Multi-well linkage drilling parameter multi-level optimization decision method and system

Abstract

The application discloses a multi-well linkage drilling parameter multi-level optimization decision method and system, and the method comprises the following steps: acquiring real-time drilling data in the drilling operation process to determine a first rock breaking parameter; in the case that the first rock breaking parameter is out of a preset range, predicting a mechanical drilling speed corresponding to the real-time drilling data according to a preset model; generating a plurality of candidate parameter groups according to the mechanical drilling speed and drilling constraint conditions, determining a function value corresponding to each candidate parameter group based on a target function of multi-objective optimization; determining a target parameter group in the plurality of candidate parameter groups according to the function value of each candidate parameter group; and determining a target drilling operation parameter according to the target parameter group, so as to take the target drilling operation parameter as a current drilling operation parameter for drilling operation. The above scheme can avoid complex downhole conditions, intelligently optimize the drilling operation parameter by using a multi-objective optimization algorithm, and comprehensively improve the drilling efficiency.

Description

Technical Field

[0001] This application relates to the field of oil and gas exploration technology, specifically to a multi-level optimization decision-making method and system for drilling parameters involving multiple wells. Background Technology

[0002] Complex oil and gas formations in deep, non-deep, and non-deep formations represent a crucial successor area for national oil and gas resource exploration and development, and also present significant challenges for drilling engineering. Given the extremely hard and difficult-to-drill deep formations, poor drillability, complex geological environments, and increased uncertainties, higher demands are placed on the precise and real-time optimization of drilling parameters. In recent years, with the rapid development of artificial intelligence technology, many scholars both domestically and internationally have widely applied it to solve the problem of recommending optimal drilling parameters during the drilling process, thereby improving drilling efficiency and achieving high-efficiency and economical drilling. However, the drilling process faces complex downhole environmental conditions, with multi-source drilling data constantly influencing fluctuations in the mechanical drilling rate, and the drilling rock-breaking mechanisms differing across different blocks, posing significant challenges to traditional drilling rate equations and intelligent prediction models. Furthermore, current drilling parameter optimization methods still have some problems. For example, traditional drilling parameter optimization methods, represented by statistical or mechanistic models, rely heavily on expert coefficients and equation assumption coefficients, resulting in limited optimization results and failing to provide field engineers with scientifically sound and reasonable drilling parameter optimization schemes. On the other hand, drilling parameter optimization methods based on intelligent models struggle to comprehensively consider complex downhole fluctuations. Furthermore, the applicability of intelligent models across different blocks is limited, and the model's portability and the rationality of the optimization parameters need improvement. In summary, both traditional drilling parameter optimization methods and existing intelligent methods rely on design data, geological data, or real-time data from the target well for calculation and optimization. These methods have low optimization efficiency, lack comparability of results, and do not fully utilize historical knowledge from adjacent wells, failing to fully explore and leverage the potential value of historical data from adjacent wells in drilling parameter optimization. Summary of the Invention

[0003] The purpose of this application is to provide a multi-level optimization decision-making method and apparatus for drilling parameters in multi-well linkage, so as to solve the problems of low data utilization efficiency and failure to consider complex downhole working conditions in the existing drilling parameter optimization methods.

[0004] To achieve the above objectives, the first aspect of this application provides a multi-level optimization decision-making method for drilling parameters in multi-well linkage, comprising:

[0005] Acquire real-time drilling data during the drilling operation, including drilling operation parameters and drill bit usage data;

[0006] The first rock-breaking parameters are determined based on real-time drilling data;

[0007] When the first rock breaking parameter is within the preset range, real-time drilling data is input into the preset model so that the preset model can output the predicted value of mechanical drilling rate. The preset model is a model trained based on historical real-time drilling data.

[0008] Multiple candidate parameter groups are generated based on the predicted mechanical drilling rate and drilling constraints. Each candidate parameter group includes multiple different types of secondary rock breaking parameters.

[0009] The function value corresponding to each candidate parameter group is determined based on the objective function. The function value is the weighted sum of all the second rock-breaking parameters included in each candidate parameter group. The objective function is solved by minimizing all normalized second rock-breaking parameters and minimizing the function value.

[0010] Determine the target parameter group from multiple candidate parameter groups based on the function value of each candidate parameter group;

[0011] The target drilling parameters are determined based on the target parameter set, and the target drilling parameters are used as the current drilling parameters for drilling operations.

[0012] In the embodiments of this application, the preset model includes a first graph learning module and a second graph learning module. Inputting real-time drilling data into the preset model to output a predicted value of the mechanical drilling rate includes: acquiring historical real-time drilling data, which includes historical drilling operating parameters, formation stratification data, and historical drill bit usage data; and based on the first graph learning module, performing dynamic clustering processing on the historical drilling operating parameters and historical drill bit usage data according to the formation stratification data to generate a formation information pre-training map, wherein the formation information pre-training map uses each type of formation as a node. Furthermore, each node is represented and embedded based on historical drilling operation parameters and historical drill bit usage data; based on the second graph learning module, the drilling operation parameters and drill bit usage data are serialized and reconstructed to generate a drilling parameter association knowledge graph; the drilling parameter association knowledge graph is extracted for temporal patterns through a temporal convolutional network to obtain temporal features; the temporal features and the adjacency matrix of the pre-trained formation information graph are fused and convolved through a graph encoding module to obtain high-dimensional hidden features; the high-dimensional hidden features are decoded and mapped to obtain the predicted value of the mechanical drilling rate output by the preset model.

[0013] In the embodiments of this application, the generation of a formation information pre-training map based on the first graph learning module's dynamic clustering of historical drilling parameters and drill bit usage data according to formation stratification data includes: performing dynamic clustering of historical drilling parameters and drill bit usage data according to formation stratification data based on the first graph learning module to determine the historical drilling parameters and drill bit usage data corresponding to each type of formation; representing each type of formation as a node in the formation information pre-training map, and embedding the historical drilling parameters and drill bit usage data corresponding to each type of formation into the corresponding node to generate the formation information pre-training map.

[0014] In the embodiments of this application, the serialization reconstruction of drilling working parameters and drill bit usage data based on the second graph learning module to generate a drilling parameter association knowledge graph includes: serializing and reconstructing drilling working parameters and drill bit usage data based on the second graph learning module to obtain a first sequence of data corresponding to the drilling working parameters and a second sequence of data corresponding to the drill bit usage data; and generating a drilling parameter association knowledge graph based on the first sequence of data and the second sequence of data.

[0015] In the embodiments of this application, the high-dimensional hidden features are determined according to the following formula (1):

[0016] (1)

[0017] in, This refers to high-dimensional hidden features. as well as middle, Both refer to the first and second sequence data, and A refers to the adjacency matrix of the stratigraphic information pre-trained map. It refers to the transpose of the adjacency matrix. This refers to the weight matrix of the learning module in the first graph. This refers to the weight matrix of the second graph learning module. This refers to the activation function of the learning module in the first image. This refers to the activation function of the learning module in the second graph.

