Model generation method, recommendation method and related device

Abstract

The invention provides a model generation method, a recommendation method and a related device. The method comprises the following steps: constructing a single upright post connecting operation process training data set; performing data treatment on the training data set in the operation process of the single upright post to obtain a treated data set; based on feature engineering, performing data feature extraction on the treated data set to obtain a single upright post connecting operation process target training data set, and determining relevance features of the extracted features; and based on the relevance features and the target training data set in the operation process of the single upright post, training to obtain an intelligent recommendation model of the full-time nodes of the operation process of the single upright post and the working parameters of the single upright post. The method is used for improving the accuracy and reliability of the intelligent recommendation model of the working procedure full-time nodes of the single connecting stand column and the working parameters of the single connecting stand column, so that the intelligent recommendation model can keep efficient and stable performance in different application scenes, and a solid foundation is provided for scientific and accurate drilling operation; and the drilling operation efficiency is further improved.

Description

Technical Field

[0001] This application relates to the field of oil and gas development drilling technology, and in particular to a model generation method, recommendation method and related apparatus. Background Technology

[0002] The single-pipe connection process is a routine and important step in drilling operations, involving connecting a single drill pipe to the drill string to extend its length and continue drilling. This step directly affects drilling efficiency and safety. Each step in the single-pipe connection process requires a high degree of coordination and precise operation. How to accurately determine the timing and procedure for the single-pipe connection process has become a hot research topic in the drilling field.

[0003] In related technologies, technicians collect and analyze drilling site data to determine the timing and operational procedures for installing a single drilling string. This process is highly subjective and prone to errors, especially when entering new areas or when the wellbore conditions are complex. The accuracy and reliability of the judgment results are low, which limits the efficiency of drilling operations. Summary of the Invention

[0004] This application provides a model generation method, a recommendation method, and related apparatus to improve the accuracy and reliability of the intelligent recommendation model for the entire time node and working parameters of the single-column drilling operation. This enables the intelligent recommendation model to maintain high efficiency and stable performance in different application scenarios, providing a solid foundation for scientific and precise drilling operations, thereby improving drilling efficiency.

[0005] Firstly, this application provides a model generation method, including:

[0006] Construct a training dataset for the single-column erection operation process;

[0007] Data governance was performed on the training dataset of the single column operation process to obtain the governed dataset.

[0008] Based on feature engineering, data features are extracted from the post-treatment dataset to obtain the target training dataset for the single column connection operation process, and the correlation features of the extracted features are determined.

[0009] Based on the correlation features and the target training dataset of the single column erection operation process, an intelligent recommendation model is trained to obtain the full time nodes of the single column erection process and the working parameters of the single column erection.

[0010] In one possible implementation, based on the correlation features and the target training dataset of the single-column erection operation process, an intelligent recommendation model is trained to obtain the full time nodes of the single-column erection process and the working parameters of the single-column erection, including:

[0011] Based on the correlation characteristics, a clustering algorithm is determined that meets the evaluation criteria for the single column assembly process.

[0012] Based on the target training dataset of the single column erection operation process, the clustering algorithm is continuously modified through iterative fitting and optimization of algorithm exponential parameters to form the first algorithm model;

[0013] Based on the correlation characteristics, a time-based optimization algorithm is determined to satisfy the requirements of the entire process of receiving a single column.

[0014] Based on the target training dataset of the single column assembly operation process and the optimization algorithm for the entire time node of the single column assembly process, the first algorithm model is continuously corrected to form the second algorithm model through iterative fitting and optimization of algorithm exponential parameters.

[0015] Based on the correlation characteristics, an optimization algorithm is determined to meet the working parameters of a single column.

[0016] Based on the target training dataset of the single column erection operation process and the single column erection working parameter optimization algorithm, the second algorithm model is continuously corrected through iterative fitting and algorithm exponential parameter optimization to form an intelligent recommendation model for the entire time node of the single column erection process and the single column erection working parameters.

[0017] In one possible implementation, the correlation features include feature indices and controlling factor indices; based on feature engineering, data features are extracted from the post-treatment dataset to obtain the target training dataset for the single-column support operation process, and the correlation features of the extracted features are determined, including:

[0018] For the drilling parameters corresponding to the operation steps of raising the drill string, stopping the rotary table, setting the slips, stopping the pump, uncoupling, setting the slips, removing the slips, starting the pump, starting the rotary table, and lowering the drill string, skewness, kurtosis, and value range are extracted to find the correlation and association characteristics between different drilling parameters, thereby obtaining the target training dataset for the single-segment connection operation process.

[0019] Based on the extracted results, feature indices and controlling factor indices are determined. Feature indices are used to quantify the characteristics of the correlation between drilling parameters in the target training dataset of the single-column connection operation process, while controlling factor indices are used to quantify the degree of correlation between drilling parameters related to the single-column connection operation process.

[0020] In one possible implementation, a training dataset for the single-column support operation process is constructed, including:

[0021] Acquire historical drilling data, which includes drilling data for different lithologies, pressure systems, well types, drilling techniques, bit sizes, caliper enlargement, wellbore trajectory control operations, bottom hole assembly (BHA), wellbore cleaning, operating conditions, and drilling parameters.

[0022] By calibrating historical drilling data, a training dataset for the single-column connection operation process is obtained.

[0023] In one possible implementation, the model generation method further includes: selecting clustering algorithms that meet the evaluation criteria for the single-column erection process from a machine learning database containing partitioning algorithms, hierarchical algorithms, density algorithms, graph-based clustering algorithms, grid algorithms, and model algorithms, based on feature indices and controlling factor indices; and / or selecting optimization algorithms for the single-column erection process across all time nodes from a machine learning database containing K-nearest neighbors, Bayesian algorithms, logistic regression, decision trees, vector machines, and neural networks, based on feature indices and controlling factor indices; and / or selecting optimization algorithms for the working parameters of the single-column erection process from a machine learning database containing Stochastic Gradient Descent (SGD), Adaptive Gradient Algorithm (Adagrad), Adaptive Moment Estimation (Adam), and Root Mean Square Propagation (RMSProp), based on feature indices and controlling factor indices.

[0024] Secondly, this application provides a recommended method, including:

[0025] Acquire real-time drilling data and basic data for single-spindle connection operations. Real-time drilling data includes actual drilled wellbore data and single-spindle connection operating parameters. Basic data for single-spindle connection operations includes actual drilling engineering technical data for the same basin, structure, zone, and well type.

