Method for autonomous update of predictive model of physical world environment perception and related device

Abstract

The application provides a physical world environment perception prediction model autonomous updating method and related equipment, which trains a constructed prediction model by using physical world environment data, decomposes a whole drilling task into multiple subtasks by using a task planning agent, inputs a geological prediction model and physical world environment data into a digital twin hybrid world model in a simulation deduction agent according to the received subtasks to perform forward deduction simulation, compares simulation deduction results with real formation data collected while drilling and actual drilling feedback data in an inversion model quality evaluation agent after drilling operation is completed, feeds back comparison results to the task planning agent, decomposes the whole drilling task again based on the comparison results, and returns to perform deduction simulation and prediction result comparison, so as to realize autonomous updating of the geological prediction model, greatly improve the geological prediction accuracy, and further improve the safety, efficiency and resource exploration accuracy of drilling operation.

Description

Technical Field

[0001] This invention relates to the interdisciplinary field of artificial intelligence and geological prediction, and in particular to a method and related equipment for autonomously updating a prediction model of physical world environmental perception. Background Technology

[0002] Deep geological drilling is a core component of mineral and oil and gas resource exploration and development. Its main challenge lies in safe and efficient drilling in unknown and complex geological environments, while well path planning and geological resource distribution prediction are the two core elements determining the success or failure of drilling. Existing technologies have developed various solutions around these two aspects, but all have significant inherent drawbacks, as follows: Traditional drilling path planning methods are mainly based on static geological prior models obtained from previous exploration, combined with engineers' human experience to design wellbore trajectories, and then local optimization is performed through conventional path planning algorithms. These methods are highly dependent on human experience, and their response to sudden geological anomalies during drilling is lagging, making it impossible to achieve real-time dynamic adjustment of the trajectory. In deep and complex strata, drilling safety accidents are very likely to occur, and it is also difficult to guarantee drilling efficiency.

[0003] With the development of deep learning and reinforcement learning technologies, data-driven intelligent path planning methods have become a research hotspot. Researchers have introduced deep reinforcement learning algorithms into drilling trajectory optimization, realizing automated trajectory generation based on geological models. However, the core of these methods still relies on fixed static geological models, lacking the ability to actively perceive and interact with unknown environments, and cannot adapt to the highly uncertain scenarios of dynamic changes in geological conditions during deep drilling.

[0004] Currently, researchers have constructed high-fidelity simulation environments such as AI2-THOR, Habitat-Sim, and iGibson, and developed methods based on deep reinforcement learning for point navigation, target navigation, and active environment exploration, achieving efficient autonomous navigation and environmental perception in static indoor scenes. Some studies have achieved exploration and map construction in unknown environments through active neural SLAM technology, and achieved near-theoretical optimal navigation results through distributed reinforcement learning algorithms. Other studies have improved the generalization of navigation in unknown environments by introducing prior knowledge of the scene. However, these navigation methods are all designed for structured and static indoor environments, and do not consider the dynamic evolution of the geological environment, high-risk engineering constraints, and the need to fuse multi-source heterogeneous drilling-while-drilling sensing information in deep drilling scenarios. They cannot be directly transferred to drilling path planning scenarios, nor can they achieve multi-objective balance optimization between drilling efficiency and safety risks.

[0005] Most existing geological resource distribution prediction models are trained offline based on previous exploration data. With fixed model parameters and structures, they cannot be updated autonomously in real time with the actual stratigraphic data obtained during drilling, resulting in a persistent discrepancy between prediction results and actual geological conditions. Some studies have introduced digital twin technology to construct virtual geological environments, enabling simulations of the drilling process. However, these lack quantitative quality assessment mechanisms for the prediction models and fail to establish a closed-loop linkage between virtual simulations and real-world feedback. This prevents automated iterative optimization of the models, leading to low safety, efficiency, and resource exploration accuracy in drilling operations. Summary of the Invention

[0006] This invention provides a method and related equipment for autonomously updating a prediction model of physical world environment perception, with the aim of improving the safety, efficiency and accuracy of drilling operations and resource exploration.

