Formation pressure while drilling early warning method and device based on model prediction

Abstract

The invention provides a while-drilling formation pressure early warning method and device based on model prediction, equipment and a medium, and relates to the technical field of low-carbon mining, and the method comprises the steps: in a pre-drilling stage, based on seismic data, adjacent well historical data and design data, analyzing and generating prediction risk data of a target monitoring well; in the drilling process, actual formation pressure data are calculated in real time by using drilling data collected for a target monitoring well in real time, and the actual formation pressure data are compared with the predicted risk data; in the drilling stage, a pressure safety window is depicted based on the predicted risk data, the cycle equivalent density of the target monitoring well calculated in real time is compared with the safety window, and the drilling safety risk is recognized and analyzed; and after drilling is finished, summarizing the predicted risk data, the actual formation pressure data and the drilling safety risk, and updating and iterating the model algorithm through the summarized data. The method has the effect of improving the accuracy of formation pressure early warning while drilling in the petroleum drilling process.

Description

Technical Field

[0001] This application relates to the technical field of low-carbon mining, and particularly relates to a method and device for predicting formation pressure while drilling based on model prediction. Background Art

[0002] The early warning and monitoring of formation pressure while drilling an oil well effectively reduces the incidence of accidents such as lost circulation and overflow by real-time identifying abnormal pressure zones and precisely controlling the downhole pressure state, significantly reduces the energy waste related to unplanned shutdowns, drilling fluid losses, and carbon emissions, and at the same time reduces the repeated operations and equipment operation loads brought about by accident handling. While ensuring operation safety, it improves the drilling operation efficiency and resource utilization rate, and is an important technical support for realizing the intelligent and green transformation of the oil exploration and development process and promoting low-carbon mining.

[0003] Traditional methods do not have an automated data processing process. During the analysis process, data needs to be manually collected and imported into desktop software for pressure data calculation, and continuous and real-time monitoring of formation pressure cannot be achieved. Pressure analysis requires engineers to make judgments based on personal experience, without a unified standard for reference, and the accuracy of early warning cannot be guaranteed. Based on the integrated promotion of integrated business work, single and professional early warnings cannot meet business needs, and the analysis results of multiple specialties need to be comprehensively considered for auxiliary judgment to improve the accuracy of early warning. Therefore, a method is needed to improve the accuracy of predicting formation pressure while drilling during the oil drilling process. Summary of the Invention

[0004] This application provides a method and device for predicting formation pressure while drilling based on model prediction, which has the effect of improving the accuracy of predicting formation pressure while drilling during the oil drilling process.

[0005] In the first aspect of this application, a method for predicting formation pressure while drilling based on model prediction is provided. The method includes: Obtaining seismic data, adjacent well historical data, and design data for a target monitoring well; In the pre-drilling stage, based on the seismic data, the adjacent well historical data, and the design data, analyzing and generating prediction risk data for the target monitoring well; During the drilling process, using the drilling data collected in real time for the target monitoring well, calculating the actual formation pressure data in real time, and comparing the actual formation pressure data with the prediction risk data; In the middle-drilling stage, based on the prediction risk data, depicting a pressure safety window, and calculating the circulating equivalent density of the target monitoring well in real time, comparing the circulating equivalent density with the safety window, and identifying and analyzing drilling safety risks; After the drilling is completed, the predicted risk data, the actual formation pressure data, and the drilling safety risks are summarized, and the model algorithm is updated and iterated based on the summarized data.

[0006] Based on the above technical solutions, preferably, before analyzing and generating the predicted risk data of the target monitoring well based on the seismic data, the historical data of adjacent wells, and the design data in the pre-drilling stage, the method further includes: Obtain the seismic data of the area where the target monitoring well is located, and the seismic data includes seismic profiles, seismic layer velocities, and seismic attribute volumes; Model the structural form, fault development, and formation continuity of the regional formation through the seismic data to obtain a seismic model; Analyze the formation pressure change characteristics of each adjacent well through the historical data of adjacent wells using machine learning algorithms, and establish a formation pressure gradient model. The historical data of adjacent wells includes adjacent well logging data, adjacent well mud logging data, adjacent well drilling parameter data, and adjacent well formation pressure data; Compare the formation pressure abnormal points of different adjacent wells through the formation pressure gradient model, and summarize the distribution data of abnormal pressures. The distribution data includes distribution patterns and influencing factors; Based on the seismic model and the distribution data, combined with the geological structure characteristics of the area where the target monitoring well is located, adopt a well-seismic joint analysis method to predict the formation pressure distribution at the well location through seismic data to obtain predicted formation pressure data.

[0007] Based on the above technical solutions, preferably, in the pre-drilling stage, analyzing and generating the predicted risk data of the target monitoring well based on the seismic data, the historical data of adjacent wells, and the design data specifically includes: Based on the predicted formation pressure data, combined with the design data of the target monitoring well, calculate the predicted formation pressure gradient of the target monitoring well. The design data includes well trajectory design, drilling fluid parameters, bit and drill string combination, and wellbore structure design; Correct and optimize the predicted formation pressure gradient through the Eaton method, and combined with the historical accident data of adjacent wells, compare and analyze the risk data of well leakage, overflow, or well collapse occurring in the target monitoring well to generate the predicted risk data.

[0008] Based on the above technical solutions, preferably, during the drilling process, use the drilling data collected in real time for the target monitoring well to calculate the actual formation pressure data in real time, and compare the actual formation pressure data with the predicted risk data. Specifically includes: Continuously acquiring the drilling data according to a real-time data acquisition system deployed in the target monitoring well, wherein the drilling data includes drilling rate, rotation speed, bit pressure, and drilling fluid density; The drilling index is calculated based on the drilling data, specifically using the following formula:

[0009] Wherein, DC is the drilling index, R is the drilling speed, N is the rotation speed, and W is the bit weight; Based on the current real-time drilling index, combined with the drilling fluid density and conventional formation pressure gradient, the formation pressure gradient is estimated using the modified drilling index method. The specific calculation is as follows:

[0010] in, is the normal formation pressure gradient, DC obs The real-time drilling index of the current drilling, DC n is the reference drilling index of adjacent wells or historical formations under normal pressure, and β is the empirical adjustment coefficient; The formation pressure gradient is integrated to obtain the actual formation pressure data.

[0011] Based on the above technical solution, preferably, during the drilling process, actual formation pressure data is calculated in real time using the drilling data collected in real time from the target monitoring well, and the actual formation pressure data is compared with the predicted risk data, which specifically further includes: Aligning the actual formation pressure data with the depth interval corresponding to the predicted formation pressure data corresponding to the predicted risk data in depth sequence, and performing layer calibration; Performing point-to-point calculation on the aligned predicted risk data and the actual formation pressure data using a residual calculation method to calculate the absolute value of the residual; Determine the relationship between the absolute value of the residual and the preset threshold. If it is determined that the absolute value of the residual is greater than or equal to the preset threshold, analyze the pressure changes of the target monitoring well in combination with the predicted formation pressure gradient of the target monitoring well corresponding to the predicted risk data.