[0026] Real-time drilling data and basic data of single-column connection operations are input into a pre-trained intelligent recommendation model for the full time nodes and working parameters of the single-column connection process, thereby obtaining the full time nodes and working parameters of the single-column connection process corresponding to the real-time drilling data. The intelligent recommendation model is generated by the model generation method described in any of the first aspects.

[0027] Thirdly, this application provides a model generation apparatus, comprising:

[0028] The building module is used to construct the training dataset for the single column connection operation process;

[0029] The governance module is used to process the training dataset of the single column operation process to obtain the governed dataset.

[0030] The determination module is used to extract data features from the post-treatment dataset based on feature engineering, obtain the target training dataset for the single column connection operation process, and determine the correlation features of the extracted features.

[0031] The training module is used to train an intelligent recommendation model for the entire time nodes of the single column assembly process and the working parameters of the single column assembly operation based on the correlation features and the target training dataset of the single column assembly operation process.

[0032] In one possible implementation, the training module is specifically used for: determining a clustering algorithm that satisfies the evaluation criteria for the single-column erection process based on correlation characteristics; continuously refining the clustering algorithm to form a first algorithm model based on the single-column erection process target training dataset through iterative fitting and algorithm exponent parameter optimization; determining an optimization algorithm that satisfies the full-time node requirements of the single-column erection process based on correlation characteristics; continuously refining the first algorithm model to form a second algorithm model based on the single-column erection process target training dataset and the single-column erection process full-time node optimization algorithm through iterative fitting and algorithm exponent parameter optimization; determining an optimization algorithm that satisfies the single-column erection working parameters based on correlation characteristics; and continuously refining the second algorithm model to form an intelligent recommendation model for the full-time node requirements and working parameters of the single-column erection process based on the single-column erection process target training dataset and the single-column erection working parameter optimization algorithm through iterative fitting and algorithm exponent parameter optimization.

[0033] In one possible implementation, the correlation features include feature indices and controlling factor indices. The determination module is specifically used to: extract skewness, kurtosis, and value ranges for drilling parameters corresponding to the operation steps of raising the drill string, stopping the rotary table, setting the slips, stopping the pump, unhooking, setting the slips, removing the slips, starting the pump, starting the rotary table, and lowering the drill string, in order to find the correlation and correlation characteristics between different drilling parameters, thereby obtaining the target training dataset for the single-stand connection operation process; and determine the feature indices and controlling factor indices based on the extracted results. The feature indices are used to quantify the characteristics of the correlation between drilling parameters in the target training dataset for the single-stand connection operation process, and the controlling factor indices are used to quantify the degree of correlation of drilling parameters related to the single-stand connection procedure.

[0034] In one possible implementation, the construction module is specifically used to: acquire historical drilling data, which includes drilling data of different lithologies, pressure systems, well types, drilling processes, drill bit sizes, wellbore enlargement, wellbore trajectory control operations, bottom hole assembly, wellbore cleaning, operating conditions, and drilling parameters; and calibrate the historical drilling data to obtain a training dataset for the single-column connection operation process.

[0035] In one possible implementation, the model generation device further includes a screening module, which is configured to: screen clustering algorithms that meet the evaluation criteria for the single-column erection process from a machine learning database containing partitioning algorithms, hierarchical algorithms, density algorithms, graph-based clustering algorithms, grid algorithms, and model algorithms, based on feature indices and controlling factor indices; and / or screen optimization algorithms that meet the full-time-node optimization criteria for the single-column erection process from a machine learning database containing K-nearest neighbors, Bayesian algorithms, logistic regression, decision trees, vector machines, and neural networks, based on feature indices and controlling factor indices; and / or screen optimization algorithms that meet the working parameter optimization criteria for the single-column erection process from a machine learning database containing SGD, Adagrad, Adam, and RMSProp, based on feature indices and the controlling factor indices.

[0036] Fourthly, this application provides a recommended device, comprising:

[0037] The acquisition module is used to acquire real-time drilling data and basic data for single-spindle connection operations. Real-time drilling data includes actual drilled wellbore data and single-spindle connection operating parameters. Basic data for single-spindle connection operations includes actual drilling engineering technical data for the same basin, structure, zone, well type, and machine type.

[0038] The processing module is used to input real-time drilling data and basic data of single-column connection operations into a pre-trained intelligent recommendation model of the full time nodes and working parameters of the single-column connection process, so as to obtain the full time nodes and working parameters of the single-column connection process corresponding to the real-time drilling data. The intelligent recommendation model is generated by the model generation method described in any one of the first aspects.

[0039] Fifthly, this application provides an electronic device, including: a memory and a processor;

[0040] The memory stores computer-executed instructions;

[0041] The processor executes computer execution instructions stored in the memory, causing the processor to perform the first aspect and / or various possible implementations of the first aspect, and / or to implement the second aspect and / or various possible implementations of the second aspect.

[0042] In a sixth aspect, this application provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the first aspect and / or various possible implementations of the first aspect, and / or to implement the second aspect and / or various possible implementations of the second aspect.

[0043] In a seventh aspect, this application provides a computer program product, including a computer program that, when executed by a processor, implements the first aspect and / or various possible implementations of the first aspect, and / or implements the second aspect and / or various possible implementations of the second aspect.

[0044] The model generation method, recommendation method, and related apparatus provided in this application construct a training dataset for the single-column drilling operation process, perform data governance and data feature extraction to ensure data quality and consistency, thereby improving the model training effect. Based on the correlation features of the extracted features and the target training dataset for the single-column drilling operation process, an intelligent recommendation model for the entire time nodes of the single-column drilling process and the working parameters of the single-column drilling operation is trained, improving the accuracy and reliability of the intelligent recommendation model, providing a solid foundation for scientific and precise drilling operations, and thus improving drilling operation efficiency. Attached Figure Description

[0045] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0046] Figure 1 A schematic diagram of a scenario for the model generation method provided in this application;

[0047] Figure 2 A flowchart illustrating the model generation method provided in this application;

[0048] Figure 3 A flowchart illustrating the recommended method provided for this application;

[0049] Figure 4 Flowchart for constructing the intelligent recommendation model for the process and parameters of connecting a single oil drilling column in this application;

[0050] Figure 5 A schematic diagram of the model generation device provided in this application;

[0051] Figure 6 A schematic diagram of the recommended device provided in this application;

[0052] Figure 7 A schematic diagram of the structure of the electronic device provided in this application.