[0007] To achieve the above objectives, this invention provides a method for autonomously updating a prediction model for physical world environmental perception, comprising: Step 1: Collect physical world environmental data during the drilling process. Physical world environmental data includes downhole image data of drilled formations, lithological and physical property data from logging while drilling, geological attribute data from advanced exploration, current wellbore trajectory data, and drilling engineering parameter data. Step 2: Train the constructed prediction model using physical world environmental data to obtain the geological prediction model; Step 3: Use the task planning agent to break down the overall drilling task into multiple sub-tasks, and distribute each sub-task to the simulation simulation agent for simulating the geological prediction model and the inversion simulation quality assessment agent for assessing the quality of the simulation simulation results based on the actual prediction results. Step 4: Based on the received sub-tasks, input the geological prediction model and physical world environmental data into the digital twin hybrid world model in the simulation agent to perform forward simulation and obtain the simulation results. Step 5: Based on the received sub-tasks, the simulation results are compared with the actual formation data and actual drilling feedback data collected during drilling operations in the inversion model quality assessment agent to obtain the comparison results. The comparison results are then fed back to the task planning agent. Based on the comparison results, the overall drilling task is further decomposed and returned to step 4 to achieve autonomous updating of the geological prediction model until the update termination condition is met. Finally, the geological prediction model is output to perform geological prediction for the un-drilled area.

[0008] Furthermore, prior to step 2, the following steps are also included: Preprocessing physical world environmental data yields basic environmental data.

[0009] Furthermore, the predictive models include: The input module is used to perform data standardization and feature embedding processing on physical world environmental data. The multi-scale convolutional encoder module is used to extract the lithology, physical properties, and geological body control distribution characteristics of strata at different depths from the data processed by the input module. A cross-scale semantic fusion module is used to fuse measured semantic features of drilled areas with advanced detection features of undrilled areas; The decoder module is used to output geological prediction results for un-drilled areas.

[0010] Furthermore, the digital twin hybrid world model includes: The 3D virtual simulation environment module is used to construct a 3D virtual drilling scene based on the preliminary exploration data of the target area and the real data of the drilled strata. The real-time physical world mapping module is used to synchronize real formation data, drilling status data, and engineering parameter data collected during drilling to the three-dimensional virtual drilling scene in real time. The forward-looking simulation module is used to perform forward-looking simulations of geological prediction models in a virtual environment and output simulation results.

[0011] Furthermore, a three-dimensional virtual drilling scenario is constructed based on the preliminary exploration data of the target area and the actual data of the drilled strata, including: The preliminary exploration data and actual data of drilled strata in the target area are cleaned, denoised, standardized, and have their spatial coordinates unified. Based on the unified spatial coordinates, the formation interface, lithological property data, current wellbore trajectory data and drilling engineering parameter data are registered and fused to construct a three-dimensional geological structure model and wellbore model of the target area. The three-dimensional geological structure model, wellbore model, and drilling boundary conditions are input into the digital twin simulation engine to generate a three-dimensional virtual drilling scene corresponding to the actual drilling process.

[0012] Furthermore, before breaking down the overall drilling mission into multiple sub-tasks, it also includes: Build a visual interactive interface in the task planning agent; The geological features of the target area, on-site operational experience, and potential risk prediction results are input into the task planning agent through a visual interactive interface. The task objectives of the task planning agent are revised based on geological patterns, field operation experience, and the results of potential risk prediction.

[0013] Furthermore, before comparing the simulation results with the actual formation data collected during drilling and the actual drilling feedback data after the drilling operation is completed, the following steps are also included: The simulation results were manually verified.

[0014] Furthermore, when the comparison result exceeds the preset threshold, the geological prediction model is fine-tuned and optimized using incremental learning and real formation data collected during drilling operations and actual drilling feedback data.