[0012] Based on the above technical solution, preferably, during the drilling stage, a pressure safety window is depicted based on the predicted risk data, and the circulating equivalent density of the target monitoring well is calculated in real time. The circulating equivalent density is compared with the safety window to identify and analyze the drilling safety risk, specifically including: Constructing a pressure safety window for well section distribution based on formation pore pressure, fracture pressure and collapse pressure data included in the predicted risk data; Calculate the equivalent static pressure at the bottom of the drill string in real time according to the drilling fluid density and the current well depth; Make a preliminary comparison and judgment between the equivalent static pressure and the pressure safety window. If it is determined that the equivalent static pressure is within the pressure safety window, use a dynamic hydraulics model to calculate the circulating equivalent density; Align the continuous curve constructed according to the circulating equivalent density with the pressure safety window in depth, and determine the drilling safety risk according to the relative size relationship between the continuous curve and the pressure safety window.

[0013] On the basis of the above technical solutions, preferably, after the drilling is completed, summarize the predicted risk data, the actual formation pressure data, and the drilling safety risk, and update and iterate the model algorithm through the summarized data, specifically including: Perform well depth consistency correction on the predicted risk data, the actual formation pressure data, and the drilling safety risk through a depth registration algorithm, organize them into a standardized format according to the horizon structure, and file the predicted risk data, the actual formation pressure data, and the drilling safety risk of each well section to form a complete risk assessment data set; Extract the deformation characteristics of the ECD curve, the fluctuation pattern of drilling parameters, and the corresponding response feedback data in the abnormal section as the input of the feature vector of the training sample; Use the key parameters adopted before the end of drilling as variables to be optimized, construct a loss function, and aim to minimize the weighted mean square error of the whole well section between the predicted pressure and the measured pressure. The key parameters include the drilling index gradient model coefficient, the Eaton method index value, and the adjacent well reference curve weight factor; For the prediction model constructed based on machine learning, use the risk assessment data set as an extended sample for incremental learning to absorb the new feature changes in the actual drilling process.

[0014] In the second aspect of the present application, a formation pressure early warning device while drilling based on model prediction is provided. The device is used to execute a formation pressure early warning method while drilling based on model prediction as described in any one of the above. The device includes an acquisition module, a processing module, and an output module, wherein: The acquisition module is used to acquire seismic data, adjacent well historical data, and design data for the target monitoring well; The processing module is used to analyze and generate predicted risk data for the target monitoring well based on the seismic data, the adjacent well historical data, and the design data in the pre-drilling stage; The processing module is used to, during the drilling process, utilize the drilling data collected in real time for the target monitoring well to calculate the actual formation pressure data in real time, and compare the actual formation pressure data with the predicted risk data; The processing module is used to, during the middle stage of drilling, characterize the pressure safety window based on the predicted risk data, calculate the circulating equivalent density of the target monitoring well in real time, compare the circulating equivalent density with the safety window, and identify and analyze the drilling safety risks; The output module is used to, after the drilling is completed, summarize the predicted risk data, the actual formation pressure data, and the drilling safety risks, and update and iterate the model algorithm through the summarized data.

[0015] Based on the above technical solutions, preferably, the acquisition module is used to acquire seismic data of the area where the target monitoring well is located, and the seismic data includes seismic profiles, seismic layer velocities, and seismic attribute bodies; The processing module is used to model the structural morphology, fault development, and formation continuity of the regional formation through the seismic data to obtain a seismic model; The processing module is used to analyze the formation pressure change characteristics of each adjacent well during drilling through the adjacent well historical data using machine learning algorithms, and establish a formation pressure gradient model. The adjacent well historical data includes adjacent well logging data, adjacent well mud logging data, adjacent well drilling parameter data, and adjacent well formation pressure data; The processing module is used to compare the formation pressure abnormal points of different adjacent wells through the formation pressure gradient model, and summarize the distribution data of abnormal pressures. The distribution data includes distribution patterns and influencing factors; The output module is used to, based on the seismic model and the distribution data, combine the geological structure characteristics of the area where the target monitoring well is located, adopt a well-seismic joint analysis method, and predict the formation pressure distribution at the well location through seismic data to obtain predicted formation pressure data.

[0016] Based on the above technical solutions, preferably, the processing module is used to calculate the predicted formation pressure gradient of the target monitoring well based on the predicted formation pressure data and in combination with the design data of the target monitoring well. The design data includes well trajectory design, drilling fluid parameters, bit and drill string assembly, and wellbore structure design; The output module is used to correct and optimize the predicted formation pressure gradient through the Eaton method, and in combination with the adjacent well historical accident data, compare and analyze the risk data of well leakage, overflow, or well collapse occurring in the target monitoring well, and generate the predicted risk data.

[0017] Based on the above technical solutions, preferably, the processing module is used to continuously obtain the drilling data according to the real-time data acquisition system deployed in the target monitoring well, where the drilling data includes the drilling rate, rotation speed, drilling pressure, and drilling fluid density; The processing module is used to calculate the drilling index according to the drilling data, and specifically calculate it through the following formula:

[0018] where DC is the drilling index, R is the drilling rate, N is the rotation speed, and W is the drilling pressure; The processing module is used to estimate the formation pressure gradient by the corrected drilling index method according to the real-time drilling index of the current drilling, in combination with the drilling fluid density and the conventional formation pressure gradient, and specifically calculate it through the following formula:

[0019] where, is the normal formation pressure gradient, DC obs is the real-time drilling index of the current drilling, DC n is the reference drilling index of the adjacent well or historical formation under normal pressure, and β is the empirical adjustment coefficient; The processing module is used to integrate the formation pressure gradient to obtain the actual formation pressure data.

[0020] Based on the above technical solutions, preferably, the processing module is used to align the actual formation pressure data in depth sequence with the depth interval corresponding to the predicted formation pressure data corresponding to the predicted risk data, and perform horizon calibration; The processing module is used to perform point-to-point calculation on the aligned predicted risk data and the actual formation pressure data by using the residual calculation method, and calculate the absolute value of the residual; The processing module is used to judge the magnitude relationship between the absolute value of the residual and the preset threshold. If it is judged that the absolute value of the residual is greater than or equal to the preset threshold, then analyze the pressure change of the target monitoring well in combination with the predicted formation pressure gradient of the target monitoring well corresponding to the predicted risk data.

[0021] Based on the above technical solutions, preferably, the processing module is used to construct a pressure safety window for the well section distribution based on the formation pore pressure, fracture pressure, and collapse pressure data included in the predicted risk data; The processing module is used to calculate the equivalent static pressure at the bottom of the drill string in real time according to the drilling fluid density and the current well depth; The processing module is configured to preliminarily compare and judge the equivalent static pressure with the pressure safety window. If it is determined that the equivalent static pressure is within the pressure safety window, a dynamic hydraulics model is used to calculate the circulating equivalent density; The processing module is configured to deeply align the continuous curve constructed according to the circulating equivalent density with the pressure safety window, and determine the drilling safety risk according to the relative size relationship between the continuous curve and the pressure safety window.