[0018] In the embodiments of this application, determining the function value corresponding to each candidate parameter group based on the objective function includes: normalizing each second rock-breaking parameter included in each candidate parameter group to obtain the processed second rock-breaking parameter; constructing an objective function based on all the processed second rock-breaking parameters corresponding to each candidate parameter group and the weight value corresponding to each second rock-breaking parameter to determine the corresponding function value.

[0019] In the embodiments of this application, the functional expression of the objective function is shown in the following formula (2):

[0020] (2)

[0021] Where J refers to the objective function value, ROP refers to the mechanical drilling rate, DRIMP refers to the drill bit wear evaluation index, which is determined based on the ratio of drilling pressure to drill bit feed rate, MSE refers to mechanical specific energy, and RF refers to the ratio factor, which is determined based on the ratio of mechanical specific energy to the drill bit wear index. , These refer to the weight and normalization coefficient corresponding to the mechanical drilling rate, respectively. 、 This refers to the weights and normalization coefficients corresponding to the drill bit wear evaluation indicators. 、 This refers to the weight and normalization coefficient corresponding to the mechanical specific energy. 、 This refers to the weights and normalization coefficients corresponding to the ratio factor.

[0022] In embodiments of this application, the method further includes: acquiring historical real-time drilling data, which includes historical drilling operating parameters, formation layering data, and historical drill bit usage data; constructing a target document based on the historical real-time drilling data; dividing the target document into blocks to obtain multiple document fragments; determining the fragment vector corresponding to each document fragment; storing all fragment vectors in a vector database; searching the vector database for the target fragment vector corresponding to the real-time drilling data; and inputting the document fragment corresponding to the target fragment vector and the real-time drilling data into a large language model to output drilling constraints related to the current drilling operation through the large language model, wherein the drilling constraints include a range of safety parameters corresponding to the drilling operating parameters.

[0023] In the embodiments of this application, the drilling operating parameters include the drill bit's pressure on the drill bit, rotational speed, and displacement, and the drilling constraints include a first parameter range for pressure on the drill bit, a second parameter range for rotational speed, and a third parameter range for displacement.

[0024] The second aspect of this application provides a multi-well linkage drilling parameter multi-level optimization decision system, including:

[0025] The memory is configured to store instructions;

[0026] The processor is configured to retrieve instructions from memory and, when executing instructions, to implement a multi-level optimization decision-making method based on the aforementioned multi-well linkage drilling parameters.

[0027] The above technical solution acquires real-time drilling data during the drilling operation, including drilling parameters and drill bit usage data. Based on this data, a first rock-breaking parameter is determined. If the first rock-breaking parameter is outside a preset range, the real-time drilling data is input into a preset model to output a predicted value of the mechanical drilling rate. This preset model is trained based on historical real-time drilling data. Multiple candidate parameter groups are generated based on the predicted mechanical drilling rate and drilling constraints. Each candidate parameter group includes multiple different types of second rock-breaking parameters. The objective function is then used to determine... The function value corresponding to each candidate parameter group is a weighted sum of all second rock-breaking parameters included in each candidate parameter group. The objective function is solved by minimizing all normalized second rock-breaking parameters and the function value. Based on the function value of each candidate parameter group, the target parameter group among multiple candidate parameter groups is determined. The target drilling parameters are determined based on the target parameter group, and the target drilling parameters are used as the current drilling parameters for drilling operations. While avoiding complex downhole conditions, the drilling parameters are intelligently optimized using a multi-objective optimization algorithm to comprehensively improve drilling efficiency.

[0028] Other features and advantages of the embodiments of this application will be described in detail in the following detailed description section. Attached Figure Description

[0029] The accompanying drawings are provided to further illustrate the embodiments of this application and form part of the specification. They are used together with the following detailed description to explain the embodiments of this application, but do not constitute a limitation on the embodiments of this application. In the drawings:

[0030] Figure 1 The illustration shows a flowchart of a multi-well linkage drilling parameter multi-level optimization decision-making method according to an embodiment of this application;

[0031] Figure 2 This illustration schematically shows a process diagram for multi-objective real-time optimization of drilling operating parameters according to an embodiment of this application;

[0032] Figure 3 The illustration shows a schematic diagram of a process for real-time prediction of mechanical drilling rate embedded with formation knowledge according to an embodiment of this application;

[0033] Figure 4 The illustration shows a schematic diagram of the process for searching and recommending drilling operating parameters according to an embodiment of this application;

[0034] Figure 5 This illustration schematically shows a flowchart of multi-level optimization decision-making for drilling operating parameters according to an embodiment of this application;

[0035] Figure 6This schematic diagram illustrates a structural block diagram of a multi-well linkage drilling parameter multi-level optimization decision system according to an embodiment of this application;

[0036] Figure 7 The illustration shows a schematic diagram of the structure of a computer device according to an embodiment of the present application. Detailed Implementation

[0037] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only for illustration and explanation of the embodiments of this application and are not intended to limit the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0038] It should be noted that if the embodiments of this application involve directional indicators (such as up, down, left, right, front, back, etc.), the directional indicators are only used to explain the relative positional relationship and movement of each component in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indicators will also change accordingly.

[0039] Furthermore, if the embodiments of this application involve descriptions such as "first" or "second," these descriptions are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, features defined with "first" or "second" may explicitly or implicitly include at least one of those features. Additionally, the technical solutions of various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. If the combination of technical solutions is contradictory or impossible to implement, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed in this application.

[0040] Figure 1 The illustration shows a flowchart of a multi-well linkage drilling parameter multi-level optimization decision-making method according to an embodiment of this application. Figure 1 As shown in the embodiment of this application, a multi-level optimization decision-making method for drilling parameters in multi-well linkage is provided, which may include the following steps.

[0041] S102, acquire real-time drilling data during the drilling operation. The real-time drilling data includes drilling operation parameters and drill bit usage data.

[0042] It is understood that drilling parameters include, but are not limited to, well depth, pressure on drill bit, torque, rotary table speed, standpipe pressure, outlet flow rate, outlet density, hook load, equivalent density, outlet conductivity, outlet temperature, and mechanical drilling speed. Drill bit usage data includes, but is not limited to, drill bit serial number, size, model, series number, depth into the well, depth out of the well, and drill bit footage.

[0043] S104, determine the first rock-breaking parameters based on real-time drilling data.

[0044] It is understandable that the first rock-breaking parameters include the ratio factor and mechanical specific energy, both of which are calculated based on real-time drilling data.

[0045] S106, when the first rock breaking parameter is within the preset range, real-time drilling data is input into the preset model so that the predicted value of mechanical drilling rate is output through the preset model. The preset model is a model trained based on historical real-time drilling data.

