Drilling data generation method and system based on conditional diffusion model

Abstract

The invention provides a drilling data generation method and system based on a conditional diffusion model, and relates to the technical field of oil and gas well drilling. The method comprises the steps that a first data set is obtained, the first data set comprises multiple pieces of drilling data, the multiple pieces of drilling data are preprocessed, and a working condition category corresponding to each piece of drilling data is determined; performing iterative training on the neural network model according to the multiple drilling data and the working condition category corresponding to each drilling data to obtain a trained neural network model; receiving an input drilling data generation instruction, wherein the drilling data generation instruction carries a target working condition category; and inputting the target working condition category into the trained neural network model, and outputting multiple drilling data corresponding to the target working condition category. According to the invention, the well drilling data corresponding to different working condition categories can be rapidly and accurately obtained based on the trained neural network model, and then the risk prediction model is trained through the well drilling data corresponding to different working conditions, so that the prediction capability of the model is improved.

Description

Technical Field

[0001] This invention relates to the field of oil and gas well drilling technology, and in particular to a method and system for generating drilling data based on a conditional diffusion model. Background Technology

[0002] With the rapid development of artificial intelligence (AI) technology, its application in the oil and gas industry is becoming increasingly widespread, especially in the field of oil and gas well drilling. The introduction of intelligent technology has greatly improved operational efficiency and safety. Through data-driven algorithm models, many complex problems in the drilling process have been effectively solved, such as real-time monitoring of the downhole environment, automated control of drilling equipment, and risk prediction of downhole accidents.

[0003] However, a core challenge of intelligent drilling technology lies in the lack of data, especially drilling data involving high-risk or special operating conditions. In reality, high-quality, comprehensive drilling data suitable for training intelligent models is often insufficient, particularly when facing risky conditions (such as blowouts and well collapses). Because these situations have a low probability of occurrence but high potential danger, obtaining large amounts of such drilling data is virtually impossible. This data scarcity directly limits the predictive power of neural network models, resulting in existing intelligent systems often failing to provide sufficiently accurate predictions and guidance when dealing with complex and high-risk conditions. For example, while the probability of accidents like blowouts during drilling is low, their occurrence can lead to serious safety and economic consequences. To effectively predict and mitigate such risks, models must be trained with a sufficient amount of drilling data corresponding to risky operating conditions. However, because such data is difficult to accumulate through actual operations, models often perform poorly when dealing with similar emergencies. In related technologies, CN116304701A provides a method for generating HRRP samples based on a conditional denoising diffusion probability model. Although it can obtain high-resolution radar pulse samples in the radar field through the conditional denoising diffusion probability model, the above probability model cannot be applied to the field of oil and gas well drilling technology because drilling data correspond to different working conditions, and thus it cannot obtain drilling data corresponding to different working conditions.

[0004] Therefore, there is an urgent need for a drilling data generation method and system based on the conditional diffusion model, which can obtain drilling data corresponding to different working conditions based on the trained conditional diffusion model, and then use the drilling data corresponding to different working conditions to train the risk prediction model and improve the model's prediction ability. Summary of the Invention

[0005] This invention provides a drilling data generation method and system based on a conditional diffusion model, which can quickly and accurately obtain drilling data corresponding to different working conditions based on a trained neural network model, and then train a risk prediction model using the drilling data corresponding to different working conditions to improve the model's predictive ability.

[0006] To achieve the above objectives, the embodiments of the present invention adopt the following technical solutions:

[0007] Firstly, a drilling data generation method based on a conditional diffusion model is provided. The method includes: acquiring a first dataset comprising multiple drilling data sets, each set including multiple drilling feature parameters corresponding to different times, such as torque, total pool volume, drilling pressure, inlet flow rate, rotary table speed, outlet flow rate, and late arrival depth; preprocessing the multiple drilling data sets, including normalization, denoising, and missing value imputation; determining the operating condition category corresponding to each drilling data set; iteratively training a neural network model based on the multiple drilling data sets and the corresponding operating condition category to obtain a trained neural network model; receiving an input drilling data generation instruction carrying a target operating condition category; inputting the target operating condition category into the trained neural network model and outputting multiple drilling data sets corresponding to the target operating condition category.

[0008] In one possible implementation of the first aspect, determining the working condition category corresponding to each drilling data includes: determining the working condition category corresponding to each drilling data through a voting classifier based on a preset classification algorithm. The preset classification algorithm includes one or more of the following: support vector machine algorithm, logistic regression algorithm, random forest algorithm, adaptive boosting algorithm, multilayer perceptron algorithm, and decision tree algorithm.

[0009] In one possible implementation of the first aspect, the neural network model is a diffusion model based on the Transformer structure. The diffusion model includes a time encoding module, an input encoding module, and a Transformer module. The time encoding module is used to convert the time corresponding to each drilling data into an embedding vector. The input encoding module is used to encode each drilling data into a feature vector through linear projection. The Transformer module is used to perform denoising operations on the embedding vector and the feature vector to obtain the drilling data.