[0053] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0054] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0055] In related technologies, the collection and analysis of drilling site data by technicians relies heavily on their experience, leading to strong subjectivity and significant errors. This can result in large differences in the efficiency of single-pipe connection during shifts among technicians of different skill levels. Furthermore, when entering new areas, technicians need to adapt to the different lithological properties of formations. The trial-and-error process during single-pipe connection connection can easily lead to complex wellbore conditions due to incomplete gas-solid phase cleaning. Human decision-making carries even greater risks and errors, ultimately hindering the efficiency of single-pipe connection operations and creating a technical problem of limited efficiency. For example, if tens of thousands of wells are completed annually, with drilling and completion costs exceeding 100 billion RMB, and single-pipe / connection procedures are integrated throughout the entire drilling process, typically accounting for 3.5% to 5% of drilling time, and a joint-stock company spends approximately 70,000 days annually on single-pipe connection operations, current technology would severely impact the efficiency of single-pipe connection operations.

[0056] With the acceleration of digitalization in oil drilling, the entire process of single-pipe connection can be automatically depicted. Through reasoning and algorithms based on wellbore-surface related elements, the operation and parameters of single-pipe connection can be continuously optimized, guiding the driller to complete the single-pipe connection operation in the most scientific and reasonable way, which helps to improve drilling efficiency. Therefore, in view of the technical problems existing in the above-mentioned prior art, the model generation method, recommendation method and related device provided in this application utilize the single-pipe connection operation process target training dataset after data governance and data feature extraction, and the correlation features based on the extracted features to jointly train an intelligent recommendation model of the entire time node of the single-pipe connection process and the working parameters of the single-pipe connection. The obtained intelligent recommendation model has high accuracy and reliability, providing a solid foundation for scientific and precise drilling operations, thereby improving drilling efficiency.

[0057] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0058] Figure 1 This is a schematic diagram illustrating an application scenario provided in one embodiment of this application, such as... Figure 1As shown, the specific application scenario of this application may include computing device 11, server 12, and terminal device 13. The user operates on computing device 11 to generate an intelligent recommendation model of the entire time node of the single column connection process and the working parameters of the single column connection. In response to the user's request, computing device 11 obtains the target training dataset and correlation features of the single column connection operation process from server 12. After multiple iterations and optimizations of the model, the intelligent recommendation model that meets the user's requirements is finally generated on computing device 11. For example, historical drilling data can be pulled through drilling data monitoring software installed on terminal device 13 and sent to server 12 for storage. Based on the idea of ​​supervised machine learning, the historical drilling data is calibrated on server 12 to increase the diversity of data and facilitate the model to learn useful features and patterns from the data. Further data governance and feature engineering are performed on the calibrated data to extract data features to generate the target training dataset and correlation features of the single column connection operation process.

[0059] It should be noted that, Figure 1 The application scenarios shown are for illustrative purposes only. Any two or three of the computing device 11, server 12, and terminal device 13 can be the same device, and there is no limit to the number of each type of device. The computing device 11 and terminal device 13 can be wearable devices, mobile phones, computers, laptops, or personal digital assistants (PDAs), etc.; terminal device 12 can also be replaced by a server cluster.

[0060] It should be noted that the above application scenarios are shown only to facilitate understanding of the spirit and principles of this application, and the implementation methods of this application are not limited to these scenarios. Figure 1 The limitations of the application scenarios shown.

[0061] Figure 2 A flowchart illustrating the model generation method provided in this application. Figure 2 As shown, the model generation method includes:

[0062] S201. Construct a training dataset for the single-column support operation process.

[0063] The training dataset is the core data source for model learning. It contains a large number of labeled samples used to train the model, enabling the model to learn the inherent patterns and characteristics of the data. A high-quality training dataset is crucial for model training.

[0064] For example, multiple sets of historical drilling data are acquired, including different drilling procedures, different working conditions, different lithological systems, different well types, different drilling parameters, different drilling technologies, etc., such as log data, video and image data, manually recorded data, sensor data, and data stored in the drilling production operation digital system.

[0065] Further feature annotation (also known as calibration) is performed on the acquired historical drilling data. For example, image and video data are annotated with the start and end times of each key step, and abnormal situations are noted. Sensor data is annotated with changes in key parameters, such as pressure and location. Log data is annotated with each operation step and its corresponding timestamp. Data standardization and normalization can also be performed on historical drilling data to diversify the training dataset for single-column operation processes, making the patterns and characteristics of the data more apparent, facilitating better model learning, and improving the model's generalization ability.

[0066] S202. Perform data governance on the training dataset of the single column operation process to obtain the governed dataset.

[0067] Data governance includes cleaning tagged data, which encompasses data file format normalization, structured data entry, unified data unit system, noise reduction of dirty data, completion of missing data, and correction of bad data. Cleaning can be understood as the process of identifying and correcting errors, missing, inconsistent, or redundant data in the dataset, ensuring data integrity, reliability, and consistency. In practical applications, appropriate cleaning methods are selected based on the quality of the acquired data.

[0068] S203. Based on feature engineering, data features are extracted from the post-treatment dataset to obtain the target training dataset for the single column connection operation process, and the correlation features of the extracted features are determined.

[0069] In drilling technology, feature engineering is a crucial step in data analysis and machine learning. The purpose of feature engineering is to extract useful features from the post-drilling dataset to improve the performance and predictive ability of the model. For example, the large amount of data generated during drilling, including raw sensor data such as pressure, temperature, flow rate, and drilling pressure, may not be effective if used directly for model training and analysis. Through feature engineering, more representative and predictive features can be extracted and constructed, thereby improving the accuracy and stability of the model.

[0070] For example, features of drilling parameters such as drilling pressure, rotational speed, torque, displacement, standpipe pressure, and temperature can be extracted, including the mean, median, difference, skewness, and kurtosis of drilling pressure. Optionally, interactive features between original features can be constructed, such as products and ratios. For instance, the product of pressure and temperature can be used as a new feature to describe the combined influence of the internal and external environments of the wellbore. Combined features between original features can also be constructed, such as combining pressure, temperature, and flow rate to construct a comprehensive feature to describe the combined state of the internal and external environments of the wellbore, and so on.

[0071] Furthermore, the correlation characteristics of the extracted features are determined, for example, by calculating the correlation between data features and identifying strongly and weakly correlated features; the degree of correlation between data features is calculated to measure the dependency between two variables.

[0072] Based on correlation and correlation results, the relationships between different drilling parameters can be interpreted. For example, high correlation may indicate a strong linear relationship between two parameters, while high mutual information may indicate a complex nonlinear relationship. Correlation and correlation results can be used to optimize the drilling process, such as identifying key parameters, adjusting operational strategies, and predicting drilling performance.

[0073] S204. Based on the correlation features and the target training dataset of the single column erection operation process, an intelligent recommendation model is trained to obtain the full time nodes of the single column erection process and the working parameters of the single column erection.