[0046] It is understandable that complex downhole conditions such as drill bit vibration, drill bit mud buildup, and drill bit wear can lead to abnormally low drilling rates. Therefore, drilling conditions can be monitored in real time using ratio factors and mechanical specific energy. Based on the SPOT (Streaming Anomaly Detection with Extreme Values) algorithm, extreme values ​​at a specified probability are calculated, and these extreme values ​​are considered the threshold for determining anomalies. This application uses this method for anomaly detection because it is simple, easy to implement, and fast, making it suitable for real-time intelligent optimization systems for drilling parameters. It is understood that corresponding preset ranges are set for anomaly detection of ratio factors and mechanical specific energy. If the ratio factor and mechanical specific energy are outside their respective preset ranges, an anomaly is considered to have occurred. Specifically, the preset ranges include a first preset range for the ratio factor and a second preset range for the mechanical specific energy. When the real-time calculated ratio factor is within the first preset range and the mechanical specific energy is within the second preset range, an adjustment to the mechanical drilling rate is triggered. It is understandable that, further, real-time drilling data is input into a preset model so that the preset model can output a predicted value of the mechanical drilling rate, which can be considered as the best working parameters for the current drilling operation.

[0047] In the embodiments of this application, reference is made to Figure 2If the ratio factor and mechanical specific energy are outside the corresponding first and second preset ranges—for example, if the ratio factor is greater than the maximum value of the first preset range and the mechanical specific energy parameter is greater than the maximum value of the second preset range—it is considered abnormal drill bit wear and drill bit mud buildup. In this case, it is recommended to pull out the drill string and replace the drill bit. If the ratio factor is much lower than the minimum threshold of the first preset range and the mechanical specific energy is greater than the maximum threshold of the second preset range, it is determined that the drill bit is vibrating, and drilling parameters need to be reduced to mitigate the abnormal situation. If the ratio factor is within the first preset range and the mechanical specific energy is within the second preset range, the mechanical drilling rate can be predicted using a preset model. Real-time drilling data is input into the preset model for real-time drilling rate prediction, and the predicted value of the mechanical drilling rate is output through the preset model. Furthermore, multi-objective optimization decisions can be made by combining the constraints of complex working conditions and recommending the optimal drilling parameters in real time.

[0048] In the embodiments of this application, the preset model includes a first graph learning module and a second graph learning module. Inputting real-time drilling data into the preset model to output a predicted value of the mechanical drilling rate includes: acquiring historical real-time drilling data, which includes historical drilling operating parameters, formation stratification data, and historical drill bit usage data; and based on the first graph learning module, performing dynamic clustering processing on the historical drilling operating parameters and historical drill bit usage data according to the formation stratification data to generate a formation information pre-training map, wherein the formation information pre-training map uses each type of formation as a node. Furthermore, each node is represented and embedded based on historical drilling operation parameters and historical drill bit usage data; based on the second graph learning module, the drilling operation parameters and drill bit usage data are serialized and reconstructed to generate a drilling parameter association knowledge graph; the drilling parameter association knowledge graph is extracted for temporal patterns through a temporal convolutional network to obtain temporal features; the temporal features and the adjacency matrix of the pre-trained formation information graph are fused and convolved through a graph encoding module to obtain high-dimensional hidden features; the high-dimensional hidden features are decoded and mapped to obtain the predicted value of the mechanical drilling rate output by the preset model.

[0049] In the embodiments of this application, the generation of a formation information pre-training map based on the first graph learning module's dynamic clustering of historical drilling parameters and drill bit usage data according to formation stratification data includes: performing dynamic clustering of historical drilling parameters and drill bit usage data according to formation stratification data based on the first graph learning module to determine the historical drilling parameters and drill bit usage data corresponding to each type of formation; representing each type of formation as a node in the formation information pre-training map, and embedding the historical drilling parameters and drill bit usage data corresponding to each type of formation into the corresponding node to generate the formation information pre-training map.

[0050] In the embodiments of this application, the serialization reconstruction of drilling working parameters and drill bit usage data based on the second graph learning module to generate a drilling parameter association knowledge graph includes: serializing and reconstructing drilling working parameters and drill bit usage data based on the second graph learning module to obtain a first sequence of data corresponding to the drilling working parameters and a second sequence of data corresponding to the drill bit usage data; and generating a drilling parameter association knowledge graph based on the first sequence of data and the second sequence of data.

[0051] It is understandable that historical drilling data is obtained by collecting and organizing the data of wells already drilled in the current block. Historical drilling parameters include, but are not limited to, historical well depth, drilling pressure, torque, rotary table speed, standpipe pressure, outlet flow rate, outlet density, hook load, equivalent density, outlet conductivity, outlet temperature, and mechanical drilling rate. Historical drill bit usage data includes, but is not limited to, historical drill bit serial number, size, model, series number, depth into the well, depth out of the well, and footage. Formation stratification data includes, but is not limited to, layer position, top depth, bottom depth, top vertical depth, bottom vertical depth, and layer depth.

[0052] The historical drilling data needs to be cleaned and processed, and the main process is as follows: First, data samples from the drilling process are screened based on logical rules. Methods such as 3σ outlier removal, linear regression interpolation, and sliding filtering are used to filter out downhole anomalies and noise data, improving data sample quality. Second, the influence of different parameters' dimensions is eliminated through data minimax normalization. The processed data is then saved as tabular data. Finally, using well depth as a data index, the historical drilling operation parameter data table, formation stratification data table, and drill bit usage records are merged to obtain multi-source historical drilling data. Subsequently, multi-well knowledge is extracted and the model is pre-trained using this multi-source historical drilling data to obtain a preset model.

[0053] It is understandable that the preset model includes a first graph learning module and a second graph learning module. In this way, a preset model of the mapping relationship between multi-source historical drilling parameters and mechanical drilling rate can be established using graph learning methods. Then, the preset model can be optimized by combining real-time drilling data to quickly adapt to the current drilling formation environment.

[0054] Specifically, refer to Figure 3Based on the first-graph learning module, historical drilling parameters (historical logging data) and historical drill bit usage data are dynamically clustered according to formation stratification data to determine the historical drilling parameters and drill bit usage data corresponding to each formation class, thus achieving automatic geological unit division. This process effectively reduces the time cost of manual division while maintaining high classification accuracy and achieves fully automated workflow. Furthermore, each formation class is represented as a node in the formation information pre-training map, and the historical drilling parameters and drill bit usage data corresponding to each formation class are embedded into the corresponding node to generate the formation information pre-training map. The specific process of the first-graph learning module involves obtaining a multi-well, multi-formation geological unit dataset, which forms a model training set. Then each stratum is a subset. Where N is the number of samples, C represents the stratigraphic type, T represents the sliding window length, and D represents the input feature dimension. Subsequently, a pre-trained stratigraphic information graph is constructed using the first graph learning module. Here, the stratigraphic type C can be used as the number of nodes in the graph. Node embedding learning is performed on the input sequence to obtain the initial node embedding matrix. , where M is the embedding dimension. Subsequently, the embedding information is learned and standardized using tanh activation function and ReLU activation function to ensure the non-negativity of the adjacency matrix of the stratigraphic information pre-trained graph.