[0010] In one possible implementation of the first aspect, the Transformer module includes multiple stacked Transformer layers. Each Transformer layer includes a self-attention layer, a cross-attention layer, and a feedforward neural network layer. Each Transformer layer is configured with residual connections and a layer normalization mechanism. The self-attention layer is used to determine the contextual information included in the feature vector of each drilling data. The cross-attention layer is used to determine the correlation between the feature vector of each expert data and the working condition category. The feedforward neural network layer is used to process the feature vector based on the contextual information included in the feature vector and the correlation between the feature vector and the working condition category to obtain the corresponding classification result.

[0011] In one possible implementation of the first aspect, the neural network model is iteratively trained based on multiple drilling data and the working condition category corresponding to each drilling data to obtain a trained neural network model, including: constructing an objective function, wherein the objective function is a mean squared error function; and iteratively training the neural network model based on the objective function, according to multiple drilling data and the working condition category corresponding to each drilling data to obtain a trained neural network model.

[0012] The beneficial effects of this invention are as follows: The method provided by this invention can obtain drilling data corresponding to different working conditions through a diffusion model based on a Transformer structure. This data is then used to train a risk prediction model, improving the model's predictive ability. In other words, this invention significantly improves the efficiency and accuracy of drilling data generation and expansion by introducing a drilling data generation algorithm based on a diffusion model. By generating drilling data under various working conditions through a diffusion model, this invention not only enriches the existing drilling data dataset but also ensures a high degree of consistency between the generated data and real working conditions, thus providing strong support for the training and optimization of intelligent drilling systems. This method reduces reliance on data collection in actual operations, especially under high-risk or special working conditions, significantly reducing the difficulty and cost of data collection. Furthermore, the automated generation process accelerates data processing speed, enabling operators to more flexibly respond to complex underground environments and emergencies. Overall, this invention optimizes the generation and management process of drilling data through technological innovation, improving the safety, accuracy, and efficiency of drilling operations, demonstrating its broad application prospects and practical value in the oil and gas drilling field.

[0013] Secondly, embodiments of the present invention provide a drilling data generation system based on a conditional diffusion model. The system includes: an acquisition unit for acquiring a first dataset, which includes multiple drilling data sets, each containing multiple drilling feature parameters corresponding to different times, including torque, total pool volume, drilling pressure, inlet flow rate, rotary table speed, outlet flow rate, and late arrival depth; a processing unit for preprocessing the multiple drilling data sets, including normalization, denoising, and missing value imputation; a determination unit for determining the operating condition category corresponding to each drilling data set; a training unit for iteratively training a neural network model based on the multiple drilling data sets and the operating condition category corresponding to each set of drilling data sets, to obtain a trained neural network model; a receiving unit for receiving an input drilling data generation instruction, which carries a target operating condition category; and a generation unit for inputting the target operating condition category into the trained neural network model and outputting multiple drilling data sets corresponding to the target operating condition category.

[0014] In one possible implementation of the second aspect, the determining unit is specifically used to: determine the working condition category corresponding to each drilling data by means of a voting classifier based on a preset classification algorithm. The preset classification algorithm includes one or more of the following: support vector machine algorithm, logistic regression algorithm, random forest algorithm, adaptive boosting algorithm, multilayer perceptron algorithm, and decision tree algorithm.

[0015] In one possible implementation of the second aspect, the neural network model is a diffusion model based on the Transformer structure. The diffusion model includes a time encoding module, an input encoding module, and a Transformer module. The time encoding module is used to convert the time corresponding to each drilling data into an embedding vector. The input encoding module is used to encode each drilling data into a feature vector through linear projection. The Transformer module is used to perform denoising operations on the embedding vector and the feature vector to obtain the drilling data.

[0016] In one possible implementation of the second aspect, the Transformer module includes multiple stacked Transformer layers. Each Transformer layer includes a self-attention layer, a cross-attention layer, and a feedforward neural network layer. Each Transformer layer is configured with residual connections and a layer normalization mechanism. The self-attention layer is used to determine the contextual information included in the feature vector of each drilling data. The cross-attention layer is used to determine the correlation between the feature vector of each expert data and the working condition category. The feedforward neural network layer is used to process the feature vector based on the contextual information included in the feature vector and the correlation between the feature vector and the working condition category to obtain the corresponding classification result.

[0017] Thirdly, an electronic device is provided, the electronic device including a memory and one or more processors; the memory is coupled to the processors; wherein the memory stores computer program code, the computer program code including computer instructions, which, when executed by the processor, cause the electronic device to perform the method as described in any implementation of the first aspect.

[0018] Fourthly, a computer-readable storage medium is provided, including computer instructions that, when executed on an electronic device, cause the electronic device to perform a method as described in any implementation of the first aspect.

[0019] Fifthly, a computer program product is provided that, when run on a computer, causes the computer to perform the methods as described in any implementation of the first aspect.