[0015] This invention also provides an autonomous updating device for a prediction model of physical world environment perception, which applies an autonomous updating method for the prediction model. The autonomous updating device for the prediction model includes: The data acquisition module is used to acquire physical world environmental data during the drilling process. The physical world environmental data includes downhole image data of drilled formations, lithological and physical property data from logging while drilling, geological attribute data from advanced exploration, current wellbore trajectory data, and drilling engineering parameter data. The training module is used to train the constructed prediction model using physical world environmental data to obtain a geological prediction model. The decomposition module is used to decompose the overall drilling task into multiple sub-tasks using the task planning agent, and to distribute each sub-task to the simulation simulation agent for simulating the geological prediction model and the inversion simulation quality assessment agent for assessing the quality of the simulation simulation results based on the actual prediction results. The simulation module is used to input geological prediction models and physical world environmental data into the digital twin hybrid world model in the simulation agent according to the received sub-tasks, and to perform forward simulation to obtain simulation results. The autonomous update module is used to compare the simulation results with the actual formation data and actual drilling feedback data collected during drilling operations in the inversion model quality assessment agent based on the received sub-tasks, obtain the comparison results, and feed the comparison results back to the task planning agent. Based on the comparison results, the overall drilling task is further decomposed and returned to step 4 to realize the autonomous update of the geological prediction model until the update termination condition is met, and the final geological prediction model is output to make geological predictions for the un-drilled areas.

[0016] The present invention also provides a terminal device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement a method for autonomously updating a prediction model for physical world environment perception.

[0017] The above-described solution of the present invention has the following beneficial effects: This invention trains a constructed prediction model using physical world environmental data. A task planning agent breaks down the overall drilling task into multiple sub-tasks. Based on the received sub-tasks, the geological prediction model and physical world environmental data are input into a digital twin hybrid world model in the simulation agent for prospective simulation. Based on the received sub-tasks, the simulation results are compared with real formation data collected during drilling and actual drilling feedback data in the inversion model quality evaluation agent. The comparison results are fed back to the task planning agent. Based on the comparison results, the overall drilling task is further broken down, and the simulation and prediction results are compared again to achieve autonomous updating of the geological prediction model. This realizes a closed-loop linkage between virtual prospective simulation and real physical world feedback. Compared with existing black-box geological prediction models and single-agent methods lacking collaboration, this significantly improves geological prediction accuracy, thereby enhancing the safety, efficiency, and resource exploration accuracy of drilling operations.

[0018] Other beneficial effects of the present invention will be described in detail in the following detailed description section. Attached Figure Description

[0019] Figure 1 This is a flowchart illustrating an embodiment of the present invention; Figure 2 This is a schematic diagram of the prediction model in an embodiment of the present invention; Figure 3 This is a schematic diagram of the predictive model autonomous update device in an embodiment of the present invention; Figure 4 This is a schematic diagram of the structure of the terminal device in an embodiment of the present invention. Detailed Implementation

[0020] To make the technical problems, solutions, and advantages of this invention clearer, a detailed description will be provided below with reference to the accompanying drawings and specific embodiments. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0021] In the description of this invention, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing the invention and for simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0022] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a locking connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0023] Furthermore, the technical features involved in the different embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0024] This invention addresses existing problems by providing a method and related equipment for autonomously updating a prediction model for physical world environmental perception.

[0025] like Figure 1 As shown, embodiments of the present invention provide an autonomous update method for a prediction model of physical world environment perception, comprising: Step 1: Collect physical world environmental data during the drilling process. Physical world environmental data includes downhole image data of drilled formations, lithological and physical property data from logging while drilling, geological attribute data from advanced exploration, current wellbore trajectory data, and drilling engineering parameter data. Step 2: Train the constructed prediction model using physical world environmental data to obtain the geological prediction model; Step 3: Use the task planning agent to break down the overall drilling task into multiple sub-tasks, and distribute each sub-task to the simulation simulation agent for simulating the geological prediction model and the inversion simulation quality assessment agent for assessing the quality of the simulation simulation results based on the actual prediction results. Step 4: Based on the received sub-tasks, input the geological prediction model and physical world environmental data into the digital twin hybrid world model in the simulation agent to perform forward simulation and obtain the simulation results. Step 5: Based on the received sub-tasks, the simulation results are compared with the actual formation data and actual drilling feedback data collected during drilling operations in the inversion model quality assessment agent to obtain the comparison results. The comparison results are then fed back to the task planning agent. Based on the comparison results, the overall drilling task is further decomposed and returned to step 4 to achieve autonomous updating of the geological prediction model until the update termination condition is met. Finally, the geological prediction model is output to perform geological prediction for the un-drilled area.