[0074] In this step, an open-source algorithm model based on machine learning and deep learning technologies can be selected as the initial training model, or an algorithm model can be built according to the actual project requirements as the initial training model. This application embodiment does not impose any restrictions.

[0075] For example, by utilizing correlation features and the target training dataset of the single column erection operation process, the initial training model is continuously trained to learn the patterns and rules in the sample data. Through continuous iteration and optimization, the initial training model gradually adjusts its parameters and structure to better fit the data distribution in the target training dataset of the single column erection operation process, forming an intelligent recommendation model that meets the requirements for the full time nodes of the single column erection process and the working parameters of the single column erection.

[0076] In this embodiment, by constructing and managing a high-quality training dataset for the single-pillar drilling operation process, the accuracy and integrity of the data are ensured, thereby improving the model training effect. Based on feature engineering and the training dataset for the single-pillar drilling operation process, an intelligent recommendation model is trained, improving the accuracy and reliability of the intelligent recommendation model. This enables the intelligent recommendation model to maintain efficient and stable performance in different application scenarios, providing a solid foundation for scientific and precise drilling operations, and thus improving drilling efficiency.

[0077] Specifically, in some embodiments, based on correlation features and the target training dataset of the single-column splicing operation process, an intelligent recommendation model is trained to obtain the full time nodes of the single-column splicing process and the working parameters of the single-column splicing operation, including:

[0078] Step 1.1: Based on the correlation characteristics, determine the clustering algorithm that meets the evaluation criteria for the single column process.

[0079] Among them, the evaluation of the single column connection process can be regarded as assessing the effectiveness of the single column connection operation through a series of indicators, including the speed of single column connection, safety of single column connection, efficiency of single column connection, and quality of single column connection operation.

[0080] Clustering algorithms are unsupervised learning methods used to find similarities and patterns among data samples, and then cluster the data samples into groups or clusters based on features. Common clustering algorithms include K-means clustering, hierarchical clustering, spectral clustering, and mean shift.

[0081] For example, evaluation criteria for the single-column splicing process can be preset, such as high splicing speed, high operational safety, and high success rate. Preprocessed correlation feature data is input into a clustering algorithm, the clustering results are analyzed, the clustering effect is evaluated, and it is determined whether the clustering algorithm meets the evaluation criteria for the single-column splicing process. This allows for the selection of one or more clustering algorithms that meet the evaluation criteria from multiple algorithms. Optionally, multiple clustering algorithms can be encapsulated in a database (algorithm library), with clustering effect evaluation rules set in the program. After the computer executes the code, it can automatically output clustering algorithms that meet the evaluation criteria for the single-column splicing process. Alternatively, based on historical experience, clustering algorithms that conform to the characteristics and patterns of the single-column splicing process can be selected, and then further verified based on correlation features to finally determine one or more clustering algorithms that meet the evaluation criteria for the single-column splicing process.

[0082] Step 1.2: Based on the target training dataset of the single column erection operation process, the clustering algorithm is continuously modified through iterative fitting and optimization of algorithm exponential parameters to form the first algorithm model.

[0083] The algorithm index parameters include torque index, drilling fluid density index, wellbore pressure index, and safety event occurrence rate index.

[0084] This step is supervised machine learning. The clustering algorithm continuously learns the data features in the training dataset of the single column construction process so that it can make accurate predictions when encountering new and unseen data.

[0085] A clustering algorithm is trained using the target training dataset of the single column erection operation process to construct the first algorithm model. That is, after each iteration, the clustering effect is evaluated and the algorithm exponent parameters are optimized. For example, based on the clustering effect, the weight and range of the exponent parameters are repeatedly adjusted until the clustering effect reaches the optimal level, thus forming the first algorithm model.

[0086] Step 1.3: Based on the correlation characteristics, determine the optimization algorithm for all time nodes of the single column receiving process.

[0087] As described in step 1.1, optionally, multiple machine learning algorithms can be encapsulated in a database (algorithm library). The program can set evaluation rules for all time nodes of the single column assembly process, such as minimum risk probability, low failure rate, and high rationality of the single column assembly step. After the computer executes the code program, it can automatically generate the optimal algorithm that satisfies all time nodes of the single column assembly process. Alternatively, based on historical experience, machine learning algorithms that conform to the characteristics and patterns of the single column assembly process can be selected, and then further verified based on correlation features to finally determine one or more optimal algorithms that satisfy all time nodes of the single column assembly process.

[0088] Step 1.4: Based on the target training dataset of the single column assembly operation process and the optimization algorithm for the entire time node of the single column assembly process, the first algorithm model is continuously modified to form the second algorithm model through iterative fitting and optimization of algorithm exponential parameters.

[0089] This step involves continuing supervised machine learning based on the first algorithm model. It utilizes the training dataset of the single column installation process and the optimization algorithm for all time nodes of the single column installation process to construct the second algorithm model. That is, after each iteration, the prediction results of the first algorithm model are evaluated and the algorithm exponent parameters are optimized. For example, based on the prediction results, the weights and ranges of the exponent parameters of the first algorithm model are repeatedly adjusted until the prediction effect reaches the optimal level, thus forming the second algorithm model.

[0090] Step 1.5: Based on the correlation characteristics, determine the optimization algorithm that satisfies the working parameters of a single column.

[0091] As described in step 1.1, optionally, multiple machine learning algorithms can be encapsulated in a database (algorithm library). The program sets evaluation rules for the working parameters of single column connection, such as minimizing the total time for connecting a single column, maximizing the success rate of connecting a single column, and minimizing the operating cost. After the computer executes the code program, it can automatically generate an optimal algorithm that meets the working parameters of connecting a single column. Alternatively, based on historical experience, machine learning algorithms that conform to the characteristics and patterns of the single column connection process can be selected, and then further verified based on correlation features to finally determine one or more optimal algorithms that meet the working parameters of connecting a single column.

[0092] Considering the characteristics of the single column connection process and the fact that the single column connection process is shorter than the drilling process, the target training dataset of the single column connection process is used as the sample data based on the final industrial production value. See step 1.6 for details.

[0093] Step 1.6: Based on the target training dataset of the single column erection operation process and the single column erection working parameter optimization algorithm, the second algorithm model is continuously corrected through iterative fitting and algorithm exponential parameter optimization to form an intelligent recommendation model for the entire time node of the single column erection process and the single column erection working parameters.