[0020] Understandably, the beneficial effects achieved by the system of the second aspect, the electronic device of the third aspect, the computer-readable storage medium of the fourth aspect, and the computer program product of the fifth aspect provided above can be referred to with reference to the beneficial effects of the first aspect and any of its possible design embodiments, which will not be repeated here. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of the hardware structure of an electronic device according to an embodiment of the present invention;

[0022] Figure 2 This is a flowchart illustrating a drilling data generation method based on a conditional diffusion model, as shown in an embodiment of the present invention.

[0023] Figure 3 This is a flowchart illustrating a drilling data generation method based on a conditional diffusion model, as shown in an embodiment of the present invention.

[0024] Figure 4 This is a schematic diagram of the hardware structure of a generation system according to an embodiment of the present invention. Detailed Implementation

[0025] The technical solutions of the embodiments of the present invention will be described below with reference to the accompanying drawings. In the description of the present invention, unless otherwise stated, " / " indicates that the objects before and after are in an "or" relationship; for example, A / B can represent A or B. "And / or" in the present invention is merely a description of the relationship between the related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone, where A and B can be singular or plural. Furthermore, in the description of the present invention, unless otherwise stated, "multiple" refers to two or more. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items.

[0026] Furthermore, to facilitate a clear description of the technical solutions of the embodiments of the present invention, the terms "first" and "second" are used in the embodiments of the present invention to distinguish identical or similar items with essentially the same function and effect. Those skilled in the art will understand that the terms "first" and "second" do not limit the quantity or execution order, and the terms "first" and "second" are not necessarily different.

[0027] In this embodiment of the invention, the terms "exemplary" or "for example" are used to indicate that something is being described as an example, illustration, or illustration. Any embodiment or design described as "exemplary" or "for example" in this embodiment of the invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of terms such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner for ease of understanding.

[0028] With the rapid development of artificial intelligence (AI) technology, its application in the oil and gas industry is becoming increasingly widespread, especially in the field of oil and gas well drilling. The introduction of intelligent technology has greatly improved operational efficiency and safety. Through data-driven algorithm models, many complex problems in the drilling process have been effectively solved, such as real-time monitoring of the downhole environment, automated control of drilling equipment, and risk prediction of downhole accidents.

[0029] However, a core challenge of intelligent drilling technology lies in the lack of data, especially drilling data involving high-risk or special operating conditions. In reality, high-quality, comprehensive drilling data suitable for training intelligent models is often insufficient, particularly when facing risky conditions (such as blowouts and well collapses). Because these situations have a low probability of occurrence but high potential danger, obtaining large amounts of such drilling data is virtually impossible. This data scarcity directly limits the predictive power of neural network models, resulting in existing intelligent systems often failing to provide sufficiently accurate predictions and guidance when dealing with complex and high-risk conditions. For example, while the probability of accidents like blowouts during drilling is low, their occurrence can lead to serious safety and economic consequences. To effectively predict and mitigate such risks, models must be trained with a sufficient amount of drilling data corresponding to risky operating conditions. However, because such data is difficult to accumulate through actual operations, models often perform poorly when dealing with similar emergencies. In related technologies, CN116304701A provides a method for generating HRRP samples based on a conditional denoising diffusion probability model. Although it can obtain high-resolution radar pulse samples in the radar field through the conditional denoising diffusion probability model, the above probability model cannot be applied to the field of oil and gas well drilling technology because drilling data correspond to different working conditions, and thus it cannot obtain drilling data corresponding to different working conditions.

[0030] Therefore, there is an urgent need for a drilling data generation method and system based on the conditional diffusion model, which can obtain drilling data corresponding to different working conditions based on the trained conditional diffusion model, and then use the drilling data corresponding to different working conditions to train the neural network model and improve the predictive ability of the neural network model.

[0031] In view of this, embodiments of the present invention provide a drilling data generation method based on a conditional diffusion model. The method includes: acquiring a first dataset, the first dataset including multiple drilling data, each drilling data including multiple drilling feature parameters corresponding to different times, the drilling feature parameters including torque, total pool volume, drilling pressure, inlet flow rate, rotary table speed, outlet flow rate, and late arrival depth; preprocessing the multiple drilling data, the preprocessing including normalization, denoising, and missing value imputation; determining the operating condition category corresponding to each drilling data; iteratively training a neural network model based on the multiple drilling data and the operating condition category corresponding to each drilling data to obtain a trained neural network model; receiving an input drilling data generation instruction, the drilling data generation instruction carrying a target operating condition category; inputting the target operating condition category into the trained neural network model, and outputting multiple drilling data corresponding to the target operating condition category.

[0032] The method provided by this invention utilizes a diffusion model based on a Transformer structure to obtain drilling data corresponding to different working conditions. This data is then used to train a risk prediction model, enhancing its predictive capabilities. In other words, this invention significantly improves the efficiency and accuracy of drilling data generation and expansion by introducing a drilling data generation algorithm based on a diffusion model. By generating drilling data under various working conditions using a diffusion model, this invention not only enriches existing drilling data datasets but also ensures a high degree of consistency between the generated data and real-world working conditions, thus providing strong support for the training and optimization of intelligent drilling systems. This method reduces reliance on data collection during actual operations, especially under high-risk or special working conditions, significantly reducing the difficulty and cost of data collection. Furthermore, the automated generation process accelerates data processing, enabling operators to respond more flexibly to complex underground environments and unexpected situations. Overall, this invention optimizes the generation and management process of drilling data through technological innovation, improving the safety, accuracy, and efficiency of drilling operations, demonstrating its broad application prospects and practical value in the oil and gas drilling field.