[0026] In this embodiment of the invention, the lithological and physical property data of logging while drilling are obtained by core analysis of the raw logging curve data collected by multi-parameter sensors such as natural gamma, dual lateral resistivity, sonic transit time, bulk density, neutron porosity, and elemental trapping energy spectrum carried by the downhole logging while drilling system.

[0027] Specifically, before step 2, the following is also included: Preprocessing physical world environmental data yields basic environmental data.

[0028] In this embodiment of the invention, the preprocessing process includes spatiotemporal synchronization, noise filtering, semantic alignment, and standardization of all physical world environmental data to eliminate spatiotemporal deviations in data from different sensors.

[0029] Specifically, such as Figure 2 As shown, the prediction model includes: The input module is used to perform data standardization and feature embedding processing on physical world environmental data. The multi-scale convolutional encoder module is used to extract the lithology, physical properties, and geological body control distribution characteristics of strata at different depths from the data processed by the input module. A cross-scale semantic fusion module is used to fuse measured semantic features of drilled areas with advanced detection features of undrilled areas; The decoder module is used to output the geological prediction results for the un-drilled area. The geological prediction results include the continuous prediction results of the geological attributes, ore body distribution, and spatial location of risk bodies in the un-drilled area.

[0030] In this embodiment of the invention, the input module is used to perform data standardization and feature embedding processing on physical world environmental data. Specifically, it includes a data receiving unit, a data preprocessing unit, and a feature embedding unit. The data receiving unit is used to receive physical world environmental data, which includes at least downhole image data of drilled formations, lithological property data from logging while drilling, geological attribute data from advanced exploration, current wellbore trajectory data, and drilling engineering parameter data. The data preprocessing unit is used to perform missing value completion, outlier removal, time synchronization, depth alignment, and normalization processing on data from different sources, with different dimensions, and different sampling frequencies. The feature embedding unit is used to map the preprocessed data into a multi-channel feature tensor of a unified dimension for input into the multi-scale convolutional encoder module.

[0031] In this embodiment of the invention, the multi-scale convolutional encoder module includes multiple convolutional branches with different receptive fields, residual connection units, and feature convergence units. Each convolutional branch encodes the multi-channel feature tensor using different convolutional kernel sizes and / or different void ratios to extract lithological features, physical properties, and geological body control distribution features of strata at different depths. The residual connection units are used to preserve shallow details and enhance feature transfer. The feature convergence units are used to splice and compress the outputs of each convolutional branch to form a multi-scale strata representation.

[0032] The cross-scale semantic fusion module includes a semantic extraction unit for drilled areas, a feature extraction unit for advanced exploration in un-drilled areas, a scale alignment unit, and a semantic fusion unit. The semantic extraction unit for drilled areas is used to extract real stratigraphic semantic features from the measured data of drilled areas. The feature extraction unit for advanced exploration in un-drilled areas is used to extract potential geological semantic features of un-drilled areas from the geological attribute data of advanced exploration. The scale alignment unit is used to perform upsampling, downsampling, or interpolation mapping on features of different resolutions. The semantic fusion unit is used to fuse the measured semantic features of drilled areas and the advanced exploration features of un-drilled areas through attention weighting or gating fusion to form cross-scale fused features.

[0033] The decoder module includes a progressive upsampling unit, a feature refinement unit, and a prediction output unit. The progressive upsampling unit is used to recover the cross-scale fused features layer by layer. The feature refinement unit is used to correct the boundary details by combining the skip connection features from the encoding stage. The prediction output unit is used to output the geological prediction results for the un-drilled area. The geological prediction results include the geological attribute regression results of the un-drilled area, the probability field of ore body distribution, and the probability field of the spatial location of the risk body.

[0034] Furthermore, the data processing of the prediction model is as follows: the input module first performs unified preprocessing on the physical world environmental data and forms a standardized feature tensor; the multi-scale convolutional encoder module performs multi-scale encoding on the standardized feature tensor; the cross-scale semantic fusion module aligns and fuses the measured semantic features of the drilled area with the advanced detection features of the un-drilled area; and the decoder module then maps the fused features into continuous geological prediction results, thereby realizing the joint prediction of geological information in the un-drilled area.

