probability at each depth point of the target well is determined based on the initial effective reservoir probability; Using the feature interaction calculation module, the effective reservoir probability at each depth point of the target well is determined based on the intermediate effective reservoir probability.

5. The method according to claim 4, characterized in that, The step of using the constraint rule execution module to determine the intermediate effective reservoir probability at each depth point of the target well based on the initial effective reservoir probability includes: Based on the contribution value of each feature in the initial effective reservoir probability to the prediction result, the corresponding feature importance coefficient is determined; If the initial effective reservoir probability does not meet the preset engineering constraint rules, the feature weights of key features in the multidimensional feature set are adjusted according to the feature importance coefficient; wherein, the preset engineering constraint rules include the minimum continuous segment length threshold, the maximum upper and lower layer difference amplitude threshold, and the fluctuation stability threshold of the effective reservoir distribution. Based on the adjusted feature weights, the initial effective reservoir probability is corrected to obtain the intermediate effective reservoir probability at each depth point of the target well.

6. The method according to claim 5, characterized in that, The step of using the feature interaction calculation module to determine the effective reservoir probability at each depth point of the target well based on the intermediate effective reservoir probability includes: Calculate the corresponding feature interaction weight coefficients based on the vector relationships between the features in the multidimensional feature set; Based on the feature interaction weight coefficient, the intermediate effective reservoir probability is dynamically corrected to obtain the effective reservoir probability at each depth point of the target well.

7. The method according to claim 1, characterized in that, The method further includes: Based on the target perforated section, a perforation parameter configuration scheme for the target well is determined; wherein, the perforation parameter configuration scheme includes perforation density, perforation diameter, perforation phase, and perforation depth.

8. A reservoir perforation interval determination device, characterized in that, include: The data acquisition module is used to acquire logging response data and core measurement data of the target well in the target area; The parameter determination module is used to determine multiple reservoir physical property parameters along the depth direction of the target well based on the logging response data and the core measured data. The feature determination module is used to determine a multi-dimensional feature set based on the reservoir physical property parameters and the logging response data; wherein the multi-dimensional feature set is used for reservoir effectiveness determination. The probability determination module is used to determine the effective reservoir probability at each depth point of the target well based on the multidimensional feature set using a preset reservoir effectiveness prediction model; wherein, the preset reservoir effectiveness prediction model includes a feature fitting module, a constraint rule execution module, and a feature interaction calculation module. The layer segment determination module is used to determine the effective reservoir interval that meets the continuity requirement based on the effective reservoir probability, and to use the effective reservoir interval as the target perforation segment of the target well; wherein, the target perforation segment is used to guide the determination of the perforation location of the target well.

9. An electronic device, characterized in that, It includes a processor and a memory for storing processor-executable instructions, wherein the processor, when executing the instructions, implements the steps of the reservoir perforation segment determination method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, It stores computer instructions that, when executed by a processor, implement the steps of the reservoir perforation segment determination method according to any one of claims 1 to 7.

Images