[0094] This step involves continuing supervised machine learning based on the second algorithm model. Using an optimization algorithm for the working parameters of the single column connection, an intelligent recommendation model is constructed for the entire time nodes of the single column connection process and the working parameters of the single column connection. That is, after each iteration, the prediction results of the second algorithm model are evaluated and the algorithm index parameters are optimized. For example, based on the prediction results, the weights and ranges of the index parameters of the second algorithm model are repeatedly adjusted until the prediction effect reaches the optimal level, thus forming an intelligent recommendation model.

[0095] This application embodiment evaluates the performance of each algorithm model by assessing the operational indicators of single-pillar connection in real time. Based on the actual results, the model parameters are continuously optimized, ensuring the long-term stability and accuracy of the intelligent recommendation model. Trained on a large amount of data and related features, the intelligent recommendation model can predict and avoid potential operational risks, reducing safety accidents caused by human error and effectively improving the efficiency and safety of single-pillar connection operations. Furthermore, by utilizing cluster analysis and supervised machine learning, this application breaks through the conventional method of determining the timing and parameters of single-pillar connection operations based on the experience of drilling engineers and operators. This allows for more precise timing of each operation and more scientifically sound working parameters.

[0096] Furthermore, in one possible implementation, the correlation features include feature indices and controlling factor indices; based on feature engineering, data features are extracted from the post-treatment dataset to obtain the target training dataset for the single-column support operation process, and the correlation features of the extracted features are determined, including:

[0097] Step 2.1: Extract the skewness, kurtosis, and value range of the drilling parameters corresponding to the operation steps of raising the drill string, stopping the rotary table, setting the slips, stopping the pump, uncoupling, setting the slips, removing the slips, starting the pump, starting the rotary table, and lowering the drill string, in order to find the correlation and association characteristics between different drilling parameters, thereby obtaining the target training dataset for the single-segment connection operation process.

[0098] Step 2.2: Determine the feature index and the main control factor index based on the extracted results. The feature index is used to quantify the characteristics of the correlation between drilling parameters in the target training dataset of the single-column connection operation process. The main control factor index is used to quantify the degree of correlation between drilling parameters related to the single-column connection process.

[0099] For example, the correlation between the extracted data features is calculated to identify strongly and weakly correlated features; the degree of association between the extracted data features is calculated to measure the dependency between two variables.

[0100] Characteristic indices can be calculated using various methods to determine the correlation coefficients between different drilling parameters, such as correlation analysis and characteristic importance scoring. Commonly used methods include Pearson correlation coefficient, Kendall correlation coefficient, and Spearman correlation coefficient. Assuming correlation analysis is used to calculate the characteristic index of other drilling parameters relative to the bit pressure, the characteristic index ranges from 0 to 1, with a larger value indicating a greater influence of the characteristic on the bit pressure.

[0101] By identifying and quantifying the key control factors, the single-spindle connection operation can be better optimized. For example, parameters with high correlation to the single-spindle connection process can be identified from a large number of drilling parameters, such as hook load, hook height, drilling pressure, rotation speed, and torque. The correlation between the above parameters and the single-spindle connection process can be quantified. The higher the index, the higher the correlation between the parameter and the single-spindle connection process, and the greater the impact on the single-spindle connection process.

[0102] In this embodiment of the application, feature engineering can be used to extract and construct more representative and predictive features, providing high-quality input for subsequent model training, thereby improving the accuracy and stability of the model.

[0103] Based on the above embodiments, a training dataset for the single-column support operation process is constructed, including:

[0104] Step 3.1: Obtain historical drilling data. Historical drilling data includes drilling data for different lithologies, pressure systems, well types, drilling techniques, drill bit sizes, wellbore enlargement, wellbore trajectory control operations, bottom hole assembly, wellbore cleaning, operating conditions, and drilling parameters.

[0105] For example, historical drilling data can be obtained from online databases, third parties, or digital systems for oil drilling production and operation. In this embodiment of the application, no restrictions are placed on the method of acquisition.

[0106] Step 3.2: Calibrate the historical drilling data to obtain the training dataset for the single-column connection operation process.

[0107] Data calibration includes the process of assigning one or more labels to each data point in historical drilling data. For example, calibrating time-domain data includes calibrating the time period of a single wellhead, equipment status, etc.

[0108] Data calibration includes manual calibration, automatic calibration, and semi-automatic calibration. Manual calibration involves human experts or annotators labeling the data. Semi-automatic calibration combines automated tools with manual correction. Automatic calibration relies entirely on algorithms and models to label the data. This application does not impose specific limitations on the data calibration method.

[0109] In this embodiment, by calibrating historical drilling data, the quality of the training dataset for single-pipe connection operations is improved. When using this dataset to train a model, the accuracy and performance of the model can be enhanced. Furthermore, accurate labels help the model better understand and predict data, and facilitate the identification and collection of diverse data samples, thereby enabling the model to better generalize to different scenarios and environments.

[0110] In some embodiments, the model generation method further includes: selecting clustering algorithms that meet the evaluation criteria for the single-column erection process from a machine learning database containing partitioning algorithms, hierarchical algorithms, density algorithms, graph-based clustering algorithms, grid algorithms, and model algorithms, based on feature indices and controlling factor indices; and / or selecting optimization algorithms for the single-column erection process across all time nodes from a machine learning database containing K-nearest neighbors, Bayesian algorithms, logistic regression, decision trees, vector machines, and neural networks, based on feature indices and controlling factor indices; and / or selecting optimization algorithms for the working parameters of the single-column erection process from a machine learning database containing Stochastic Gradient Descent (SGD), Adaptive Gradient Algorithm (Adagrad), Adaptive Moment Estimation (Adam), and Root Mean Square Propagation (RMSProp), based on feature indices and controlling factor indices.

[0111] Based on the above embodiments, the intelligent recommendation model for the entire time nodes and working parameters of the single column connection process generated by the model generation method can be applied to the real-time push scenario of the single column connection process and parameters. The prediction results of the intelligent recommendation model can be transmitted online to the engineering and technical management personnel to guide the driller to complete the single column connection operation in the most scientific and reasonable way.

[0112] Figure 3 A flowchart illustrating the recommended method provided in this application is shown below. Figure 3 As shown, recommended methods include:

[0113] S301. Obtain real-time drilling data and basic data for single-spindle connection operations. Real-time drilling data includes actual drilled wellbore data and single-spindle connection operating parameters. Basic data for single-spindle connection operations includes actual drilling engineering technical data for the same basin, structure, zone, and well type.

[0114] The actual drilled wellbore data includes various wellbore-related data collected during the drilling process using various sensors and monitoring equipment, such as well depth, wellbore diameter, wellbore trajectory, pressure, temperature, drilling speed, drilling fluid data, and formation data. The single-stand connection parameters include various parameters involved in the single-stand connection operation, such as hook load, drilling pressure, displacement, rotational speed, torque, and standpipe pressure.