[0035] Specifically, the digital twin hybrid world model includes: The 3D virtual simulation environment module is used to construct a 3D virtual drilling scene based on the preliminary exploration data of the target area and the real data of the drilled strata, so as to fully restore the equipment characteristics of the real drilling system and the geological laws and physical properties of the strata environment. The physical world real-time mapping module is used to establish a real-time data channel between the physical world and the virtual environment, and to synchronize the real formation data, drilling status data and engineering parameter data collected during drilling to the three-dimensional virtual drilling scene in real time, so as to achieve spatiotemporal alignment and state synchronization between the virtual scene and the real physical world. The forward-looking simulation module is used to perform forward-looking simulations of geological prediction models in a virtual environment and output simulation results.

[0036] In this embodiment of the invention, the forward simulation of the geological prediction model in a virtual environment includes simulating drilling responses, risk evolution, and stratigraphic changes under different trajectories and geological conditions.

[0037] Specifically, a three-dimensional virtual drilling scene is constructed based on preliminary exploration data of the target area and actual data of drilled strata, including: The preliminary exploration data and actual data of drilled strata in the target area are cleaned, denoised, standardized, and have their spatial coordinates unified. Based on the unified spatial coordinates, the formation interface, lithological property data, current wellbore trajectory data and drilling engineering parameter data are registered and fused to construct a three-dimensional geological structure model and wellbore model of the target area. The three-dimensional geological structure model, wellbore model, and drilling boundary conditions are input into the digital twin simulation engine to generate a three-dimensional virtual drilling scene corresponding to the actual drilling process, so as to provide a basic environment for subsequent forward-looking simulation.

[0038] In this embodiment of the invention, the preliminary exploration data includes regional geological data, seismic exploration data, geological mapping data, geophysical data, geochemical data, historical borehole data, well logging interpretation data, formation interface interpretation results, fault and structural distribution data, ore body or reservoir prediction data, etc., acquired before the formal drilling or the current drilling stage; the actual data of the drilled formations include downhole image data, core logging data, cuttings logging data, logging-while-drilling data, lithological property test data, wellbore trajectory data, drilling fluid parameters, drilling pressure, rotation speed, torque, mechanical drilling speed, pump pressure and other drilling engineering parameters, as well as actual drilling feedback data such as stuck pipe, lost circulation, wellbore instability, etc.

[0039] Specifically, before breaking down the overall drilling mission into multiple sub-tasks, it also includes: Build a visual interactive interface in the task planning agent; The geological features of the target area, on-site operational experience, and potential risk prediction results are input into the task planning agent through a visual interactive interface. By refining the task objectives of the task planning agent based on geological patterns, field operation experience, and potential risk predictions, the task scheduling of the agent can be guided to meet the actual needs of the project.

[0040] Specifically, before comparing the simulation results with the actual formation data collected during drilling and the actual drilling feedback data after the drilling operation is completed, the following steps are also included: The simulation results are manually verified to correct misjudgments by the AI, supplement professional geological interpretations, and improve the accuracy of model evaluation.

[0041] In this embodiment of the invention, the inversion model quality assessment agent is responsible for receiving the simulation results output by the simulation and the real geological data and drilling feedback data collected during drilling in the real physical world. By comparing the simulation results with the data collected during drilling in the real physical world through multi-dimensional indicators, the agent quantitatively assesses the prediction quality of the current geological prediction model, thereby generating the optimization direction and constraints for model updates.

[0042] Specifically, when the comparison result exceeds the preset threshold, the geological prediction model is fine-tuned and optimized using incremental learning and real formation data collected during drilling operations and actual drilling feedback data.

[0043] In this embodiment of the invention, an experience replay library is also built, which standardizes and stores all the real geological data, model update data and agent decision data in each drilling process into the experience replay library, thereby reproducing the training and updating process of the geological prediction model.