[0022] Based on the above technical solutions, preferably, the processing module is configured to perform well depth consistency correction on the predicted risk data, the actual formation pressure data, and the drilling safety risk through a depth registration algorithm, organize them into a standardized format according to the horizon structure, and file the predicted risk data, the actual formation pressure data, and the drilling safety risk of each well section to form a complete risk assessment data set; The processing module is configured to extract the ECD curve deformation characteristics, the drilling parameter fluctuation patterns, and the corresponding response feedback data in the abnormal section as the feature vectors of the training samples for input; The processing module is configured to use the key parameters adopted before the end of drilling as variables to be optimized, construct a loss function, and aim to minimize the weighted mean square error of the whole well section between the predicted pressure and the measured pressure. The key parameters include the drilling index gradient model coefficient, the Eaton method index value, and the adjacent well reference curve weight factor; The output module is configured to perform incremental learning on the prediction model constructed based on machine learning using the risk assessment data set as an extended sample to absorb the new feature changes in the actual drilling process.

[0023] In a third aspect of the present application, an electronic device is provided, including a processor, a memory, a user interface, and a network interface. The memory is used to store instructions. Both the user interface and the network interface are used to communicate with other devices. The processor is used to execute the instructions stored in the memory so that the electronic device executes the method described in any one of the above.

[0024] In a fourth aspect of the present application, a computer-readable storage medium is provided. The computer-readable storage medium stores instructions, and when the instructions are executed, the method described in any one of the above is executed.

[0025] In summary, one or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages: 1. This application realizes the highly coupled prediction, monitoring and feedback by constructing a multi-source data fusion mechanism and a model closed-loop iterative system throughout the whole process of pre-drilling, drilling and post-drilling, breaks through the limitations of traditional reliance on single data and static judgment, uses seismic, offset well and design data to jointly establish a high-precision risk prediction model, dynamically calculates the actual pressure through real-time drilling data during the drilling process and compares it with the prediction results, and intelligently identifies drilling risks based on the real-time correlation analysis of the safety window and the circulating equivalent density. In the post-drilling stage, a feedback correction mechanism is introduced to perform incremental updates on the model while maintaining its structure, thus significantly improving the comprehensive performance of the formation pressure early warning while drilling in terms of timeliness, accuracy and adaptability.

[0026] 2. By introducing steps of seismic modeling and excavation of abnormal pressure laws in offset wells before generating prediction risk data, this application fully integrates multi-source geological information such as seismic profiles, interval velocities and attribute volumes, constructs a formation pressure gradient model by combining multi-dimensional historical data of offset wells, summarizes the distribution patterns and formation mechanisms of abnormal pressures, and accurately predicts the formation pressure distribution of the target monitoring well based on well-seismic joint analysis, so as to achieve high-resolution and high-reliability risk identification in the pre-drilling stage, provide a more targeted and regionally adaptable initial judgment basis for early warning while drilling, and significantly improve the accuracy of subsequent formation pressure early warning.

[0027] 3. By introducing a mechanism for depth alignment and horizon calibration of actual formation pressure and prediction data during the drilling process, combined with point-to-point absolute residual value calculation and threshold judgment, the dynamic quantitative identification of formation pressure deviation is realized, and multi-dimensional pressure evolution analysis is carried out in combination with the predicted pressure gradient under the condition of exceeding the threshold, thus significantly improving the sensitivity of perception and response timeliness to abnormal pressure changes during the drilling process, effectively enhancing the accuracy of early warning judgment, the pertinence of risk identification and the safety control ability during the operation process.

[0028] 4. By constructing a pressure safety window including pore pressure, fracture pressure and collapse pressure, and introducing a phased judgment mechanism for equivalent static pressure and circulating equivalent density, the real-time dynamic assessment of the pressure control state of the target well section is realized. Combining the depth alignment and deviation judgment of the circulating equivalent density curve and the safety window, potential well leakage, well collapse or overflow risks during the drilling process can be accurately identified, and the real-time performance of pressure early warning and the refinement degree of judgment in the drilling stage are significantly improved.

[0029] 5. After the drilling is completed, this application introduces a mechanism for multi-source data depth registration, feature extraction, and model reconstruction to construct a risk assessment data set with consistent well depths and clear horizons, refine the key drilling response features of abnormal sections, and construct a residual-driven loss function with historical key parameters as the optimization object. Combining the incremental learning strategy of the machine learning model, it realizes the precise correction and adaptive evolution of the prediction model, thereby significantly improving the generalization ability and warning accuracy of the model under the conditions of new wells and new areas, and constructs an intelligent formation pressure warning system with the ability of continuous learning and dynamic optimization. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 FIG. is a schematic flow chart of a method for predicting formation pressure while drilling based on model prediction disclosed in an embodiment of this application; Figure 2 FIG. is a schematic diagram of a pressure safety window; Figure 3 FIG. is a schematic block diagram of a device for predicting formation pressure while drilling based on model prediction disclosed in an embodiment of this application; Figure 4 FIG. is a schematic structural diagram of an electronic device.

[0031] Description of the reference numerals: 301, acquisition module; 302, processing module; 303, output module; 401, processor; 402, communication bus; 403, user interface; 404, network interface; 405, memory. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0032] In order to enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of this specification. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all of the embodiments.

[0033] In the description of the embodiments of this application, words such as "for example" or "for illustration" are used to indicate examples, illustrations, or explanations. Any embodiment or design solution described as "for example" or "for illustration" in the embodiments of this application should not be construed as being more preferred or having more advantages than other embodiments or design solutions. Exactly speaking, the use of words such as "for example" or "for illustration" is intended to present relevant concepts in a specific manner.

[0034] In the description of the embodiments of the present application, the term "plurality" means two or more. For example, a plurality of systems means two or more systems, and a plurality of screen terminals means two or more screen terminals. In addition, the terms "first" and "second" are only used for descriptive purposes and cannot be construed as indicating or implying relative importance or implicitly specifying the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include one or more of such features. The terms "comprising", "including", "having" and their variants all mean "including but not limited to", unless otherwise specifically emphasized in other ways.

[0035] To improve the accuracy of real-time formation pressure warning during oil drilling, it is necessary to break through the limitations of traditional methods that rely on manual experience and scattered data processing, and construct an intelligent warning method with automated processing capabilities and a multi-source data fusion mechanism to achieve continuous real-time monitoring and accurate identification of formation pressure, reduce the occurrence rate of risk events such as well loss and overflow, improve operation efficiency and resource utilization rate, and promote the transformation of oil exploration and development towards intelligence and low carbon.

[0036] This embodiment discloses a real-time formation pressure warning method during drilling based on model prediction, referring to Figure 1 , and includes the following steps S110 - S150: S110, obtain seismic data, historical data of adjacent wells, and design data for the target monitoring well.

[0037] A real-time formation pressure warning method during drilling based on model prediction disclosed in the embodiments of the present application is applied to a server. The server includes, but is not limited to, electronic devices such as mobile phones, tablets, wearable devices, PCs (Personal Computers), etc., and can also be a background server running a real-time formation pressure warning method during drilling based on model prediction. The server can be implemented by an independent server or a server cluster composed of multiple servers.