[0033] In some embodiments, the drilling data generation method based on the conditional diffusion model provided in this invention can be executed by a drilling data generation system 100 based on the conditional diffusion model (hereinafter referred to as generation system 100). As an example, generation system 100 can be any electronic device 200 with data processing capabilities, such as a general-purpose computer, personal computer, laptop computer, switch, or tablet computer, etc. The specific implementation of the procurement list generation system 100 is not limited here.

[0034] Figure 1 A schematic diagram of the hardware structure of an electronic device provided in an embodiment of the present invention is shown. The electronic device 200 includes a processor 210, a memory 220, and a communication interface 230.

[0035] Processor 210 may include one or more processing cores. Processor 210 connects to various parts within electronic device 200 using various interfaces and lines, and performs various functions and processes data of electronic device 200 by running or executing instructions, programs, code sets, or instruction sets stored in memory 220, and by calling data stored in memory 220. Optionally, processor 210 may be implemented using at least one of the following hardware forms: Central Processing Unit (CPU), Graphics Processing Unit (GPU), Digital Signal Processing (DSP), Field-Programmable Gate Array (FPGA), and Programmable Logic Array (PLA).

[0036] The memory 220 may include random access memory (RAM) or read-only memory (ROM). Optionally, the memory 220 may include non-transitory computer-readable storage medium. The memory 220 may be used to store instructions, programs, code, code sets, or instruction sets. The memory 220 may include a stored program area. The stored program area may store instructions for implementing an operating system, instructions for implementing at least one function (such as data acquisition functions, model training functions, etc.), and instructions for implementing the various method embodiments described above.

[0037] The communication interface 230 is used to communicate with other devices, equipment, or communication networks, such as data storage devices, image processing devices, or Ethernet, wireless access networks (RAN), wireless local area networks (WLAN), etc.

[0038] In terms of physical implementation, the aforementioned devices (such as processor 210, memory 220, and communication interface 230) can each be devices within the same device (such as a laptop computer). Alternatively, at least two of these devices can be located within the same device, i.e., as different devices within the same device, similar to the deployment of devices or components in a distributed system.

[0039] It is understood that the structure illustrated in this embodiment does not constitute a specific limitation on the electronic device 200. In other embodiments of the present invention, the electronic device 200 may include more or fewer components than illustrated, or combine some components, or split some components, or have different component arrangements. The illustrated components may be implemented in hardware, software, or a combination of software and hardware.

[0040] The drilling data generation method based on the conditional diffusion model provided by the embodiments of the present invention will be described below with reference to the accompanying drawings.

[0041] Figure 2 This is a flowchart illustrating a drilling data generation method based on a conditional diffusion model, provided as an embodiment of the present invention. Optionally, this method can be... Figure 1 The illustrated electronic device 200 performs this operation, that is, the production system 100 performs it. The method may include the following steps:

[0042] S1. Obtain the first dataset, which includes multiple drilling data.

[0043] Specifically, each drilling data point includes multiple drilling characteristic parameters corresponding to different times, including torque, total pool volume, drilling pressure, inlet flow rate, rotary table speed, outlet flow rate, and late well depth.

[0044] For example, see Table 1, which is a drilling data table provided in an embodiment of the present invention, including drilling data 1, drilling data 2 and drilling data 3.

[0045] Table 1

[0046]

[0047]

[0048] S2. Preprocessing of multiple drilling data includes normalization, noise reduction, and missing value imputation.

[0049] S3. Determine the operating condition category corresponding to each drilling data.

[0050] In one example, the working condition categories include jacking up, directional drilling, tripping, running down, circulation, and casing running. It should be understood that the above working condition categories are merely illustrative and may include more or fewer items than those in the above examples. The embodiments of the present invention do not impose any particular limitations on this.

[0051] In some embodiments, S3 specifically includes the following steps:

[0052] Based on a pre-defined classification algorithm, a voting classifier is used to determine the working condition category corresponding to each drilling data. The pre-defined classification algorithm includes one or more of the following: support vector machine algorithm, logistic regression algorithm, random forest algorithm, adaptive boosting algorithm, multilayer perceptron algorithm, and decision tree algorithm.

[0053] S4. Iteratively train the neural network model based on multiple drilling data and the corresponding working condition category for each drilling data to obtain the trained neural network model.

[0054] In one possible implementation, the neural network model is a diffusion model based on the Transformer structure. The diffusion model includes a time encoding module, an input encoding module, and a Transformer module. The time encoding module is used to convert the time corresponding to each drilling data into an embedding vector. The input encoding module is used to encode each drilling data into a feature vector through linear projection. The Transformer module is used to perform denoising operations on the embedding vector and the feature vector to obtain the drilling data.