[0055] The adjacency matrix of the stratigraphic information pre-training map is expressed as the following formula (3):

[0056] (3)

[0057] in, , , It is the transformed feature hiding representation 1. Feature hiding representation 2 after transformation It is a randomly initialized node embedding matrix 1. It is another randomly initialized matrix 2. , These are all parameters that can be learned during model training. It is a learnable transformation matrix 1. It is a learnable transformation matrix 2, used to map the original features to a new feature space. It refers to the adjacency matrix of the pre-trained stratigraphic information map. available Each feature and Similarity of features This reflects the asymmetry of the learned adjacency matrix.

[0058] Another part of the preset model is the second graph learning module, which can use the sliding window method to serialize and reconstruct the real-time acquired drilling working parameters and drill bit usage data to obtain the first sequence data corresponding to the drilling working parameters and the second sequence data corresponding to the drill bit usage data. These are represented as nodes in the new graph, so as to generate a drilling parameter association knowledge graph based on the first sequence data and the second sequence data.

[0059] Furthermore, the pre-defined model also includes a knowledge hybrid encoder, which can extract, transform, and represent multivariate features from the learned formation information pre-training map and drilling parameter association knowledge graph. It encodes and represents the different mechanical drilling rate representation relationships between different formations and the potential mapping relationships contained in drilling working parameters within the same formation. Finally, the output layer obtains the predicted value of the mechanical drilling rate. Specifically, in the embodiments of this application, a temporal convolutional network and a graph coding module are selected as the hybrid encoder, which is constructed by cross-concatenating temporal coding and graph coding. The dilated convolutional blocks of the temporal convolutional network (TCN) are used to extract temporal patterns of different lengths from the two sequences included in the drilling parameter association knowledge graph to obtain temporal features. Compared with traditional convolutional networks, they can capture more varied features at different scales, thereby effectively extracting pattern information from temporal data and increasing the receptive field. The graph coding module can perform fusion convolution on the temporal features and the adjacency matrix of the formation information pre-training map to obtain high-dimensional hidden features. Furthermore, the obtained high-dimensional hidden features are decoded and mapped to the predicted target parameter, i.e., mechanical drilling rate, using a fully connected layer to obtain the predicted value of the mechanical drilling rate output by the preset model.

[0060] In the embodiments of this application, the high-dimensional hidden features are determined according to the following formula (1):

[0061] (1)

[0062] in, This refers to high-dimensional hidden features. as well as middle, Both refer to the first and second sequence data, and A refers to the adjacency matrix of the stratigraphic information pre-trained map. It refers to the transpose of the adjacency matrix. This refers to the weight matrix of the learning module in the first graph. This refers to the weight matrix of the second graph learning module. This refers to the activation function of the learning module in the first image. This refers to the activation function of the learning module in the second graph.

[0063] The above scheme constructs a pre-trained formation information graph and a knowledge graph relating drilling parameters using graph learning methods. Specifically, it performs node embedding and adaptive graph learning for each formation and its multi-source historical drilling data. The pre-trained formation information graph links the relationships between drilling parameters across different formations and provides knowledge guidance for the currently drilling formation. The knowledge hybrid encoder can be composed of various deep learning networks, including but not limited to fully connected networks (FNN), convolutional networks (CNN), recurrent neural networks (RNN), and graph networks (GNN). Model training hyperparameters include learning rate, batch size, and number of training epochs.

[0064] In the embodiments of this application, after accessing real-time drilling data from the field, a sliding window can be used to reconstruct its time series. Next, a hybrid encoder identical to the preset model is established and learned in real-time from the current formation's real-time drilling data. Simultaneously, only the weight parameters of the last layer of the preset model are fine-tuned, while the weights of the remaining layers are frozen. The feature output of the preset model is concatenated with the feature information of the real-time drilling data, and the final predicted drilling rate is output through a feature fusion layer.

[0065] In the embodiments of this application, after the preset model is established, the predictive performance of the model is verified using three model evaluation metrics, and the optimal model is determined by comprehensively considering the three metrics. The evaluation metrics selected are Mean Absolute Percentage Error (MAPE), Mean Absolute Error (MAE), and Empirical Correlation Coefficient (CORR). Their calculation formulas are shown in the following formulas (4), (5), and (6):

[0066] (4)

[0067] (5)

[0068] (6)

[0069] Where n is the number of samples. Let i be the true value of the i-th sample. Let be the model prediction value for the i-th sample. To fully utilize real-time drilling data and improve model accuracy, the model needs to be continuously trained at each fixed drilling depth.

[0070] S108 generates multiple candidate parameter groups based on the predicted mechanical drilling rate and drilling constraints. Each candidate parameter group includes multiple different types of second rock breaking parameters.

[0071] It is understandable that after predicting the mechanical rate of penetration (MRP), other drilling parameters need to be adjusted accordingly. These drilling parameters are constrained by the complex drilling environment and conditions; that is, drilling constraints are specific to the drilling parameters. For example, drilling constraints can limit the range of drilling parameters. Therefore, the predicted RPP and the range of drilling parameters are used to calculate multiple different types of secondary rock-breaking parameters. These multiple types of secondary rock-breaking parameters can form a candidate parameter set, from which the optimal parameter set is selected for subsequent optimization algorithms to perform the drilling operation.

[0072] In the embodiments of this application, the drilling operating parameters include the drill bit's pressure on the drill bit, rotational speed, and displacement, and the drilling constraints include a first parameter range for pressure on the drill bit, a second parameter range for rotational speed, and a third parameter range for displacement.

[0073] Specifically, drilling constraints are determined by the maximum safe operating parameters of the drilling rig and the range of parameters related to complex downhole conditions. The range of drilling constraints is shown in the following formula (7):

[0074] (7)

[0075] in, , These are the minimum and maximum values ​​of drilling pressure (Wob), rotational speed (Rpm), and displacement (Q), respectively.

[0076] S110, determine the function value corresponding to each candidate parameter group based on the objective function, where the function value is the weighted sum of all the second rock-breaking parameters included in each candidate parameter group, and the objective function is solved by minimizing all normalized second rock-breaking parameters and minimizing the function value.