[0038] When obtaining seismic data for the target monitoring well, first, based on the three-dimensional seismic exploration results of the target block, retrieve the standardized seismic profile of the area, extract the well profile along the well of the target well design trajectory in the seismic volume, accurately extract seismic layer velocity information through well-seismic joint analysis technology, and simultaneously obtain parameter body data of seismic attribute bodies such as amplitude, frequency, and wave impedance, so as to provide the attribute input required for subsequent formation structure identification, velocity modeling, and pressure prediction, and achieve the full-scale acquisition of seismic data in spatial and attribute dimensions.

[0039] When obtaining the historical data of adjacent wells, it is necessary to screen the adjacent wells within a certain radius around the target monitoring well, collect and standardize the logging data of each adjacent well, such as density curve, acoustic travel time, natural gamma, etc., the mud logging data, such as cuttings description, gas logging anomaly, oil and gas show record, etc., the drilling parameters, such as weight on bit, rotary speed, displacement, standpipe pressure and other engineering data, as well as the measured or predicted curves of formation pressure obtained from each adjacent well. By unifying the data format, eliminating data missing and anomalies, and establishing the corresponding horizon relationship between wells, a sample data set with spatio-temporal consistency is provided for subsequent multi-well comparative analysis and model training.

[0040] During the process of obtaining the design data, it is necessary to extract the three-dimensional spatial information of the well trajectory from the target well design document, including parameters such as vertical depth, displacement, well inclination, azimuth, etc., in order to construct a wellbore spatial model; synchronously retrieve the designed drilling fluid performance parameters, including density, viscosity, water loss, type of weighting agent, etc., for subsequent calculation of pressure window and circulating equivalent density; obtain the bit parameters such as type, size, number of teeth and the matching drill string assembly, including the configuration information of drill collars, stabilizers, heavy drill pipes, etc., to assist in simulating downhole mechanical and hydraulic behaviors; finally, it is also necessary to obtain the wellbore structure design, such as casing program, well diameter distribution, etc., to establish the formation exposure conditions and mechanical boundaries during the drilling process, ensuring the integrity and rationality of the model simulation.

[0041] S120. In the pre-drilling stage, based on the seismic data, adjacent well historical data, and design data, analyze and generate the predicted risk data of the target monitoring well.

[0042] In a possible implementation manner, before analyzing and generating the predicted risk data of the target monitoring well based on the seismic data, the adjacent well historical data, and the design data in the pre-drilling stage, the method further includes: obtaining the seismic data of the area where the target monitoring well is located, where the seismic data includes seismic profiles, seismic layer velocities, and seismic attribute volumes; modeling the structural morphology, fault development, and formation continuity of the regional formation through the seismic data to obtain a seismic model; analyzing the formation pressure change characteristics of each adjacent well during drilling using a machine learning algorithm through the adjacent well historical data, and establishing a formation pressure gradient model, where the adjacent well historical data includes adjacent well logging data, adjacent well mud logging data, adjacent well drilling parameter data, and adjacent well formation pressure data; comparing the formation pressure anomaly points of different adjacent wells through the formation pressure gradient model, and summarizing the distribution data of abnormal pressures, where the distribution data includes distribution patterns and influencing factors; based on the seismic model and the distribution data, combined with the geological structure characteristics of the area where the target monitoring well is located, using a well-seismic joint analysis method, predicting the formation pressure distribution at the well location through the seismic data to obtain the predicted formation pressure data.

[0043] Specifically, when obtaining seismic data in the area where the target monitoring well is located, first extract the well profile and key horizon slices from the 3D seismic exploration results, identify tectonic units using reflection seismic phases, and invert the seismic layer velocity volume V(z) and attribute volume data such as amplitude A(x, y, z), wave impedance Z(x, y, z), and frequency distribution from pre-stack or post-stack data. The velocity acquisition method is usually the conversion from RMS velocity to Interval velocity, that is

[0044] where V int (z) is the Interval velocity within a certain depth interval, representing the true formation velocity, V(t1) is the RMS velocity at the two-way travel time t1 of the seismic profile, V(t2) is the RMS velocity at the two-way travel time t2 of the seismic profile, t1 is the two-way travel time corresponding to the upper bound of the time window, and t2 is the two-way travel time corresponding to the lower bound of the time window.

[0045] In the seismic modeling stage, based on the seismic layer velocity and attribute volume data, construct a regional 3D stratigraphic model, identify the fault distribution boundary through the tectonic surface, and combine the horizon consistency algorithm and the seismic amplitude alignment method along the axis to generate a spatially continuous tectonic morphology and stratigraphic connection model, assisting in identifying pressure anomaly-prone locations such as fracture-intensive areas and pinch-out horizons, providing a geometric structure basis for subsequent formation pressure prediction.

[0046] Subsequently, using the historical data of adjacent wells, relying on machine learning methods such as support vector regression (SVR) or multi-layer perceptron (MLP), input the logging data of adjacent wells such as density ρ(z), acoustic travel time Δt(z), natural gamma γ(z), and mud logging and drilling parameters, label the data as the measured formation pressure P(z) of the corresponding well section, construct a formation pressure prediction model, and extract gradient features to establish a formation pressure gradient model:

[0047] Through model regression analysis of the pressure change trend with depth, identify abnormal gradient sections and perform cluster analysis on abnormal points to generate pressure anomaly distribution data, including the location, frequency, associated lithology, and tectonic location of abnormal high-pressure sections.

[0048] On this basis, extract the formation pressure anomaly sections of adjacent wells, summarize their spatial distribution patterns and main influencing factors such as tectonic stress, stratigraphic discontinuity, and fracture density, and form a standardized distribution feature index matrix. Using this distribution data as a model constraint condition, combined with the regional seismic model and the designed trajectory of the target well position, adopt the well-seismic joint analysis method to align the seismic inversion velocity and logging data, and interpolate and predict the pressure state along the target well in the spatial domain.

[0049] Fracture pressure P fIt is calculated using the following formula:

[0050] where σ H is the maximum horizontal principal stress, P p is the pore pressure, and T0 is the tensile strength of the rock.

[0051] The collapse pressure is calculated based on the Mohr-Coulomb criterion:

[0052] where Pc is the collapse pressure, which is the minimum pressure that the mud needs to maintain when the wellbore is unstable. σ V is the vertical principal stress, usually obtained by integrating the rock density , σ h is the minimum horizontal stress, which can be estimated through a geostress model or acoustic logging. is the internal friction angle of the rock, usually obtained from core experiments or a lithology experience database, generally 25–40°, and Pp is the pore pressure.

[0053] The final output result is the multi-curve data of the formation pressure curve, pore pressure, fracture pressure, and collapse pressure along the well depth, providing a high-resolution quantitative basis for pre-drilling risk modeling.