[0055] Furthermore, the Transformer module includes multiple stacked Transformer layers. Each Transformer layer includes a self-attention layer, a cross-attention layer, and a feedforward neural network layer. Each Transformer layer is configured with residual connections and a layer normalization mechanism. The self-attention layer is used to determine the contextual information included in the feature vector of each drilling data. The cross-attention layer is used to determine the correlation between the feature vector of each expert data and the working condition category. The feedforward neural network layer is used to process the feature vector based on the contextual information included in the feature vector and the correlation between the feature vector and the working condition category to obtain the corresponding classification result.

[0056] Specifically, the neural network model's operation is divided into two stages: forward diffusion and backward generation. In the forward diffusion stage, the neural network model gradually "randomizes" each drilling data point by adding noise, generating normally distributed random variables. The backward generation stage is the core of this invention. Through a reverse diffusion process, i.e., by gradually removing noise, the neural network model learns and reconstructs the true distribution of the drilling data. In particular, by introducing drilling condition categories as a condition during the generation process, the neural network model can generate drilling data corresponding to different drilling condition categories. This mechanism ensures that the generated drilling data not only has high realism but also effectively reflects the characteristics of different drilling condition categories, greatly improving the model's applicability. Furthermore, this invention employs a diffusion model based on the Transformer structure, which effectively improves the correlation and accuracy of the drilling data during the generation process.

[0057] Specifically, the Transformer structure, as the backbone of the denoising network, can capture the complex relationships between different features in the input data through its powerful self-attention mechanism. Meanwhile, this invention introduces a Cross-Attention module into the Transformer. The Cross-Attention module can deeply correlate the conditional information of the working condition category with the input noisy data, ensuring that each set of drilling data generated conforms to the feature distribution of the corresponding working condition category.

[0058] In the denoising process, the Cross-Attention module plays a crucial role: by using the working condition category as the input query, the neural network model can effectively capture the interaction between the working condition category and other drilling parameters (such as drilling pressure, torque, and drilling speed), and guide each step of the denoising operation accordingly. This mechanism ensures that the generated drilling data is not only highly consistent with the working condition category, but also preserves the interdependencies between various parameters at the feature level.

[0059] Unlike traditional diffusion models, the Transformer structure employed in this invention possesses stronger global contextual understanding capabilities, enabling more accurate restoration of the global data distribution during denoising. Furthermore, the introduction of the Cross-Attention module further enhances the model's sensitivity to operating condition categories, ensuring that the generated data accurately captures feature changes under complex operating conditions. This design not only improves the realism of the generated drilling data but also significantly reduces data bias caused by changes in operating condition categories, making the generated drilling data highly applicable across different operating condition categories.

[0060] Optionally, to ensure that the diffusion model can accurately simulate drilling data corresponding to various working conditions, this invention also includes a strategy for training and optimizing the neural network model. First, the neural network model is initially trained using existing actual drilling data, learning the basic distribution of different working condition categories. Second, the neural network model is continuously adaptively optimized using simulated drilling data, ensuring that the drilling data generated by the neural network model can more accurately reflect the complex situations corresponding to different working condition categories in actual operation. Therefore, the method provided by this invention has broad application prospects, not only applicable to expanding training data for intelligent drilling systems, but also providing important data support for drilling equipment optimization, working condition simulation, and equipment failure prediction, further promoting the development of intelligent drilling technology. By generating diverse drilling data, this invention can effectively solve the problem of data scarcity, especially the risk prediction and safety control problems under high-risk working condition categories.

[0061] In some embodiments, see Figure 3The above S4 specifically includes the following steps:

[0062] S41. Construct the objective function, which is the mean square error function;

[0063] To facilitate understanding of this solution, the neural network model provided in the embodiments of the present invention will be further described below. The method provided by the present invention uses the mean square error function (MSE) to calculate the error between the predicted noise and the real noise, with the aim of making the predicted noise as close as possible to the real noise.

[0064] S42. Based on the objective function, the neural network model is iteratively trained according to multiple drilling data and the working condition category corresponding to each drilling data to obtain the trained neural network model.

[0065] The method provided in this invention uses multiple drilling data to iteratively train a neural network model. The neural network model captures the correlation between working condition categories and various features through cross-attention layers, introducing additional contextual information in the training of the denoising model. This enables a better understanding and capture of the relationship between input data and conditions (such as input noise level, semantic information, or other external features), ensuring that the generated data is consistent with the actual working condition categories.

[0066] Unlike traditional U-Net-based denoising networks, Transformer-based neural network models can capture global dependencies across the entire feature sequence. U-Net, primarily reliant on convolutional operations, has a limited receptive field, especially in early layers. This makes Transformer architectures better suited for handling complex long-range dependencies, thus better generating drilling data over a given period. Traditional U-Net convolutional layers apply fixed weights to features, while Transformer architectures assign dynamic weights through self-attention, allowing for flexible adjustment of the importance of different features. This dynamic weight allocation is particularly helpful in processing diffusion models containing significant noise, enabling more accurate separation of detailed information from different drilling data sets.