[0115] In this step, real-time drilling data can be actively acquired or passively received. For example, it can be collected from the monitoring system in real time or periodically, or sent periodically by the monitoring system, or sent when the monitoring system detects new data updates, and so on.

[0116] The basic data for single-column splicing operations can be understood as the data of single-column splicing operations carried out under similar conditions in the past. It is a historical data record and a basic database that can be used for reference and analysis of subsequent single-column splicing operations to optimize future single-column splicing plans. For example, the basic data for single-column splicing operations can be pre-set in the storage space of the recommendation device or the storage space of the monitoring system, etc.

[0117] It should be noted that the embodiments of this application do not impose specific restrictions on the acquisition format and storage location of real-time drilling data and basic data for single-pipe operation.

[0118] S302. Input the real-time drilling data and basic data of the single-column connection operation into the pre-trained intelligent recommendation model of the full time nodes and working parameters of the single-column connection process, and obtain the full time nodes and working parameters of the single-column connection process corresponding to the real-time drilling data.

[0119] The intelligent recommendation model is generated by executing any of the above model generation method embodiments. The entire timeline for connecting a single drill string includes: raising the drill string, stopping the rotary table, setting the slips, stopping the pump, uncoupling, setting the slips, removing the slips, starting the pump, starting the rotary table, and lowering the drill string. The working parameters for connecting a single drill string include drilling pressure, rotation speed, drilling rate, well cleaning cleanliness, torque, and displacement.

[0120] The pre-trained intelligent recommendation model for the entire time nodes and working parameters of the single column splicing process can accurately predict based on the input data and output accurate data for the entire time nodes and working parameters of the single column splicing process.

[0121] In this embodiment, a pre-trained intelligent recommendation model is used to analyze data and optimize the entire time node and working parameters of the single-column connection process in real time. This guides drilling engineers and operators to make adjustment decisions on the operation steps and working parameters of the single-column connection based on the optimization results, making the timing of each operation in the single-column connection process more accurate and the working parameters more scientific and reasonable, thereby improving the efficiency of drilling operations.

[0122] Figure 4 The flowchart for the intelligent recommendation model construction of the single-pole drilling procedure and parameters provided in this application is as follows: Figure 4 As shown, this embodiment, based on the above embodiment, provides a detailed description of the recommended method for drilling connection of a single support column, including the following steps:

[0123] Step 1: Expert Certification

[0124] This step includes the calibration of the label set for single-stand connection operations. This label set includes data on drilling data for different lithologies, pressure systems, well types, drilling techniques, bit sizes, enlargement operations, wellbore trajectory control operations, borehole harmonics (BHA), wellbore cleaning, drilling parameters, and operating conditions. Specifically, the calibration of the label set involves experts calibrating the time-domain data for drilling data across different lithologies, pressure systems, well types, drilling techniques, bit sizes, enlargement operations, wellbore trajectory control operations, BHA, wellbore cleaning, drilling parameters, and operating conditions to construct training label data.

[0125] Step Two: Data Governance

[0126] This step includes cleaning the training label data, also known as data governance. Specifically, this includes normalizing data file formats, structuring data for storage, standardizing data units, denoising dirty data, filling in missing data, and correcting bad data. As described in step S202, it will not be repeated here.

[0127] Step 3: Feature Engineering

[0128] This step includes feature extraction from the training label data, as described in step S203, and will not be repeated here.

[0129] Step 4: Algorithm Selection

[0130] This step includes screening of single column connection procedures and parameter recommendation algorithms, specifically screening and parameter optimization of single column connection procedure index evaluation algorithms, screening and parameter optimization of automatic optimization algorithms for all time nodes of single column connection procedures, and screening and parameter optimization of automatic optimization algorithms for single column connection working parameters.

[0131] The selection and parameter tuning of the single-column erection process evaluation algorithm includes selecting clustering algorithms that meet the single-column erection process evaluation criteria from a machine learning database containing partitioning algorithms, hierarchical algorithms, density algorithms, graph theory clustering algorithms, grid algorithms, and model algorithms, as described in steps 1.1 and 1.2, which will not be repeated here. The selection and parameter tuning of the automatic optimization algorithm for the single-column erection process across all time nodes includes selecting optimization algorithms that meet the single-column erection process across all time nodes from a machine learning database containing K-nearest neighbors, Bayesian, logistic regression, decision trees, vector machines, and neural networks, as described in steps 1.3 and 1.4, which will not be repeated here. The selection and parameter tuning of the automatic optimization algorithm for the single-column erection working parameters includes selecting optimization algorithms that meet the single-column erection working parameter optimization criteria from a machine learning database containing SGD, Adagrad, Adam, and RMSProp, as described in steps 1.5 and 1.6, which will not be repeated here.

[0132] Step 5: Recommended Processes

[0133] This step includes real-time push notifications of the single column connection process and parameters, which are the real-time push notifications of the entire time node of the single column connection process and the working parameters of the single column connection process.

[0134] For example, real-time drilling data from batch wells is integrated into an intelligent recommendation model. This data includes drilling pressure, hook load, displacement, torque, and standpipe pressure. This enables the automatic optimization of all time nodes and working parameters for the single standpipe connection process, which are then transmitted online to engineering and technical management personnel to guide the driller in completing the single standpipe connection operation in the most scientific and reasonable manner.

[0135] Next, a specific embodiment will be used to illustrate the intelligent recommendation method for drilling connection procedures and parameters for a single drill string provided in this application. The specific application steps are as follows:

[0136] Step 4.1: Utilize the digital system for oil drilling production and operation to collect actual drilling engineering technical data from the same basin, structure, zone, well type, and machine type, and compile them into a database for single-spindle operation.

[0137] Step 4.2: Utilize the petroleum engineering technology Internet of Things system to access the wellbore engineering parameters of batch well drill bits in real time, including real-time data such as drilling fluid properties, formation lithology, pressure system, and drilling parameters.

[0138] Step 4.3: Utilize the intelligent recommendation model for the entire time nodes and working parameters of the single column connection process developed in this application to realize the automatic evaluation of the indicators of the single column connection process, the automatic optimization of the entire time nodes and working parameters of the workflow, and transmit them online to the engineering and technical management personnel to guide the driller to complete the single column connection operation in the most scientific and reasonable manner.