[0054] In a possible implementation, during the pre-drilling stage, based on seismic data, adjacent well historical data, and design data, predictive risk data for the target monitoring well is analyzed and generated, specifically including: based on the predicted formation pressure data, combined with the design data of the target monitoring well, calculating the predicted formation pressure gradient of the target monitoring well. The design data includes well trajectory design, drilling fluid parameters, bit and drill string assembly, and wellbore structure design; correcting and optimizing the predicted formation pressure gradient through the Eaton method, and combining the adjacent well historical accident data to compare and analyze the risk data of lost circulation, kick, or well collapse occurring in the target monitoring well, generating predictive risk data. Specifically, during the pre-drilling stage, first, the 3D seismic data of the area where the target monitoring well is located is imported through the seismic data management platform, and the seismic profile containing the target well trajectory is extracted. The structural interpretation module combines the fault identification algorithm and the seismic attribute analysis algorithm to analyze the seismic attribute volume to generate a structural wireframe. And the velocity analysis method is used to perform velocity modeling on each structural layer to obtain the 3D seismic layer velocity volume, and then a seismic geological model containing the structural form, fault distribution, and formation continuity is constructed to support subsequent predictive analysis.

[0055] Subsequently, extract the historical data of adjacent wells to clean, standardize, and depth-register the logging curves, mud logging data, drilling parameters, and historical actual formation pressure data of the adjacent wells around the target well. Construct a multi-source feature dataset based on the horizons of adjacent wells, and use machine learning algorithms such as Gradient Boosting Decision Tree (GBDT), Random Forest (RF), or Long Short-Term Memory Network (LSTM) to learn the pressure response patterns during the drilling process and establish a regional formation pressure gradient prediction model. The typical pressure gradient calculation formula is:

[0056] where P g is the pressure gradient, ΔP is the pressure difference between adjacent measurement points, and ΔD is the depth difference.

[0057] Based on the trained pressure gradient model, select multiple key adjacent well sections as comparison objects, extract the abnormal inflection points in their predicted pressure curves, and identify abnormal pressure zones in combination with geological interpretation information. Further, statistically analyze the distribution characteristics of each abnormal pressure point, including geological location, relevant tectonic background, lithological characteristics, prediction error, etc., to generate a knowledge graph reflecting the abnormal pressure distribution pattern and genetic mechanism.

[0058] On this basis, project the designed trajectory of the target well into the seismic model and combine it with fault and tectonic deformation information. Use the well-seismic joint analysis method to perform velocity-pressure conversion on each drilled interval of the target well, and calculate the predicted formation pressure using Bowers' empirical relationship based on the seismic interval velocity volume:

[0059] where P p is the predicted pore pressure, P o is the pressure under normal compaction trend, V is the measured interval velocity, V o is the velocity under normal compaction trend, and n is the empirical exponential parameter, usually ranging from 1.2 to 1.5.

[0060] Fuse the obtained predicted formation pressure data with the designed data of the target well, input parameters such as drilling fluid density, well diameter, well inclination, bit structure, and drill string assembly parameters, and decompose the well section in combination with the well trajectory model to further calculate the predicted formation pressure gradient and drilling fluid density window for different well sections:

[0061]

[0062] where MW min is the minimum density corresponding to the pore pressure, MW max is the maximum density corresponding to the fracture pressure, D is the well depth, and P f is the formation fracture pressure.

[0063] Finally, the Eaton method is used to correct the above predicted pressure gradient values, and its calculation formula is:

[0064] Among them, σ v is the overburden rock pressure, P n is the normal pore pressure, Δt is the measured acoustic travel time difference, and Δt n is the acoustic travel time difference of the normal compaction trend. x is an empirical coefficient, usually 3.0.

[0065] Perform horizon comparison on the corrected predicted pressure curve, introduce accident records such as historical well losses, overflows, and well collapses of adjacent wells, and conduct cross-verification against key well sections to determine the risk level of the target well at each depth section, so as to generate a predicted risk data map covering the entire well section, including identification of pressure anomaly intervals, risk type classification, risk intensity grading, and preliminary warning suggestions, etc.

[0066] S130. During the drilling process, use the drilling data collected in real time for the target monitoring well to calculate the actual formation pressure data in real time, and compare the actual formation pressure data with the predicted risk data.

[0067] During the drilling process, first based on the real-time data acquisition system deployed at the target monitoring well site, continuously receive and process the data uploaded in real time by equipment such as the comprehensive mud logging instrument and the drilling parameter collector, including key parameters such as the drilling rate (R, in meters per hour), the rotary speed (N, in revolutions per minute), the weight on bit (W, in kilonewtons), and the drilling fluid density. After timestamp verification, all data is written into the drill-in data stream at second-level precision and subjected to sliding window filtering and outlier rejection processing to ensure the continuity and reliability of the input data.

[0068] Based on the drilling data obtained in real time, the system automatically calculates the drilling index DC at set time intervals. This index is used to characterize the comprehensive characteristics of formation drillability and downhole mechanical response, and its calculation formula is:

[0069] Where R is the drilling rate, N is the rotational speed, and W is the weight on bit. After the calculation results are smoothed by a moving average filter, a dynamic drilling index curve for the current well section is formed, providing real-time input for subsequent pressure gradient estimation. The logarithm used in the formula is base-10 logarithm, which is a commonly used logarithmic system in engineering calculations and is different from the natural logarithm ln. In the early practical formulas in petroleum engineering, base-10 logarithms were widely used to handle the non-linear responses of multi-order parameters to simplify data fitting and trend expression. This is a standard logarithmic dimension normalization method, making the unit change amplitude linearly approximate to the physical response. This term comes from the unit conversion between the drilling rate and the rotational speed. To make the drilling rate R (the unit is often m / h or ft / h) and the rotational speed N (the unit is rev / min) form a dimensionless expression in the ratio, minutes need to be converted to hours, so 60 is introduced. R / 60N is the penetration per revolution, indicating the distance the bit advances per revolution, which is an important reflection of the drillability of the rock formation. The faster the drilling, the larger this ratio is. Derived from the normalization structure of the DC index, it is used to standardize the influence of the weight on bit on the drilling index. The reason for choosing 12 is that through a large number of statistical analyses of actual wells during the formula construction stage, it is found that using 12 as the reference value for weight-on-bit normalization can make the weight-on-bit response curves under different pressure systems have the maximum block adaptability and relative linear stability. In other words, as a normalization coefficient, it is an empirically determined constant, which is a compromise treatment equivalent to converting the non-linear influence of the weight on bit into a linear superposition term.

[0070] The system further calls the conventional formation pressure gradient (set according to regional geological characteristics or obtained by regression from adjacent well data) and the normal drilling index D Cn (the reference drilling index established after normalization of the same type of well section) as reference parameters, and substitutes the current drilling index DC obs into the formula of the modified drilling index method for formation pressure gradient estimation:

[0071] Where β is the empirical adjustment coefficient, which is determined according to different geological blocks and lithological characteristics (generally between 0.5 and 1.5). The system can continuously optimize this parameter through machine learning methods to adapt to different drilling environments, and finally output the dynamic formation pressure gradient value.

[0072] Finally, the system performs depth integration processing on the above-mentioned real-time estimated pressure gradient to calculate the formation pressure values corresponding to each drilling point:

[0073] Among them, D is the current well depth. The integration uses a layer-by-layer recursive algorithm to improve the calculation efficiency. At the same time, the obtained actual formation pressure data is compared layer by layer with the predicted risk data established in the pre-drilling stage to analyze abnormal pressure responses, assist in identifying whether the current drilling operation is approaching risk boundaries such as lost circulation and overflow, and provide real-time early warnings and decision-making bases for on-site operators.