[0067] S5. Receive the input drilling data generation command, which carries the target working condition category;

[0068] S6. Input the target working condition category into the trained neural network model and output multiple drilling data corresponding to the target working condition category.

[0069] Specifically, during the drilling data generation phase, the generation system uses the identified target working condition category as a condition, inputting it into the trained neural network model. The neural network model progressively denoises based on the target working condition category, thereby generating drilling data with high accuracy corresponding to the target working condition category. For example, if the input target working condition category is directional drilling, the neural network model will output drilling data corresponding to the directional drilling condition. The generated drilling data includes multiple drilling characteristic parameters corresponding to different times, including torque, total pool volume, drilling pressure, inlet flow rate, rotary table speed, outlet flow rate, and late well depth; the multiple drilling characteristic parameters included in each drilling data set are consistent in time and space.

[0070] For example, see Table 2, which shows multiple drilling data corresponding to the target working condition category shown in the embodiment of the present invention, namely drilling data 4, drilling data 5 and drilling data 6.

[0071] Table 2

[0072]

[0073] As described in S1-S6 above, the method provided by this invention can obtain drilling data corresponding to different working conditions through a diffusion model based on a Transformer structure. This data is then used to train a risk prediction model, improving its predictive capabilities. In other words, this invention significantly improves the efficiency and accuracy of drilling data generation and expansion by introducing a drilling data generation algorithm based on a diffusion model. By generating drilling data under various working conditions through a diffusion model, this invention not only enriches the existing drilling data dataset but also ensures a high degree of consistency between the generated data and real working conditions, thus providing strong support for the training and optimization of intelligent drilling systems. This method reduces reliance on data collection in actual operations, especially under high-risk or special working conditions, significantly reducing the difficulty and cost of data collection. Furthermore, the automated generation process accelerates data processing, enabling operators to more flexibly respond to complex underground environments and emergencies. Overall, this invention optimizes the generation and management process of drilling data through technological innovation, improving the safety, accuracy, and efficiency of drilling operations, demonstrating its broad application prospects and practical value in the oil and gas drilling field.

[0074] The beneficial effects of the method provided in this invention are as follows: Firstly, it enriches the drilling data dataset and enhances the predictive capabilities of intelligent drilling systems. The method automatically generates drilling data under various operating conditions through a conditional diffusion model, particularly high-risk and special operating condition categories that are difficult to collect through actual operations, significantly enriching existing drilling datasets. By generating more diverse data, the training data for the risk prediction model is expanded, further improving the accuracy and reliability of intelligent drilling systems in predicting complex operating conditions, and helping to reduce potential operational risks. Secondly, it improves the accuracy and applicability of data generation. By introducing a Transformer structure as a denoising network and integrating a Cross-Attention module within it, this invention can better capture the complex relationship between operating condition categories and various drilling feature data, ensuring that the generated data highly conforms to actual operating conditions. This innovative design significantly improves the accuracy of the generated data, making it highly applicable under different operating conditions and truly reflecting the changes in key parameters during the drilling process. Thirdly, it enhances the flexibility and scalability of the model. The method provided in this embodiment of the invention is highly flexible, dynamically adjusting parameters and conditions during the generation process according to different working conditions to adapt to the needs of various drilling scenarios. Whether in simple or complex conditions, the algorithm can generate highly adaptable simulation data, providing accurate support for drilling operations. Simultaneously, the model structure of this invention has good scalability, continuously optimizing and adapting to other drilling-related data generation tasks as needed. Fourthly, the method provided in this embodiment of the invention improves drilling decision support efficiency and reduces operational risks. By generating highly realistic drilling data, this invention significantly improves decision support efficiency during the drilling process. Maintenance personnel can rely on this drilling data for accurate working condition analysis, equipment status prediction, and optimized drilling parameter settings, reducing blind spots and risks in actual operations. This not only improves the safety and economy of operations but also provides strong data support for future intelligent and unmanned drilling operations. Finally, the method provided in this embodiment of the invention is applicable to the generation and expansion of various drilling data. This invention is not only applicable to drilling condition data, but also to the generation and expansion of any drilling-related data, including drilling equipment parameters, downhole environment data, drilling fluid parameters, etc. Whether for equipment maintenance, drilling parameter optimization, or simulation of the downhole working environment, this invention can generate high-quality data, further promoting the comprehensive development of intelligent drilling.

[0075] The foregoing primarily describes the solutions of the embodiments of the present invention from a methodological perspective. It is understood that, in order to achieve the above-mentioned functions, the generation system includes at least one of the hardware structures and software modules corresponding to the execution of each function. Those skilled in the art should readily recognize that, in conjunction with the units and algorithm steps of the various examples described in the embodiments disclosed herein, the embodiments of the present invention can be implemented in hardware or a combination of hardware and computer software. Whether a function is executed in hardware or by computer software driving hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the embodiments of the present invention.