[0044] The embodiments of this invention comprehensively verify the feasibility, stability, and superiority of this method through simulation tests and field pilot tests in deep shale gas drilling and deep metal mine drilling. The specific verification results are as follows: Simulation Verification: Based on real geological data of a deep shale gas block in a basin, a high-fidelity virtual drilling simulation environment was constructed. Various complex geological risk scenarios, such as faults, high-pressure layers, karst caves, and fracture zones, were set up. Multiple sets of comparative tests were conducted with traditional manual planning methods, conventional reinforcement learning path planning methods, and fixed offline geological prediction models. The comparison results of its core performance indicators are shown in Table 1 below. Table 1. Comparison of core performance indicators under simulation.

[0045] Test results show that the method provided in this embodiment of the invention achieves a 100% success rate in avoiding unknown risk areas and reduces the average drilling footage by 15.8%. The prediction accuracy of the geological prediction model is 41.2% higher than that of the traditional offline model, and the response time for the model's autonomous update is less than 5 minutes, fully meeting the engineering requirements for real-time decision-making while drilling.

[0046] Field Test: A pilot test was conducted at a scientific drilling site at a depth of 2000 meters in a metal mine in a certain region. The geological conditions of the test block were complex, with multiple unexplored fault fracture zones and abrupt changes in stratum dip angle. The key engineering indicators were compared as shown in Table 2 below: Table 2 Comparison of Key Engineering Indicators between Field Tests and Adjacent Wells in the Same Block

[0047] During the experiment, the embodiment of the present invention achieved real-time autonomous adjustment of the drilling path. According to Table 2, the actual total drilling cycle of the test well in this embodiment was 38 days, a 22.4% reduction compared to the 49 days of adjacent well 1 in the same block, and a 25.5% reduction compared to the 51 days of adjacent well 2 in the same block. There was one drilling accident (minor stuck pipe), an 80% reduction compared to the 5 accidents of adjacent well 1 in the same block, and a 75% reduction compared to the 4 accidents of adjacent well 2 in the same block. Through continuous autonomous updates of on-site drilling data, the geological prediction model achieved an 89% accuracy rate in predicting deep orebody boundaries, a 100% trajectory qualification rate across the entire well section, and an average mechanical drilling speed of 8.2 m / h, providing precise geological support for subsequent resource reserve assessment and exploration and development.

[0048] This invention utilizes physical world environmental data to train a constructed prediction model. A task planning agent breaks down the overall drilling task into multiple sub-tasks. Based on the received sub-tasks, the geological prediction model and physical world environmental data are input into a digital twin hybrid world model within the simulation agent for prospective simulation. Based on the received sub-tasks, the simulation results are compared with real formation data collected during drilling and actual drilling feedback data in the inversion model quality evaluation agent. The comparison results are then fed back to the task planning agent. Based on the comparison results, the overall drilling task is further broken down, and the simulation and prediction results are compared again to achieve autonomous updating of the geological prediction model. This realizes a closed-loop linkage between virtual prospective simulation and real physical world feedback. Compared to existing black-box geological prediction models and single-agent methods lacking collaboration, this invention significantly improves geological prediction accuracy, thereby enhancing the safety, efficiency, and resource exploration accuracy of drilling operations.

[0049] like Figure 3 As shown, this embodiment of the invention also provides a predictive model autonomous update device 100 for physical world environment perception, which applies a predictive model autonomous update method. The predictive model autonomous update device 100 includes: The acquisition module 101 is used to acquire physical world environmental data during the drilling process. The physical world environmental data includes downhole image data of drilled formations, lithological and physical property data from logging while drilling, geological attribute data from advanced exploration, current wellbore trajectory data, and drilling engineering parameter data. Training module 102 is used to train the constructed prediction model using physical world environmental data to obtain a geological prediction model; The decomposition module 103 is used to decompose the overall drilling task into multiple sub-tasks using the task planning agent, and to send each sub-task to the simulation simulation agent for simulating the geological prediction model and the inversion simulation quality assessment agent for assessing the quality of the simulation simulation results based on the actual prediction results. The simulation module 104 is used to input the geological prediction model and physical world environment data into the digital twin hybrid world model in the simulation and inference agent according to the received sub-tasks, and to perform forward-looking inference simulation to obtain the simulation and inference results. The autonomous update module 105 is used to compare the simulation results with the real formation data and actual drilling feedback data collected during drilling operations in the inversion model quality assessment agent according to the received sub-tasks, obtain the comparison results, and feed the comparison results back to the task planning agent. Based on the comparison results, the overall drilling task is further decomposed and returned to step 4 to realize the autonomous update of the geological prediction model until the update termination condition is met, and the final geological prediction model is output to make geological predictions for the un-drilled areas.