[0077] Specifically, under the constraints of drilling parameters, a multi-objective optimization algorithm is used to screen out a set of candidate feasible solutions that meet the conditions. Different decision-making methods are then used to calculate the objective scores of different schemes within the feasible solution set, and the parameter combination with the highest score is selected as the final recommended parameter scheme. Specifically, the multi-objective optimization algorithm is implemented through an objective function, which is constructed by weighted summation of each normalized secondary rock-breaking parameter. The secondary rock-breaking parameters include mechanical drilling rate, bit wear index, ratio factor, and mechanical energy. The trend of the bit wear index can monitor bit wear. Mechanical energy refers to the energy consumed by the drill bit to break a unit volume of rock. The ratio factor is used to quickly identify complex working conditions of drill bit vibration and wear. The mechanical drilling rate can be used to calculate the bit feed rate, and other secondary rock-breaking parameters can be calculated from the bit feed rate. The bit feed rate helps analyze the drill bit's penetration into the formation and its wear condition; normally, a larger bit feed rate results in a faster mechanical drilling rate. Therefore, the objective function aims to minimize the solutions for the drill bit wear index, ratio factor, and mechanical specific energy, while maximizing the solution for the mechanical drilling rate. The objective function is solved by minimizing its value. The mechanical drilling rate can be normalized, and the normalized mechanical drilling rate also aims to minimize its solution, thus minimizing the function value obtained by weighted summation of the normalized second rock-breaking parameters.

[0078] In the embodiments of this application, determining the function value corresponding to each candidate parameter group based on the objective function includes: normalizing each second rock-breaking parameter included in each candidate parameter group to obtain the processed second rock-breaking parameter; constructing an objective function based on all the processed second rock-breaking parameters corresponding to each candidate parameter group and the weight value corresponding to each second rock-breaking parameter to determine the corresponding function value.

[0079] In the embodiments of this application, the functional expression of the objective function is shown in the following formula (2):

[0080] (2)

[0081] Where J refers to the objective function value, ROP refers to the mechanical drilling rate, DRIMP refers to the drill bit wear evaluation index, which is determined based on the ratio of drilling pressure to drill bit feed rate, MSE refers to mechanical specific energy, and RF refers to the ratio factor, which is determined based on the ratio of mechanical specific energy to the drill bit wear index. , These refer to the weight and normalization coefficient corresponding to the mechanical drilling rate, respectively. 、 This refers to the weights and normalization coefficients corresponding to the drill bit wear evaluation indicators. 、 This refers to the weight and normalization coefficient corresponding to the mechanical specific energy. 、 This refers to the weights and normalization coefficients corresponding to the ratio factor.

[0082] Specifically, the calculation process for the second rock-breaking parameter is as follows:

[0083] (1) Drill bit feed rate: The drill bit feed rate can help analyze the drill bit's penetration into the formation and the drill bit's wear. Under normal circumstances, the larger the drill bit feed rate, the faster the mechanical drilling speed. Calculate the drill bit feed rate (DOC) according to the following formula (8):

[0084] (8)

[0085] in, It is the mechanical drilling rate (ft / hr). It refers to the rotational speed (rpm).

[0086] (2) Drill bit wear index: Its changing trend can monitor the wear of the drill bit. The drill bit wear index (DRIMP) is calculated according to the following formula (9):

[0087] (9)

[0088] In the formula, It is a drill bit wear evaluation index (lb / in). It is drilling pressure (lbs). It is the drill bit feed rate (in).

[0089] (3) Mechanical specific energy: The energy consumed by the drill bit to break a unit volume of rock. The mechanical specific energy (MSE) is calculated according to the following formula (10):

[0090] (10)

[0091] In the formula, It is mechanical specific energy (Psi); It is the drill bit diameter (in). It is the drill bit torque (ft×lbf).

[0092] (4) Ratio Factor: This index is used to quickly identify complex working conditions such as drill bit vibration and wear. The ratio factor (RF) is calculated according to the following formula (11):

[0093] (11)

[0094] In the formula, RF is the ratio factor (ft / in). It is particularly important to note that when RF and WOB increase simultaneously, the system is considered to have drill bit vibration; when RF decreases and WOB increases, the system is considered to have drill bit wear.

[0095] S112, determine the target parameter group from multiple candidate parameter groups based on the function values ​​of each candidate parameter group. For example, the candidate parameter group with the smallest function value can be selected as the target parameter value.

[0096] Specifically, the optimization process of the above objective function can utilize various swarm optimization algorithms, such as NSGA-II, genetic algorithm, particle swarm optimization, whale optimization, and gray wolf optimization. Subsequently, optimal decisions are made based on parameter fluctuations, recommending the parameter combination with the highest comprehensive score as the optimal parameter recommendation scheme for the formation to be drilled.

[0097] In another specific embodiment, the current rock-breaking parameters can be calculated based on real-time drilling parameters. After selecting a set of candidate feasible solutions that meet the conditions using a multi-objective optimization algorithm employed by the objective function, the Euclidean distance between each candidate drilling parameter and the real-time drilling parameters can be calculated for each parameter group in the candidate feasible solution set. The candidate drilling parameter with the smallest Euclidean distance is then selected as the optimal parameter recommendation for the formation to be drilled. This minimizes the adjustment range of various operating parameters by the drilling equipment, thereby improving drilling efficiency.

[0098] S114, Determine the target drilling parameters based on the target parameter set, and use the target drilling parameters as the current drilling parameters for drilling operations.

[0099] In embodiments of this application, the method further includes: acquiring historical real-time drilling data, which includes historical drilling operating parameters, formation layering data, and historical drill bit usage data; constructing a target document based on the historical real-time drilling data; dividing the target document into blocks to obtain multiple document fragments; determining the fragment vector corresponding to each document fragment; storing all fragment vectors in a vector database; searching the vector database for the target fragment vector corresponding to the real-time drilling data; and inputting the document fragment corresponding to the target fragment vector and the real-time drilling data into a large language model to output drilling constraints related to the current drilling operation through the large language model, wherein the drilling constraints include a range of safety parameters corresponding to the drilling operating parameters.

[0100] refer to Figure 4The target document refers to the drilling knowledge base, i.e., the drilling document, constructed by statistically analyzing drilling parameters from different formations and combining expert experience and industry standards. Utilizing retrieval enhancement techniques, the knowledge base information is fully leveraged by segmenting the drilling document into multiple document fragments, generating corresponding embedding vectors (i.e., fragment vectors). Users can retrieve information from the vector database by asking questions. The retrieved fragments are then merged with the original question to form richer contextual information. This enables the large model to perform reasonable and scientific boundary reasoning for drilling parameters, outputting drilling constraints related to the current drilling operation through the large language model—that is, the safe parameter range corresponding to the drilling parameters under the current drilling conditions.