[0076] In this embodiment of the invention, the generation system can be divided into functional units according to the above method example. For example, the generation system can be divided into functional units corresponding to each function, or two or more functions can be integrated into one processing unit. The integrated unit can be implemented in hardware or as a software functional unit. It should be noted that the unit division in this embodiment of the invention is illustrative and only represents one logical functional division; other division methods may be used in actual implementation.

[0077] For example, Figure 4 A schematic diagram of the hardware structure of a generation system provided in an embodiment of the present invention is shown. The generation system 100 includes: an acquisition unit 110, used to acquire a first dataset, the first dataset including multiple drilling data, each drilling data including multiple drilling feature parameters corresponding to different times, the drilling feature parameters including torque, total pool volume, drilling pressure, inlet flow rate, rotary table speed, outlet flow rate, and late well depth; a processing unit 120, used to preprocess the multiple drilling data, the preprocessing including normalization processing, noise reduction processing, and missing value imputation; a determination unit 130, used to determine the working condition category corresponding to each drilling data; a training unit, used to iteratively train a neural network model based on the multiple drilling data and the working condition category corresponding to each drilling data to obtain a trained neural network model; a receiving unit 140, used to receive an input drilling data generation instruction, the drilling data generation instruction carrying a target working condition category; and a generation unit 150, used to input the target working condition category into the trained neural network model and output the multiple drilling data corresponding to the target working condition category.

[0078] Optionally, the determining unit 130 is specifically used to: determine the working condition category corresponding to each drilling data through a voting classifier based on a preset classification algorithm. The preset classification algorithm includes one or more of the following: support vector machine algorithm, logistic regression algorithm, random forest algorithm, adaptive boosting algorithm, multilayer perceptron algorithm, and decision tree algorithm.

[0079] Optionally, the neural network model is a diffusion model based on the Transformer structure. The diffusion model includes a time encoding module, an input encoding module, and a Transformer module. The time encoding module is used to convert the time corresponding to each drilling data into an embedding vector. The input encoding module is used to encode each drilling data into a feature vector through linear projection. The Transformer module is used to perform denoising operations on the embedding vector and the feature vector to obtain the drilling data.

[0080] Optionally, the Transformer module includes multiple stacked Transformer layers. Each Transformer layer includes a self-attention layer, a cross-attention layer, and a feedforward neural network layer. Each Transformer layer is configured with residual connections and a layer normalization mechanism. The self-attention layer is used to determine the contextual information included in the feature vector of each drilling data. The cross-attention layer is used to determine the correlation between the feature vector of each expert data and the working condition category. The feedforward neural network layer is used to process the feature vector based on the contextual information included in the feature vector and the correlation between the feature vector and the working condition category to obtain the corresponding classification result.

[0081] It should be understood that specific descriptions of the above-mentioned optional methods can be found in the foregoing method embodiments, and will not be repeated here. Furthermore, explanations of any of the above-provided generation systems 100 and descriptions of their beneficial effects can be found in the corresponding method embodiments, and will not be repeated here.

[0082] This invention also provides a computer-readable storage medium storing at least one computer instruction, which is loaded and executed by a processor to implement the methods of the various embodiments described above. Explanations of the relevant content and descriptions of the beneficial effects of any of the computer-readable storage media provided above can be found in the corresponding embodiments described above, and will not be repeated here.

[0083] This invention also provides a chip. This chip integrates a control circuit for implementing the functions of the aforementioned generation system 100 and one or more ports. Optionally, the functions supported by this chip are as described above and will not be repeated here.

[0084] Those skilled in the art will understand that the program for implementing all or part of the steps of the above embodiments, which can be executed by a program instructing related hardware, can be stored in a computer-readable storage medium. The storage medium mentioned above can be a read-only memory, a random access memory, etc. The processing unit or processor mentioned above can be a central processing unit, a general-purpose processor, an application-specific integrated circuit (ASIC), a microprocessor (DSP), a field-programmable gate array (FPGA), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof.

[0085] This invention also provides a computer program product containing instructions that, when executed on a computer, cause the computer to perform any of the methods described in the above embodiments. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the flow or function according to the embodiments of this invention is generated. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, computer instructions may be transmitted from one website, computer, server, or data center to another via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium accessible to a computer or a data storage device such as a server or data center that integrates one or more available media. The available medium may be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., SSD), etc.

[0086] It should be noted that the devices for storing computer instructions or computer programs provided in the embodiments of the present invention, such as, but not limited to, the aforementioned memory, computer-readable storage medium, and communication chip, are all non-transitory. Those skilled in the art should recognize that the functions described in the embodiments of the present invention in one or more of the above examples can be implemented using hardware, software, firmware, or any combination thereof. When implemented using software, these functions can be stored in a computer-readable storage medium or transmitted as one or more instructions or code on a computer-readable storage medium. Computer-readable storage media include computer storage media and communication media, wherein communication media include any medium that facilitates the transmission of computer programs from one place to another. Storage media can be any available medium accessible to general-purpose or special-purpose computers.