[0050] It should be noted that the information interaction and execution process between the above-mentioned devices / units are based on the same concept as the method embodiments of this application. For details on their specific functions and technical effects, please refer to the method embodiments section, and they will not be repeated here.

[0051] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments 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. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0052] This invention also provides a terminal device, such as... Figure 4 As shown, the terminal device D10 of this embodiment includes: at least one processor D100 ( Figure 4 The diagram shows only one processor, a memory D101, and a computer program D102 stored in the memory D101 and executable on the at least one processor D100. When the processor D100 executes the computer program D102, it implements the above-described method for autonomously updating the prediction model of physical world environment perception.

[0053] The terminal device D10 can be a desktop computer, laptop, handheld computer, server, server cluster, or cloud server, etc. This terminal device may include, but is not limited to, a processor D100 and a memory D101. Those skilled in the art will understand that... Figure 4 This is merely an example of terminal device D10 and does not constitute a limitation on terminal device D10. It may include more or fewer components than shown in the figure, or combine certain components, or different components, such as input / output devices, network access devices, etc.

[0054] The processor D100 can be a central processing unit (CPU), or it can be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor.

[0055] In some embodiments, the memory D101 may be an internal storage unit of the terminal device D10, such as a hard disk or memory of the terminal device D10. In other embodiments, the memory D101 may be an external storage device of the terminal device D10, such as a plug-in hard disk, smart media card (SMC), secure digital card (SD), flash card, etc., equipped on the terminal device D10. Furthermore, the memory D101 may include both internal and external storage units of the terminal device D10. The memory D101 is used to store the operating system, applications, bootloader, data, and other programs, such as the program code of the computer program. The memory D101 can also be used to temporarily store data that has been output or will be output.

[0056] It should be noted that the information interaction and execution process between the above-mentioned devices / units are based on the same concept as the method embodiments of this application. For details on their specific functions and technical effects, please refer to the method embodiments section, and they will not be repeated here.

[0057] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments 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. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0058] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for autonomously updating a predictive model of physical world environmental perception, characterized in that, include: Step 1: Collect physical world environmental data during the drilling process. The physical world environmental data includes downhole image data of drilled formations, lithological and physical property data from logging while drilling, geological attribute data from advanced exploration, current wellbore trajectory data, and drilling engineering parameter data. Step 2: Use the physical world environmental data to train the constructed prediction model to obtain the geological prediction model; Step 3: Use the task planning agent to break down the overall drilling task into multiple sub-tasks, and distribute each sub-task to the simulation simulation agent for simulating the geological prediction model and the inversion simulation quality assessment agent for assessing the quality of the simulation simulation results based on the actual prediction results. Step 4: Based on the received sub-tasks, input the geological prediction model and the physical world environment data into the digital twin hybrid world model in the simulation and inference agent to perform forward-looking simulation and inference, and obtain the simulation and inference results; Step 5: Based on the received sub-task, the simulation results are compared with the actual formation data and actual drilling feedback data collected during drilling operations in the inversion model quality assessment agent to obtain the comparison results. The comparison results are then fed back to the task planning agent. Based on the comparison results, the overall drilling task is further decomposed, and the process returns to step 4 to achieve autonomous updating of the geological prediction model until the update termination condition is met. Finally, the geological prediction model is output to perform geological prediction for the un-drilled area.

2. The method for autonomously updating the prediction model of physical world environment perception according to claim 1, characterized in that, Before step 2, the following is also included: The physical world environmental data is preprocessed to obtain basic environmental data.