[0074] In a possible implementation manner, during the drilling process, using the drilling data collected in real time for the target monitoring well, the actual formation pressure data is calculated in real time, and the actual formation pressure data is compared with the predicted risk data. Specifically, it further includes: aligning the actual formation pressure data in the depth sequence with the depth interval corresponding to the predicted formation pressure data of the predicted risk data, and performing horizon calibration; using the residual calculation method to perform point-to-point calculation on the aligned predicted risk data and the actual formation pressure data, and calculating the absolute value of the residual; judging the size relationship between the absolute value of the residual and the preset threshold. If it is judged that the absolute value of the residual is greater than or equal to the preset threshold, then combined with the predicted formation pressure gradient of the target monitoring well corresponding to the predicted risk data, analyze the pressure change of the target monitoring well.

[0075] Specifically, first, accurately align the actual formation pressure data and the predicted risk data in the depth dimension. Interpolate or resample the real-time pressure data according to the sampling depth through the automatic horizon registration algorithm to match the formation horizon structure corresponding to the predicted data. Combine the well trajectory design data and the well deviation correction parameters to perform true vertical depth correction on different measurement positions to ensure the one-to-one correspondence relationship of the two sets of data in the geological horizon, and perform horizon anchoring through the formation marker bed or fault intersection point to improve the matching accuracy.

[0076] After completing the data alignment, for the real-time formation pressure value P real (z) and the predicted formation pressure value P pred (z) at the same depth point, perform point-by-point residual calculation, using the absolute value of the residual function, specifically as follows:

[0077] Among them, ΔP n (z) is the absolute value of the residual, and δ is a positive number to avoid the denominator approaching zero. This formula is convenient for the unified judgment standard between different pressure scale segments and improves the comparability of cross-horizon identification. According to the absolute value of the residual, generate a pressure offset sequence covering the entire well section. This sequence can be smoothed through a sliding window to eliminate the influence of local disturbances, and at the same time mark the position of the maximum residual and its depth information to form a residual distribution map for subsequent analysis.

[0078] The system then introduces a preset residual threshold to perform a threshold judgment on the residual values of all depth points. When the absolute value of the full residual is greater than or equal to the preset threshold, the depth segment is marked as an outlier and further analysis logic is activated. At this time, the system automatically retrieves the predicted formation pressure gradient of the depth segment in the predicted risk data, and calculates the gradient difference and change rate based on the current real-time pressure change trend. The following formula is used to quantify the degree of deviation between the actual pressure change and the prediction model:

[0079] Based on the above-mentioned gradient difference, residual intensity and predicted risk level of the layer, the system classifies the pressure change in the well section as one of three situations: high pressure ahead, overpressure lag or abnormal leakage pressure, and presents it graphically in the comprehensive well string diagram, providing a reliable basis for real-time early warning during drilling and subsequent adjustment of control parameters.

[0080] S140, during the drilling stage, characterizes the pressure safety window based on the predicted risk data, and calculates the circulating equivalent density of the target monitoring well in real time. The circulating equivalent density is compared with the safety window to identify and analyze drilling safety risks.

[0081] In one possible implementation, during the drilling stage, a pressure safety window is depicted based on the predicted risk data, and the circulating equivalent density of the target monitoring well is calculated in real time. The circulating equivalent density is compared with the safety window to identify and analyze the drilling safety risk, specifically including: constructing a pressure safety window for the well section distribution based on the formation pore pressure, fracture pressure and collapse pressure data contained in the predicted risk data; calculating the equivalent static pressure at the bottom of the drill string in real time based on the drilling fluid density and the current well depth; making a preliminary comparison and judgment between the equivalent static pressure and the pressure safety window. If it is determined that the equivalent static pressure is in the pressure safety window, the circulating equivalent density is calculated using a dynamic fluid dynamics model; depth-aligning the continuous curve constructed based on the circulating equivalent density with the pressure safety window, and determining the drilling safety risk based on the relative size relationship between the continuous curve and the pressure safety window.

[0082] Specifically, firstly, the pore pressure, fracture pressure and collapse pressure data of the corresponding well section are extracted based on the formation pressure prediction results established in the pre-drilling stage, and the data are interpolated and completed according to the well depth sequence and unified in unit. Combined with the well trajectory data of the target monitoring well, they are mapped to the true vertical depth domain of the wellbore to form three pressure envelope curves covering the entire well section, namely, the pore pressure curve is the lower limit, the fracture pressure curve is the upper limit, and the collapse pressure curve is the wellbore stability control benchmark. Figure 2 The three together constitute the pressure safety window of the target well. This window can be visualized as an effective density window that changes with depth, that is, the minimum density is the lowest safety density to avoid well collapse, and the maximum density is the critical density to avoid formation rupture.

[0083] Subsequently, in combination with the drilling fluid density ρ collected in real time at the well site m and the current well depth D, the following formula is used:

[0084] Calculate the equivalent static pressure P at the bottom of the drill string in real time ESD , which is used to quickly determine whether the current static pressure falls within the safety window. If P ESD is within the range of [P c , P f , it is determined that the preliminary safety condition is established, and the system enters the next dynamic monitoring process.

[0085] On this basis, the system constructs a dynamic hydraulics framework based on the drilling parameters collected in real time, such as displacement, pump pressure, drill string structure and wellbore diameter structure data, using a friction model and a wellbore hydraulics model to calculate the additional pressure ΔP in the annulus ECD , and superimpose it with the static pressure to obtain the real-time circulating equivalent density (ECD) according to the formula:

[0086] Obtain the ECD curve in the continuous depth domain, forming a high-precision pressure response curve that dynamically reflects the bottom hole annulus pressure. Among them, 0.052 is a commonly used conversion coefficient for converting the drilling fluid density (in pounds per gallon, ppg) and the well depth (in feet, ft) into the formation pressure (in pounds per square inch, psi). The source of this coefficient is derived from the following physical relationships and unit conversions: The basic formula for formation pressure is:

[0087] where: P is the pressure (Pa), ρ is the density (kg / m³), g is the acceleration due to gravity (9.81 m / s²), and h is the depth (m). In order to adapt to the commonly used unit system in the petroleum industry, the units need to be converted. The density is converted to pounds per gallon (ppg), the depth is converted to feet (ft), and the pressure is converted to psi. After unit conversion: 1 ppg = 119.8264 kg / m 3 , 1 ft = 0.3048 m, 1 psi = 6894.76 Pa.

[0088] After substituting into the formula and completing the unit conversion, the approximate conversion relationship is obtained:

[0089] Therefore, 0.052 is an empirical coefficient obtained through physical constants and unit conversions, which is used to simplify the rapid calculation of formation pressure and density depth in drilling engineering.