[0139] In summary, this application has at least the following advantages:

[0140] 1. By introducing supervised machine learning, cluster analysis, and other artificial intelligence methods, this invention breaks through the conventional method of determining the entire time node and working parameters of the single-spindle connection process based on the experience of engineers or drillers. This allows for more precise timing of each operation in the single-spindle connection process and more scientifically sound working parameters. Taking drilling operations in a joint-stock company as an example, after optimizing the single-spindle connection process using this invention, the average time efficiency of single-spindle connection is improved by 20%, equivalent to directly saving the use of 50 drilling rigs annually.

[0141] 2. The batch processing of drilling and single-pole connection procedures for multiple wells can be automatically optimized in real time, which can greatly reduce the complexity of accidents caused by improper single-pole connection operations, significantly shorten the single-pole connection time, and thus greatly improve drilling operation efficiency and reduce drilling engineering costs.

[0142] The intelligent recommendation model for the entire time nodes and working parameters of the single column connection process developed in this application covers drilling and completion projects for conventional sandstone oil and gas reservoirs and carbonate oil and gas reservoirs, unconventional shale oil and gas reservoirs and tight oil and gas reservoirs, conglomerate oil reservoirs, coalbed (rock) gas, geothermal and other types of drilling rigs from 20 to 150.

[0143] Figure 5 A schematic diagram of the model generation device provided in this application. Figure 5 As shown, the model generation device 50 includes: a construction module 51, a governance module 52, a determination module 53, and a training module 54.

[0144] in:

[0145] Module 51 is used to build a training dataset for the single column connection operation process;

[0146] The governance module 52 is used to perform data governance on the training dataset of the single column operation process to obtain the governed dataset.

[0147] Module 53 is used to extract data features from the post-treatment dataset based on feature engineering, obtain the target training dataset for the single-column support operation process, and determine the correlation features of the extracted features.

[0148] Training module 54 is used to train an intelligent recommendation model for the entire time nodes of the single column assembly process and the working parameters of the single column assembly based on the correlation features and the target training dataset of the single column assembly operation process.

[0149] In one possible implementation, the training module 54 is specifically used for: determining a clustering algorithm that satisfies the evaluation criteria for the single-column erection process based on correlation characteristics; continuously refining the clustering algorithm to form a first algorithm model based on the single-column erection process target training dataset through iterative fitting and algorithm exponent parameter optimization; determining an optimization algorithm that satisfies the full-time node requirements of the single-column erection process based on correlation characteristics; continuously refining the first algorithm model to form a second algorithm model based on the single-column erection process target training dataset and the single-column erection process full-time node optimization algorithm through iterative fitting and algorithm exponent parameter optimization; determining an optimization algorithm that satisfies the single-column erection working parameters based on correlation characteristics; and continuously refining the second algorithm model to form an intelligent recommendation model for the full-time node requirements and working parameters of the single-column erection process based on the single-column erection process target training dataset and the single-column erection working parameter optimization algorithm through iterative fitting and algorithm exponent parameter optimization.

[0150] In one possible implementation, the correlation features include feature indices and controlling factor indices. The determining module 53 is specifically used to: extract skewness, kurtosis, and value ranges for drilling parameters corresponding to the operation steps of raising the drill string, stopping the rotary table, setting the slips, stopping the pump, unhooking, setting the slips, removing the slips, starting the pump, starting the rotary table, and lowering the drill string, in order to find the correlation and correlation characteristics between different drilling parameters, thereby obtaining a target training dataset for the single-segment connection operation process; and determine the feature indices and controlling factor indices based on the extracted results. The feature indices are used to quantify the characteristics of the correlation between drilling parameters in the target training dataset for the single-segment connection operation process, while the controlling factor indices are used to quantify the degree of correlation of drilling parameters related to the single-segment connection procedure.

[0151] In one possible implementation, the construction module 51 is specifically used to: acquire historical drilling data, which includes drilling data of different lithologies, pressure systems, well types, drilling processes, drill bit sizes, wellbore enlargement, wellbore trajectory control operations, bottom hole drill string assemblies, wellbore cleaning, operating conditions, and drilling parameters; and calibrate the historical drilling data to obtain a training dataset for the single-column connection operation process.

[0152] In one possible implementation, the model generation apparatus further includes a screening module 55, which is configured to: screen clustering algorithms that meet the evaluation criteria for the single-column erection process from a machine learning database containing partitioning algorithms, hierarchical algorithms, density algorithms, graph-based clustering algorithms, grid algorithms, and model algorithms, based on feature indices and controlling factor indices; and / or screen optimization algorithms that meet the full-time-node optimization criteria for the single-column erection process from a machine learning database containing K-nearest neighbors, Bayesian algorithms, logistic regression, decision trees, vector machines, and neural networks, based on feature indices and controlling factor indices; and / or screen optimization algorithms that meet the working parameter optimization criteria for the single-column erection process from a machine learning database containing SGD, Adagrad, Adam, and RMSProp, based on feature indices and the controlling factor indices.

[0153] The model generation device 50 in this embodiment can execute the technical solution shown in the above-described model generation method embodiment. Its implementation principle and beneficial effects are similar, and will not be described again here.

[0154] Figure 6 A schematic diagram of the recommended device provided in this application. (See attached diagram.) Figure 6 As shown, the recommendation device 60 includes: an acquisition module 61 and a processing module 62. Wherein:

[0155] The acquisition module 61 is used to acquire real-time drilling data and basic data for connecting single-column operations. The real-time drilling data includes actual drilled wellbore data and working parameters for connecting single-column operations. The basic data for connecting single-column operations includes actual drilling engineering technical data for the same basin, structure, zone, well type, and machine type.

[0156] The processing module 62 is used to input real-time drilling data and basic data of single-column connection operation into a pre-trained intelligent recommendation model of the full time nodes of the single-column connection process and the working parameters of the single-column connection, so as to obtain the full time nodes of the single-column connection process and the working parameters of the single-column connection corresponding to the real-time drilling data. The intelligent recommendation model is generated by the model generation method described in any one of the first aspects.

[0157] The recommending device 60 in this embodiment can execute the technical solution shown in the above-described recommending method embodiment. Its implementation principle and beneficial effects are similar, and will not be described again here.

[0158] Figure 7 A schematic diagram of the structure of the electronic device provided in this application. Figure 7 As shown, the electronic device 70 provided in this embodiment includes at least one processor 701 and a memory 702. Optionally, the electronic device 70 further includes a communication component 703. The processor 701, memory 702, and communication component 703 are connected via a bus 704.

[0159] In a specific implementation, at least one processor 701 executes computer execution instructions stored in memory 702, causing at least one processor 701 to perform the above-described method.