3. The method for autonomously updating the prediction model of physical world environment perception according to claim 1, characterized in that, The prediction model includes: The input module is used to perform data standardization and feature embedding processing on physical world environmental data. A multi-scale convolutional encoder module is used to extract lithology, physical properties, and geological body control distribution characteristics of strata at different depths from the data processed by the input module. A cross-scale semantic fusion module is used to fuse measured semantic features of drilled areas with advanced detection features of undrilled areas; The decoder module is used to output geological prediction results for un-drilled areas.

4. The method for autonomously updating the prediction model of physical world environment perception according to claim 1, characterized in that, The digital twin hybrid world model includes: The 3D virtual simulation environment module is used to construct a 3D virtual drilling scene based on the preliminary exploration data of the target area and the real data of the drilled strata. The physical world real-time mapping module is used to synchronize real formation data, drilling status data, and engineering parameter data collected during drilling to the three-dimensional virtual drilling scene in real time; The forward-looking simulation module is used to perform forward-looking simulations of the geological prediction model in a virtual environment and output the simulation results.

5. The method for autonomously updating the prediction model of physical world environment perception according to claim 4, characterized in that, A three-dimensional virtual drilling scene is constructed based on the preliminary exploration data of the target area and the actual data of the drilled strata, including: The preliminary exploration data and actual data of drilled strata in the target area are cleaned, denoised, standardized, and have their spatial coordinates unified. Based on the unified spatial coordinates, the formation interface, the lithological property data, the current wellbore trajectory data and the drilling engineering parameter data are registered and fused to construct a three-dimensional geological structure model and wellbore model of the target area. The three-dimensional geological structure model, the wellbore model, and the drilling boundary conditions are input into the digital twin simulation engine to generate a three-dimensional virtual drilling scene corresponding to the actual drilling process.

6. The method for autonomously updating the prediction model of physical world environment perception according to claim 1, characterized in that, Before breaking down the overall drilling mission into multiple sub-missions, it also includes: A visual interactive interface is built in the task planning agent; The geological features of the target area, on-site operational experience, and potential risk prediction results are input into the task planning agent through the visual interactive interface. The task objectives of the task planning agent are revised based on the geological patterns, field operation experience, and potential risk prediction results.

7. The method for autonomously updating the prediction model of physical world environment perception according to claim 1, characterized in that, Before comparing the simulation results with the actual formation data collected during drilling and the actual drilling feedback data after the drilling operation is completed, the following steps are also included: The simulation results are manually verified.

8. The method for autonomously updating the prediction model of physical world environment perception according to claim 1, characterized in that, When the comparison result is greater than the preset threshold, the geological prediction model is fine-tuned and optimized by using incremental learning and real formation data collected during drilling after the drilling operation is completed, as well as actual drilling feedback data.

9. A device for autonomously updating a predictive model of physical world environmental perception, characterized in that, The predictive model autonomous update method according to any one of claims 1-8, wherein the predictive model autonomous update device comprises: The acquisition module is used to acquire physical world environmental data during the drilling process. The physical world environmental data includes downhole image data of drilled formations, lithological and physical property data from logging while drilling, geological attribute data from advanced exploration, current wellbore trajectory data, and drilling engineering parameter data. The training module is used to train the constructed prediction model using the physical world environmental data to obtain a geological prediction model. The decomposition module is used to decompose the overall drilling task into multiple sub-tasks using the task planning agent, and to distribute each sub-task to the simulation simulation agent for simulating the geological prediction model and the inversion simulation quality assessment agent for assessing the quality of the simulation simulation results based on the actual prediction results. The simulation module is used to input the geological prediction model and the physical world environment data into the digital twin hybrid world model in the simulation and inference agent according to the received sub-tasks to perform forward-looking inference simulation and obtain simulation results. The autonomous update module is used to compare the simulation results with the actual formation data and actual drilling feedback data collected during drilling operations in the inversion model quality assessment agent according to the received sub-tasks, obtain the comparison results, and feed the comparison results back to the task planning agent. Based on the comparison results, the overall drilling task is further decomposed, and the process returns to step 4 to realize the autonomous update of the geological prediction model until the update termination condition is met, and the final geological prediction model is output to make geological predictions for the un-drilled areas.

10. A terminal device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method for autonomously updating the prediction model of physical world environment perception as described in any one of claims 1 to 8.

Images