[0090] Align the real-time updated ECD curve with the existing pressure safety window in a one-to-one and in-depth manner, compare the ECD with the pore pressure line and the fracture pressure line. If ECD > P f / (0.052·D), it is determined that there is a risk of overflow or well leakage. If ECD < P c / (0.052·D), there is a risk of wellbore shrinkage or wellbore instability. The system will trigger a warning signal according to the above logic judgment and mark the current risk level and location with a risk label in the real-time well string diagram, realizing the safety dynamic identification and feedforward response mechanism during the drilling process. At the same time, it supports setting a safety margin band and outputs trend risk suggestions in advance within the critical interval near the upper and lower boundaries to ensure that the downhole pressure operates within the dynamic safety interval.

[0091] S150. After the drilling is completed, summarize the predicted risk data, actual formation pressure data, and drilling safety risks, and update and iterate the model algorithm through the summarized data.

[0092] In a possible implementation manner, after the drilling is completed, summarize the predicted risk data, actual formation pressure data, and drilling safety risks, and update and iterate the model algorithm through the summarized data. Specifically, it includes: for the predicted risk data, actual formation pressure data, and drilling safety risks, perform well depth consistency correction through the depth registration algorithm and organize them into a standardized format according to the stratigraphic structure. Archive the predicted risk data, actual formation pressure data, and drilling safety risks for each well section to form a complete risk assessment data set; extract the deformation characteristics of the ECD curve, the fluctuation pattern of drilling parameters, and the corresponding response feedback data in the abnormal section as the feature vectors of the training samples for input; use the key parameters adopted before the drilling is completed as the variables to be optimized, construct a loss function, and aim to minimize the weighted mean square error of the whole well section between the predicted pressure and the measured pressure. The key parameters include the drilling index gradient model coefficient, the Eaton method index value, and the adjacent well reference curve weight factor; for the prediction model constructed based on machine learning, use the risk assessment data set as an extended sample for incremental learning to absorb the new feature changes in the actual drilling process.

[0093] Specifically, first, after the drilling is completed, the system uniformly organizes the collected predicted risk data, actual formation pressure data, and drilling safety risk event records, and uses the depth registration algorithm to accurately align various data in the well depth dimension to solve the depth error problem caused by different data sources and inconsistent sampling granularities. Subsequently, standardize and organize the data according to the stratigraphic structure, and structurally encapsulate the predicted values, measured values, and risk labels for each well section to form a risk assessment data set with spatial continuity and label integrity, providing a high-quality sample basis for subsequent modeling and iterative training.

[0094] Then, the system automatically identifies the abnormal sections with high residuals, non-linear disturbances or risk events that have occurred in the well section, extracts the local deformation characteristics of the corresponding real-time ECD curve, and the fluctuation patterns of drilling parameters such as WOB, RPM, and pump pressure. Combining with the data of on-site disposal responses, representative input feature vectors are constructed, and their response results and risk types are labeled as training samples for supervised learning or reinforcement learning models. This process is assisted by sliding window analysis and frequency domain feature extraction algorithms to ensure the integrity and temporal continuity of the samples.

[0095] Furthermore, in the model parameter optimization step, key model parameters adopted before drilling, such as the adjustment coefficient in the drilling exponent gradient model, the Eaton method exponent value, the fitting weight factor of the adjacent well reference curve, etc., are used as adjustable variables to construct a loss function of the weighted mean square error of the residuals for the entire well section, that is:

[0096] where P real (z) is the real-time formation pressure value, P pred (z) is the predicted formation pressure value, and w(z) is the depth weighting factor dynamically set in combination with the risk level. The above variables are iteratively solved through backpropagation or least squares optimization algorithms to correct the local error response ability of the model and improve its fitting accuracy for non-linear pressure changes.

[0097] Finally, for the formation pressure prediction models constructed based on machine learning, such as random forest, support vector machine, deep neural network, etc., the above summarized risk assessment data set is used as an incremental training set to update the parameters and fine-tune the model without resetting the model structure, so that it can absorb new samples, new abnormal patterns and new response paths that appear in the actual drilling process. Integrate the predicted risk data, actual formation pressure data and drilling safety risks into a structured and high-quality risk assessment data set, extract the key response characteristics of the abnormal sections and construct representative input feature vectors, and at the same time establish a residual loss function to drive the optimization of key model parameters, providing a data basis with spatio-temporal consistency, label completeness and feature discrimination for model training; and it is precisely supported by the data structure, feature system and optimization logic constructed in the above steps that the new formed sample set is injected into the original machine learning model by using the incremental learning mechanism for parameter fine-tuning and generalization ability update under the condition of maintaining the structure, so that the model can gradually improve the prediction accuracy and risk identification ability under new wells, new areas and new structures by absorbing the continuously rich complex scenarios and abnormal evolution laws in actual operations, and realize the transition of the model from static mapping to dynamic self-evolution.

[0098] This incremental learning process can adopt online learning or batch update mechanism, and cross-validation evaluation is performed after training is completed to ensure that the improved model accuracy can be stably applied to the real-time risk prediction of subsequent target well sections, thereby realizing the continuous adaptation and intelligent evolution of the model in engineering applications.

[0099] This embodiment also discloses a formation pressure warning device based on model prediction while drilling, which is used to execute any one of the above-mentioned formation pressure warning methods based on model prediction while drilling, referring to Figure 3 The device includes an acquisition module 301, a processing module 302 and an output module 303, wherein: The acquisition module 301 is used to acquire seismic data, historical data of adjacent wells, and design data for a target monitoring well.

[0100] The processing module 302 is used to analyze and generate predicted risk data of the target monitoring well based on seismic data, historical data of adjacent wells and design data in the pre-drilling stage.

[0101] The processing module 302 is used to calculate the actual formation pressure data in real time during the drilling process using the drilling data collected in real time from the target monitoring well, and compare the actual formation pressure data with the predicted risk data.

[0102] The processing module 302 is used to characterize the pressure safety window based on the predicted risk data during the drilling stage, and calculate the circulating equivalent density of the target monitoring well in real time, compare the circulating equivalent density with the safety window, and identify and analyze drilling safety risks.

[0103] The output module 303 is used to summarize the predicted risk data, actual formation pressure data and drilling safety risk after drilling is completed, and update the model algorithm based on the summarized data.

[0104] In a possible implementation, the acquisition module 301 is configured to acquire seismic data in the area where the target monitoring well is located, where the seismic data includes seismic profiles, seismic layer velocities, and seismic attribute volumes.

[0105] The processing module 302 is used to model the structural morphology, fault development, and stratum continuity of the regional strata using seismic data to obtain a seismic model.

[0106] Processing module 302 is used to analyze the formation pressure change characteristics of each adjacent well during drilling using a machine learning algorithm based on the historical data of adjacent wells, and establish a formation pressure gradient model. The historical data of adjacent wells includes adjacent well logging data, adjacent well recording data, adjacent well drilling parameter data, and adjacent well formation pressure data.

[0107] The processing module 302 is configured to compare the formation pressure anomaly points of different adjacent wells through a formation pressure gradient model, and summarize the distribution data of abnormal pressure, where the distribution data includes distribution patterns and influencing factors.

[0108] The output module 303 is configured to predict the formation pressure distribution at the well position through seismic data by using a well-seismic joint analysis method based on the seismic model and the distribution data, and in combination with the geological structure characteristics of the area where the target monitoring well is located, so as to obtain predicted formation pressure data.