[0160] The specific implementation process of processor 701 can be found in the above method embodiments, and its implementation principle and technical effect are similar. It will not be repeated here.

[0161] In the above embodiments, it should be understood that the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor.

[0162] The memory may include random access memory (RAM) and may also include non-volatile memory (NVM), such as at least one disk storage device.

[0163] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of illustration, the buses shown in the accompanying drawings are not limited to a single bus or a single type of bus.

[0164] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0165] This application also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the above-described method.

[0166] The aforementioned readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The readable storage medium can be any available medium accessible to a general-purpose or special-purpose computer.

[0167] An exemplary readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can reside in an Application Specific Integrated Circuit (ASIC). Alternatively, the processor and the readable storage medium can exist as discrete components in the device.

[0168] The division of units is merely a logical functional division; in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0169] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0170] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0171] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0172] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0173] Finally, it should be noted that other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This invention is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein, and is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.

Claims

1. A model generation method characterized by comprising: The method comprises the following steps: constructing a single-string column operation process training data set; performing data governance on the single-string column operation process training data set to obtain a governed data set; based on feature engineering, performing data feature extraction on the governed data set to obtain a single-string column operation process target training data set, and determining the relevance features of the extracted features; training an intelligent recommendation model of single-string column process full time node and single-string column working parameter based on the relevance features and the single-string column operation process target training data set.

2. The model generation method according to claim 1, characterized by, The training of the intelligent recommendation model of single-string column process full time node and single-string column working parameter based on the relevance features and the single-string column operation process target training data set comprises: determining a clustering algorithm that meets the single-string column process index evaluation according to the relevance features; forming a first algorithm model by iteratively fitting and optimizing the algorithm index parameters and constantly correcting the clustering algorithm according to the single-string column operation process target training data set; determining a single-string column process full time node optimization algorithm according to the relevance features; forming a second algorithm model by iteratively fitting and optimizing the algorithm index parameters and constantly correcting the first algorithm model according to the single-string column operation process target training data set and the single-string column process full time node optimization algorithm; determining a single-string column working parameter optimization algorithm according to the relevance features; forming an intelligent recommendation model of single-string column process full time node and single-string column working parameter by iteratively fitting and optimizing the algorithm index parameters and constantly correcting the second algorithm model according to the single-string column operation process target training data set and the single-string column working parameter optimization algorithm.

3. The model generation method according to claim 1 or 2, characterized by, The relevance features include feature indexes and master factor indexes; the data feature extraction on the governed data set to obtain a single-string column operation process target training data set and determine the relevance features of the extracted features based on feature engineering comprises: extracting the skewness, kurtosis and value range of the drilling parameters corresponding to the operation steps of lifting the drilling tool, stopping the rotary table, seating the slip, stopping the pump, unhooking, hooking, taking the slip, starting the pump, starting the rotary table and lowering the drilling tool to find the correlation and relevance degree features between different drilling parameters, thereby obtaining a single-string column operation process target training data set; determining the feature indexes and master factor indexes according to the extracted results, wherein the feature indexes are used to quantify the correlation between the drilling parameters in the single-string column operation process target training data set, and the master factor indexes are used to quantify the correlation degree of the drilling parameters related to the single-string column process.

4. The model generation method according to claim 1, characterized by, The construction of the single-string column operation process training data set comprises: obtaining historical drilling data, which includes drilling data of different lithology, pressure system, well type, drilling technology, drill bit size, hole enlargement, well trajectory control operation, bottom hole assembly (BHA), hole cleaning, working condition and drilling parameter; calibrating the historical drilling data to obtain the single-string column operation process training data set.

5. The model generation method according to claim 3, wherein The model generation method further comprises: According to the feature index and the master factor index, a clustering algorithm meeting the single root column process procedure index evaluation is screened from a machine learning database comprising a division algorithm, a hierarchical algorithm, a density algorithm, a graph theory clustering algorithm, a grid algorithm and a model algorithm, and / or, based on the feature index and the master factor index, a single root column process full time node optimization algorithm meeting the single root column process procedure is screened from a machine learning database comprising K-neighbors, Bayesian, logistic regression, decision tree, vector machine and neural network, and / or, based on the feature index and the master factor index, a single root column work parameter optimization algorithm meeting the single root column work parameter optimization algorithm is screened from a machine learning database comprising a stochastic gradient descent (SGD), an adaptive gradient (Adagrad), an adaptive moment estimation (Adam) and a root mean square propagation (RMSProp).

6. A recommendation method characterized by comprising: It comprises: Real-time drilling data and single root column operation basic data are acquired, the real-time drilling data comprising real drilling wellbore data and single root column working condition parameters, and the single root column operation basic data comprising real drilling engineering technical data of the same basin, the same structure, the same zone and the same well type; The real-time drilling data and the single root column operation basic data are input into a pre-trained intelligent recommendation model of single root column process full time node and single root column work parameter to obtain single root column process full time node and single root column work parameter corresponding to the real-time drilling data, the intelligent recommendation model being generated by the model generation method in any one of claims 1 to 5.

7. A model generation apparatus characterized by comprising: It comprises: A construction module is configured to construct a single root column operation process training data set; A governance module is configured to perform data governance on the single root column operation process training data set to obtain a governed data set; A determination module is configured to perform data feature extraction on the governed data set based on feature engineering to obtain a single root column operation process target training data set, and determine a relevance feature training module of the extracted features, and train an intelligent recommendation model of single root column process full time node and single root column work parameter based on the relevance features and the single root column operation process target training data set.

8. A recommendation device, characterized in that It comprises: An acquisition module is configured to acquire real-time drilling data and single root column operation basic data, the real-time drilling data comprising real drilling wellbore data and single root column working condition parameters, and the single root column operation basic data comprising real drilling engineering technical data of the same basin, the same structure, the same zone, the same well type and the same machine type; A processing module is configured to input the real-time drilling data and the single root column operation basic data into a pre-trained intelligent recommendation model of single root column process full time node and single root column work parameter to obtain single root column process full time node and single root column work parameter corresponding to the real-time drilling data, the intelligent recommendation model being generated by the model generation method in any one of claims 1 to 5.

9. An electronic device, comprising: It comprises: A memory, a processor; The memory stores computer execution instructions; The processor executes the computer-executed instructions stored in the memory, so that the processor executes the method as claimed in any one of claims 1 to 5, and / or to implement the method as claimed in claim 6.

10. A computer-readable storage medium, characterized in that, The computer-executed instructions stored in the computer-readable storage medium are executed to implement the method as claimed in any one of claims 1 to 5, and / or to implement the method as claimed in claim 6.

Images