[0087] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A drilling data generation method based on a conditional diffusion model, characterized in that, The method includes: Obtain a first dataset, which includes multiple drilling data. Each drilling data includes multiple drilling characteristic parameters corresponding to different times. The drilling characteristic parameters include torque, total pool volume, drilling pressure, inlet flow rate, rotary table speed, outlet flow rate, and late well depth. The multiple drilling data are preprocessed, including normalization, noise reduction, and missing value imputation. Determine the operating condition category corresponding to each drilling data; The neural network model is iteratively trained based on the multiple drilling data and the working condition category corresponding to each drilling data to obtain the trained neural network model. Receive an input drilling data generation instruction, the drilling data generation instruction carrying a target working condition category; Input the target working condition category into the trained neural network model, and output multiple drilling data corresponding to the target working condition category.

2. The method according to claim 1, characterized in that, Determining the operating condition category corresponding to each drilling data includes: Based on a preset classification algorithm, a voting classifier is used to determine the working condition category corresponding to each drilling data. The preset classification algorithm includes one or more of the following: support vector machine algorithm, logistic regression algorithm, random forest algorithm, adaptive boosting algorithm, multilayer perceptron algorithm, and decision tree algorithm.

3. The method according to claim 2, characterized in that, The neural network model is a diffusion model based on the Transformer structure. The diffusion model includes a time encoding module, an input encoding module, and a Transformer module. The time encoding module is used to convert the time corresponding to each drilling data into an embedding vector. The input encoding module is used to encode each drilling data into a feature vector through linear projection. The Transformer module is used to perform denoising operations on the embedding vector and the feature vector to obtain the drilling data.

4. The method according to claim 3, characterized in that, The Transformer module includes multiple stacked Transformer layers. Each Transformer layer includes a self-attention layer, a cross-attention layer, and a feedforward neural network layer. Each Transformer layer is configured with residual connections and a layer normalization mechanism. The self-attention layer is used to determine the contextual information included in the feature vector of each drilling data. The cross-attention layer is used to determine the correlation between the feature vector of each expert data and the working condition category. The feedforward neural network layer is used to process the feature vector based on the contextual information included in the feature vector and the correlation between the feature vector and the working condition category to obtain the corresponding classification result.

5. The method according to claim 4, characterized in that, The step of iteratively training the neural network model based on the multiple drilling data and the corresponding working condition category for each drilling data to obtain the trained neural network model includes: Construct an objective function, which is the mean squared error function; Based on the objective function, the neural network model is iteratively trained according to the multiple drilling data and the working condition category corresponding to each drilling data to obtain the trained neural network model.

6. A drilling data generation system based on a conditional diffusion model, characterized in that, The system includes: The acquisition unit is used to acquire a first dataset, which includes multiple drilling data. Each drilling data includes multiple drilling characteristic parameters corresponding to different times. The drilling characteristic parameters include torque, total pool volume, drilling pressure, inlet flow rate, rotary table speed, outlet flow rate, and late well depth. The processing unit is used to preprocess the multiple drilling data, including normalization, noise reduction, and missing value imputation. The determination unit is used to determine the operating condition category corresponding to each drilling data; The training unit is used to iteratively train the neural network model based on the multiple drilling data and the working condition category corresponding to each drilling data, so as to obtain the trained neural network model. A receiving unit is used to receive an input drilling data generation instruction, wherein the drilling data generation instruction carries a target working condition category; The generation unit is used to input the target working condition category into the trained neural network model and output multiple drilling data corresponding to the target working condition category.

7. The system according to claim 6, characterized in that, The determining unit is specifically used for: Based on a preset classification algorithm, a voting classifier is used to determine the working condition category corresponding to each drilling data. The preset classification algorithm includes one or more of the following: support vector machine algorithm, logistic regression algorithm, random forest algorithm, adaptive boosting algorithm, multilayer perceptron algorithm, and decision tree algorithm.

8. The system according to claim 7, characterized in that, The neural network model is a diffusion model based on the Transformer structure. The diffusion model includes a time encoding module, an input encoding module, and a Transformer module. The time encoding module is used to convert the time corresponding to each drilling data into an embedding vector. The input encoding module is used to encode each drilling data into a feature vector through linear projection. The Transformer module is used to perform denoising operations on the embedding vector and the feature vector to obtain the drilling data.

9. The system according to claim 8, characterized in that, The Transformer module includes multiple stacked Transformer layers. Each Transformer layer includes a self-attention layer, a cross-attention layer, and a feedforward neural network layer. Each Transformer layer is configured with residual connections and a layer normalization mechanism. The self-attention layer is used to determine the contextual information included in the feature vector of each drilling data. The cross-attention layer is used to determine the correlation between the feature vector of each expert data and the working condition category. The feedforward neural network layer is used to process the feature vector based on the contextual information included in the feature vector and the correlation between the feature vector and the working condition category to obtain the corresponding classification result.

10. An electronic device, characterized in that, include: processor; Memory used to store the processor's executable instructions; The processor is configured to execute the instructions to implement the drilling data generation method based on the conditional diffusion model as described in any one of claims 1-5.

Images