[0101] Specifically, drilling operation knowledge is extracted from historical drilling data, expert experience, and formation patterns to construct a drilling operation acceleration knowledge base. The knowledge base contains acceleration cases for different formations and different drill string combinations, as well as the effective range of drilling operation parameters such as drilling fluid parameters, bit pressure, and rotation speed. The acceleration knowledge base can be presented in text form, which is the target document. The question-and-answer recommendation module can provide recommended parameter ranges and different drilling tips based on formation changes. The acceleration knowledge base includes: (1) Formation, drill string combination, and drilling parameters: Based on different formation characteristics, such as shale layers, sandstone layers, etc., and drill string combinations, the bit pressure and rotation speed values ​​used are recorded. (2) Acceleration case library: Collect acceleration operation cases with the shortest pure drilling time from historical drilling data, including optimal drilling speed and corresponding drilling parameter cases in the optimization process under different formation and drill string conditions.

[0102] The search and recommendation layer can query the most relevant cases and experiences from the drilling speed-up knowledge base based on real-time data and formation conditions of the current drilling operation, and provide parameter ranges and precautions for drilling operations. For example, input: formation characteristics and drill string type of the current well section. Output: recommended safe parameter ranges for drilling operations and precautions for different formations (rapid drilling in mudstone formations, easy drill bit wear in sandstone formations, etc.).

[0103] Upon completion of the target well operation, the summarized drilling text knowledge needs to be updated in the database to continuously expand the locally built drilling knowledge base, laying a solid foundation for drilling parameter optimization. Specifically, methods for acquiring drilling text knowledge include expert experience based on field drilling engineers and text obtained through real-time data analysis, as shown below:

[0104] "-At a well depth of around 4309m, ROP showed a significant decrease (average decrease of 21.1%), from 4.00m / h to 3.15m / h. -Wob: 8.4 tone, Rpm: 41 rpm, riser pressure: 11.53MPa, torque: 28.26kN·m."

[0105] This analysis primarily focuses on knowledge extraction for cases of abnormally low drilling speeds. It calculates the rate of change of drilling speed over a fixed time period using a sliding window. If the rate of change is less than a set threshold, the logging parameters for that segment are converted into natural language text and stored in a locally constructed knowledge base.

[0106] In the embodiments of this application, a multi-well linkage drilling parameter multi-level optimization decision system is provided. For example... Figure 5 As shown, the system mainly consists of three levels. Level L0 is used for multi-well data collection and management. Level L1 divides the well into layers based on the specific geological conditions of the well to be drilled, and integrates expert experience to form a drilling acceleration knowledge base. Using retrieval enhancement technology and a large model, it retrieves safe parameter ranges and precautions for drilling operations that meet the current conditions from the local drilling acceleration knowledge base, and then generates personalized drilling acceleration plans through reasoning. Level L2 can construct a mapping relationship between mechanical drilling rate and various drilling parameters through data-driven models, establishing a predictive model of mechanical drilling rate embedded with bottom-layer information. Furthermore, based on knowledge of drilling rock-breaking mechanisms, it calculates multiple rock-breaking parameters such as mechanical drilling rate, drill bit wear index, and ratio factor, constructs constraint boundaries for complex downhole conditions, and uses multi-objective optimization algorithms to achieve real-time optimization of drilling parameters. Based on levels L1 and L2, a collaborative working mechanism of first searching, then optimizing, and then providing feedback is formed. Combining historical drilling data, expert experience, and formation properties, it effectively achieves scientific recommendation and dynamic decision-making of drilling parameters. The aforementioned scheme aims to rapidly adapt to various complex formation conditions encountered during drilling, fully utilize data from adjacent drilled wells within the block to guide current drilling, and significantly improve oil and gas drilling efficiency. Furthermore, it can extract and represent knowledge from historical drilling data within a specific block, and based on a multi-level recommendation framework, achieves first-level recommended parameter ranges and second-level real-time drilling optimization, providing scientific and efficient drilling optimization solutions for the field.

[0107] Through the above technical solution, when drilling operations are completed, the system can feed back operational experience to the speed-up knowledge base in real time based on the relevant drilling operation descriptions in the drilling log, for optimization and adjustment of subsequent operations. The knowledge base is continuously enriched through historical data and expert experience, making the recommendation results increasingly accurate. The multi-well linkage drilling parameter multi-level optimization decision-making system can also include multiple data processing modules. Specifically, the data governance module receives drilling engineering parameters acquired during drilling, extracts key information through outlier removal, regression interpolation, and sliding filtering, and filters out downhole anomalies and noise data to improve the density and quality of data samples. The drilling parameter search and recommendation module, based on the local speed-up knowledge base, uses open-source large models and retrieval enhancement generation technology to recommend drilling parameters under specific formation and drilling tool conditions. The real-time drilling parameter optimization module, based on a mechanical drilling rate prediction model embedded with prior formation knowledge, updates the model in real time using drilling data from specific blocks to achieve optimal model performance. Based on a multi-objective optimization function, the system intelligently optimizes decision variables such as drilling pressure and rotational speed while avoiding complex downhole conditions, thereby comprehensively improving drilling efficiency. The feedback update module is triggered when the target well completes its work. It evaluates the drilling efficiency of the well based on on-site expert experience and data analysis, and stores the resulting formatted text in a local knowledge base as historical prior information for the next well to be drilled.

[0108] The aforementioned scheme significantly improves drilling efficiency and enhances the rationality and scientific validity of the optimization results. It not only fully utilizes historical information from multiple drilling blocks but also achieves in-depth data mining and knowledge extraction. Through a local acceleration knowledge base and large-scale model reasoning techniques, it performs search-based parameter recommendations, avoiding the failure of data-driven models in certain complex formations, thus ensuring a reasonable parameter range in the optimization of drilling parameters for the target well. Furthermore, the system has continuous learning capabilities, updating its knowledge base by constantly learning new data and expert experience, forming a dynamic optimization closed loop. In this way, it provides a new optimization approach for drilling acceleration, greatly improving data utilization and drilling efficiency, and has high practical value and broad application prospects.

[0109] Figure 1 This is a flowchart illustrating a multi-level optimization decision-making method for drilling parameters involving multiple wells in one embodiment. It should be understood that, although... Figure 1 The steps in the flowchart are shown sequentially as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order in which these steps are executed, and they can be performed in other orders. Figure 1At least some of the steps in the process may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least some of the sub-steps or stages of other steps.

[0110] Figure 6 The diagram schematically illustrates a multi-level optimization decision-making system for drilling parameters in multi-well linkage according to an embodiment of this application. Figure 6 As shown in the embodiments of this application, a multi-well linkage drilling parameter multi-level optimization decision system is provided, which may include:

[0111] The memory is configured to store instructions; the processor is configured to retrieve instructions from the memory and, when executing instructions, to implement the aforementioned multi-well linkage drilling parameter multi-level optimization decision-making method.

[0112] Specifically, in this embodiment of the application, the processor can be configured as follows:

[0113] Acquire real-time drilling data during the drilling operation, including drilling operation parameters and drill bit usage data;

[0114] The first rock-breaking parameters are determined based on real-time drilling data;

[0115] When the first rock breaking parameter is outside the preset range, real-time drilling data is input into the preset model so that the preset model can output the predicted value of mechanical drilling rate. The preset model is a model trained based on historical real-time drilling data.

[0116] Multiple candidate parameter groups are generated based on the predicted mechanical drilling rate and drilling constraints. Each candidate parameter group includes multiple different types of secondary rock breaking parameters.

[0117] The function value corresponding to each candidate parameter group is determined based on the objective function. The function value is the weighted sum of all the second rock-breaking parameters included in each candidate parameter group. The objective function is solved by minimizing all normalized second rock-breaking parameters and minimizing the function value.

[0118] Determine the target parameter group from multiple candidate parameter groups based on the function value of each candidate parameter group;

[0119] The target drilling parameters are determined based on the target parameter set, and the target drilling parameters are used as the current drilling parameters for drilling operations.

[0120] In the embodiments of this application, the preset model includes a first graph learning module and a second graph learning module. Inputting real-time drilling data into the preset model to output a predicted value of the mechanical drilling rate includes: acquiring historical real-time drilling data, which includes historical drilling operating parameters, formation stratification data, and historical drill bit usage data; and based on the first graph learning module, performing dynamic clustering processing on the historical drilling operating parameters and historical drill bit usage data according to the formation stratification data to generate a formation information pre-training map, wherein the formation information pre-training map uses each type of formation as a node. Furthermore, each node is represented and embedded based on historical drilling operation parameters and historical drill bit usage data; based on the second graph learning module, the drilling operation parameters and drill bit usage data are serialized and reconstructed to generate a drilling parameter association knowledge graph; the drilling parameter association knowledge graph is extracted for temporal patterns through a temporal convolutional network to obtain temporal features; the temporal features and the adjacency matrix of the pre-trained formation information graph are fused and convolved through a graph encoding module to obtain high-dimensional hidden features; the high-dimensional hidden features are decoded and mapped to obtain the predicted value of the mechanical drilling rate output by the preset model.

[0121] In the embodiments of this application, the generation of a formation information pre-training map based on the first graph learning module's dynamic clustering of historical drilling parameters and drill bit usage data according to formation stratification data includes: performing dynamic clustering of historical drilling parameters and drill bit usage data according to formation stratification data based on the first graph learning module to determine the historical drilling parameters and drill bit usage data corresponding to each type of formation; representing each type of formation as a node in the formation information pre-training map, and embedding the historical drilling parameters and drill bit usage data corresponding to each type of formation into the corresponding node to generate the formation information pre-training map.

[0122] In the embodiments of this application, the serialization reconstruction of drilling working parameters and drill bit usage data based on the second graph learning module to generate a drilling parameter association knowledge graph includes: serializing and reconstructing drilling working parameters and drill bit usage data based on the second graph learning module to obtain a first sequence of data corresponding to the drilling working parameters and a second sequence of data corresponding to the drill bit usage data; and generating a drilling parameter association knowledge graph based on the first sequence of data and the second sequence of data.

[0123] In the embodiments of this application, the high-dimensional hidden features are determined according to the following formula (1):

[0124] (1)

[0125] in, This refers to high-dimensional hidden features. as well as middle, Both refer to the first and second sequence data, and A refers to the adjacency matrix of the stratigraphic information pre-trained map. It refers to the transpose of the adjacency matrix. This refers to the weight matrix of the learning module in the first graph. This refers to the weight matrix of the second graph learning module. This refers to the activation function of the learning module in the first image. This refers to the activation function of the learning module in the second graph.

[0126] In the embodiments of this application, determining the function value corresponding to each candidate parameter group based on the objective function includes: normalizing each second rock-breaking parameter included in each candidate parameter group to obtain the processed second rock-breaking parameter; constructing an objective function based on all the processed second rock-breaking parameters corresponding to each candidate parameter group and the weight value corresponding to each second rock-breaking parameter to determine the corresponding function value.

[0127] In the embodiments of this application, the functional expression of the objective function is shown in the following formula (2):

[0128] (2)

[0129] Where J refers to the objective function value, ROP refers to the mechanical drilling rate, DRIMP refers to the drill bit wear evaluation index, which is determined based on the ratio of drilling pressure to drill bit feed rate, MSE refers to mechanical specific energy, and RF refers to the ratio factor, which is determined based on the ratio of mechanical specific energy to the drill bit wear index. , These refer to the weight and normalization coefficient corresponding to the mechanical drilling rate, respectively. 、 This refers to the weights and normalization coefficients corresponding to the drill bit wear evaluation indicators. 、 This refers to the weight and normalization coefficient corresponding to the mechanical specific energy. 、 This refers to the weights and normalization coefficients corresponding to the ratio factor.

[0130] In embodiments of this application, the method further includes: acquiring historical real-time drilling data, which includes historical drilling operating parameters, formation layering data, and historical drill bit usage data; constructing a target document based on the historical real-time drilling data; dividing the target document into blocks to obtain multiple document fragments; determining the fragment vector corresponding to each document fragment; storing all fragment vectors in a vector database; searching the vector database for the target fragment vector corresponding to the real-time drilling data; and inputting the document fragment corresponding to the target fragment vector and the real-time drilling data into a large language model to output drilling constraints related to the current drilling operation through the large language model, wherein the drilling constraints include a range of safety parameters corresponding to the drilling operating parameters.

[0131] This application also provides a machine-readable storage medium storing instructions that cause a machine to execute the above-described multi-well linkage drilling parameter multi-level optimization decision-making method.

[0132] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 7 As shown. The computer device includes a processor A01, a network interface A02, a memory (not shown), and a database (not shown) connected via a system bus. The processor A01 provides computational and control capabilities. The memory includes internal memory A03 and a non-volatile storage medium A04. The non-volatile storage medium A04 stores an operating system B01, a computer program B02, and a database (not shown). The internal memory A03 provides an environment for the operation of the operating system B01 and the computer program B02 stored in the non-volatile storage medium A04. The database stores multi-level optimization decision data for drilling parameters in multi-well linkage. The network interface A02 communicates with external terminals via a network connection. When the processor A01 executes the computer program B02, it implements a multi-level optimization decision method for drilling parameters in multi-well linkage.

[0133] Those skilled in the art will understand that Figure 7 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0134] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0135] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0136] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0137] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0138] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0139] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0140] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0141] It should also be noted that 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 process, method, article, or apparatus. Unless otherwise specified, 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 that element.

[0142] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A multi-level optimization decision-making method for drilling parameters in multi-well linkage, characterized in that, The method includes: Acquire real-time drilling data during the drilling operation, including drilling operation parameters and drill bit usage data; The first rock-breaking parameters are determined based on the real-time drilling data; When the first rock-breaking parameter is within a preset range, the real-time drilling data is input into a preset model to output a predicted value of the mechanical drilling rate through the preset model. The preset model is a model trained based on historical real-time drilling data. The preset model includes a first graph learning module and a second graph learning module. Multiple candidate parameter groups are generated based on the predicted mechanical drilling rate and drilling constraints. Each candidate parameter group includes multiple different types of second rock breaking parameters. The function value corresponding to each candidate parameter group is determined based on the objective function, wherein the function value is the weighted sum of all the second rock-breaking parameters included in each candidate parameter group, and the objective function is solved by minimizing all normalized second rock-breaking parameters and minimizing the function value; The target parameter group among the plurality of candidate parameter groups is determined based on the function value of each candidate parameter group; Target drilling parameters are determined based on the target parameter set, and these target drilling parameters are used as the current drilling parameters for drilling operations. The step of inputting the real-time drilling data into a preset model to output a predicted value of the mechanical drilling rate through the preset model includes: The historical real-time drilling data is acquired, including historical drilling operating parameters, formation stratification data, and historical drill bit usage data. Based on the first graph learning module, the historical drilling operation parameters and historical drill bit usage data are dynamically clustered according to the formation layering data to generate a formation information pre-training graph. The formation information pre-training graph uses each type of formation as a node, and each node is embedded with a representation according to the historical drilling operation parameters and historical drill bit usage data. Based on the second graph learning module, the drilling working parameters and the drill bit usage data are serialized and reconstructed to generate a knowledge graph related to drilling parameters; Temporal patterns are extracted from the drilling parameter association knowledge graph using a temporal convolutional network to obtain temporal features; The adjacency matrix of the temporal features and the pre-trained stratigraphic information graph is fused and convolved by the graph coding module to obtain high-dimensional hidden features; The high-dimensional hidden features are decoded and mapped to obtain the predicted value of the mechanical drilling rate output by the preset model.

2. The multi-level optimization decision-making method for drilling parameters in multi-well linkage according to claim 1, characterized in that, The step of generating the pre-trained formation information map by dynamically clustering the historical drilling parameters and drill bit usage data based on the formation layering data using the first map learning module includes: Based on the first graph learning module, the historical drilling operation parameters and the drill bit usage data are dynamically clustered according to the formation layering data to determine the historical drilling operation parameters and drill bit usage data corresponding to each type of formation; Each type of formation is represented as a node in the formation information pre-training map, and the historical drilling operation parameters and drill bit usage data corresponding to each type of formation are embedded into the corresponding node to generate the formation information pre-training map.

3. The multi-level optimization decision-making method for drilling parameters in multi-well linkage according to claim 1, characterized in that, The step of sequentially reconstructing the drilling operating parameters and the drill bit usage data based on the second graph learning module to generate a drilling parameter association knowledge graph includes: Based on the second graph learning module, the drilling working parameters and the drill bit usage data are serialized and reconstructed to obtain the first sequence data corresponding to the drilling working parameters and the second sequence data corresponding to the drill bit usage data. A knowledge graph relating drilling parameters is generated based on the first sequence data and the second sequence data.

4. The multi-level optimization decision-making method for drilling parameters in multi-well linkage according to claim 1, characterized in that, The high-dimensional hidden features are determined according to the following formula (1): （1） in, This refers to the high-dimensional hidden features. as well as middle, Both refer to the first sequence data and the second sequence data, and A refers to the adjacency matrix of the pre-trained stratigraphic information map. This refers to the transpose of the adjacency matrix. This refers to the weight matrix of the first graph learning module. This refers to the weight matrix of the second graph learning module. This refers to the activation function of the first image learning module. This refers to the activation function of the learning module in the second graph.

5. The multi-level optimization decision-making method for drilling parameters in multi-well linkage according to claim 1, characterized in that, The process of determining the function value corresponding to each candidate parameter group based on the objective function includes: Normalize each second rock-breaking parameter included in each candidate parameter group to obtain the processed second rock-breaking parameter; The objective function is constructed based on all the processed second rock-breaking parameters corresponding to each candidate parameter group and the weight value corresponding to each second rock-breaking parameter, so as to determine the corresponding function value.

6. The multi-level optimization decision-making method for drilling parameters in multi-well linkage according to claim 5, characterized in that, The function expression of the objective function is shown in the following formula (2): （2） Wherein, J refers to the function value of the objective function, ROP refers to the mechanical drilling rate, DRIMP refers to the drill bit wear evaluation index, which is determined based on the ratio of drilling pressure to drill bit feed rate, MSE refers to mechanical specific energy, and RF refers to the ratio factor, which is determined based on the ratio of mechanical specific energy and the drill bit wear index. , These refer to the weight and normalization coefficient corresponding to the mechanical drilling rate, respectively. 、 This refers to the weights and normalization coefficients corresponding to the drill bit wear evaluation indicators. 、 This refers to the weight and normalization coefficient corresponding to the mechanical specific energy. 、 This refers to the weights and normalization coefficients corresponding to the ratio factor.

7. The multi-level optimization decision-making method for drilling parameters in multi-well linkage according to claim 1, characterized in that, The method further includes: The historical drilling data is obtained, including historical drilling parameters, formation stratification data, and historical drill bit usage data; Based on the historical and real-time drilling data, target documents related to drilling conditions are constructed according to formation intervals, drilling parameter variation characteristics, and drill bit usage status. The target document is segmented into semantic blocks based on working conditions, resulting in multiple document fragments. Each document fragment is vectorized and encoded to determine the fragment vector corresponding to each document fragment. All fragmented vectors are stored in a vector database to form a knowledge vector space for drilling conditions; Based on real-time drilling data, target fragment vectors related to the semantics of the current drilling condition are retrieved from the vector database. The document fragments corresponding to the target fragment vector and the real-time drilling data are input into the large language model to perform drilling condition constraint reasoning through the large language model, and output drilling constraint conditions that match the current drilling operation. The drilling constraint conditions include the safety parameter range of drilling operation parameters dynamically generated based on the real-time drilling data.

8. The multi-level optimization decision-making method for drilling parameters in multi-well linkage according to claim 1, characterized in that, The drilling parameters include the drill bit's pressure on the drill bit, rotational speed, and displacement. The drilling constraints include a first parameter range for the pressure on the drill bit, a second parameter range for the rotational speed, and a third parameter range for the displacement.

9. A multi-well linkage drilling parameter multi-level optimization decision system, characterized in that, include: The memory is configured to store instructions; The processor is configured to retrieve the instructions from the memory and, when executing the instructions, to implement the multi-well linkage drilling parameter multi-level optimization decision-making method according to any one of claims 1 to 8.

Images