[0109] In a possible implementation manner, the processing module 302 is configured to calculate the predicted formation pressure gradient of the target monitoring well based on the predicted formation pressure data and in combination with the design data of the target monitoring well, where the design data includes well trajectory design, drilling fluid parameters, bit and drill string assembly, and wellbore structure design.

[0110] The output module 303 is configured to correct and optimize the predicted formation pressure gradient through the Eaton method, and in combination with the historical accident data of adjacent wells, compare and analyze the risk data of lost circulation, overflow or well collapse occurring in the target monitoring well, and generate predicted risk data.

[0111] In a possible implementation manner, the processing module 302 is configured to continuously obtain drilling data according to the real-time data acquisition system deployed in the target monitoring well, where the drilling data includes drilling speed, rotation speed, drilling pressure, and drilling fluid density.

[0112] The processing module 302 is configured to calculate the drilling index according to the drilling data, and specifically calculate it through the following formula:

[0113] where DC is the drilling index, R is the drilling speed, N is the rotation speed, and W is the drilling pressure.

[0114] The processing module 302 is configured to estimate the formation pressure gradient through the modified drilling index method according to the real-time drilling index of the current drilling, in combination with the drilling fluid density and the conventional formation pressure gradient, and specifically calculate it through the following formula:

[0115] where is the normal formation pressure gradient, DC obs is the real-time drilling index of the current drilling, DC n is the reference drilling index of adjacent wells or historical formations under normal pressure, and β is the empirical adjustment coefficient.

[0116] The processing module 302 is configured to integrate the formation pressure gradient to obtain actual formation pressure data.

[0117] In a possible implementation, the processing module 302 is configured to align the actual formation pressure data with the depth interval corresponding to the predicted formation pressure data corresponding to the predicted risk data in depth sequence, and perform layer calibration.

[0118] The processing module 302 is configured to perform point-to-point calculation on the aligned predicted risk data and the actual formation pressure data using a residual calculation method to calculate the absolute value of the residual.

[0119] Processing module 302 is used to determine the relationship between the absolute value of the residual and a preset threshold. If it is determined that the absolute value of the residual is greater than or equal to the preset threshold, the pressure change of the target monitoring well is analyzed in combination with the predicted formation pressure gradient of the target monitoring well corresponding to the predicted risk data.

[0120] In a possible implementation, the processing module 302 is configured to construct a pressure safety window for well section distribution based on formation pore pressure, fracture pressure, and collapse pressure data included in the predicted risk data.

[0121] The processing module 302 is used to calculate the equivalent static pressure at the bottom of the drill string in real time according to the drilling fluid density and the current well depth.

[0122] The processing module 302 is used to perform a preliminary comparison and judgment between the equivalent static pressure and the pressure safety window. If it is determined that the equivalent static pressure is within the pressure safety window, the circulating equivalent density is calculated using a dynamic fluid dynamics model.

[0123] The processing module 302 is configured to align the depth of the continuous curve constructed according to the circulating equivalent density with the pressure safety window, and determine the drilling safety risk according to the relative size relationship between the continuous curve and the pressure safety window.

[0124] In one possible implementation, the processing module 302 is used to perform well depth consistency correction on the predicted risk data, actual formation pressure data, and drilling safety risk through a depth registration algorithm, and organize them into a standardized format according to the layer structure, and archive the predicted risk data, actual formation pressure data, and drilling safety risk of each well section to form a complete risk assessment data set.

[0125] The processing module 302 is used to extract the ECD curve deformation characteristics, drilling parameter fluctuation patterns and corresponding response feedback data in the abnormal section as feature vector inputs of the training samples.

[0126] Processing module 302 is used to construct a loss function using the key parameters used before the end of drilling as variables to be optimized, with the goal of minimizing the weighted mean square error between the predicted pressure and the measured pressure in the entire well section. The key parameters include the drilling index gradient model coefficient, the Eaton method index value, and the weight factor of the adjacent well reference curve.

[0127] An output module 303 is configured to perform incremental learning on a prediction model built based on machine learning, using a risk assessment data set as an extended sample, so as to absorb new feature changes in the actual drilling process.

[0128] It should be noted that when the device provided in the above embodiment realizes its functions, only the division of the above functional modules is used for illustration. In actual applications, the above functions can be allocated to different functional modules according to needs, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. In addition, the device and method embodiments provided in the above embodiment belong to the same concept, and the specific implementation process can be seen in the method embodiment, which will not be elaborated here.

[0129] This embodiment also discloses an electronic device. Referring to Figure 4 , the electronic device may include: at least one processor 401, at least one communication bus 402, a user interface 403, a network interface 404, and at least one memory 405.

[0130] Among them, the communication bus 402 is used to realize the connection and communication between these components.

[0131] Among them, the user interface 403 may include a display screen (Display) and a camera (Camera). Optionally, the user interface 403 may further include a standard wired interface and a wireless interface.

[0132] Among them, the network interface 404 may optionally include a standard wired interface and a wireless interface (such as a WI-FI interface).

[0133] Processor 401 may include one or more processing cores. Using various interfaces and circuits, processor 401 connects to various components within the server. It executes instructions, programs, code sets, or instruction sets stored in memory 405, as well as accesses data stored in memory 405, to perform various server functions and process data. Optionally, processor 401 may be implemented using at least one of the following hardware forms: a digital signal processing (DSP), a field-programmable gate array (FPGA), or a programmable logic array (PLA). Processor 401 may integrate one or a combination of a central processing unit (CPU), a graphics processing unit (GPU), and a modem. The CPU primarily processes the operating system, user interface, and application programs; the GPU is responsible for rendering and drawing content displayed on the display screen; and the modem handles wireless communications. It is understood that the modem may also be implemented independently of the processor 401 and implemented as a separate chip.

[0134] Memory 405 may include random access memory (RAM) or read-only memory (ROM). Optionally, the memory may include non-transitory computer-readable storage medium. Memory 405 may be used to store instructions, programs, code, code sets, or instruction sets. Memory 405 may include a program storage area and a data storage area. The program storage area may store instructions for implementing an operating system, instructions for at least one function (such as a touch function, sound playback function, image playback function, etc.), instructions for implementing each of the aforementioned method embodiments, and the data storage area may store data related to each of the aforementioned method embodiments. Memory 405 may also optionally be at least one storage device located remotely from the processor 401. Memory 405, as a computer storage medium, may include an operating system, a network communication module, a user interface 403 module, and an application for a model-based prediction-based formation pressure warning method while drilling.

[0135] exist Figure 4In the electronic device shown, the user interface 403 is mainly used to provide an input interface for the user and obtain data input by the user; and the processor 401 can be used to call an application stored in the memory 405 for a method for early warning of formation pressure while drilling based on model prediction. When executed by one or more processors 401, the electronic device executes one or more methods in the above-mentioned embodiments.

[0136] It should be noted that for the aforementioned method embodiments, for simplicity of description, they are all expressed as a series of action combinations, but those skilled in the art should be aware that this application is not limited by the order of the actions described, because according to this application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required for this application.

[0137] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0138] In the several embodiments provided in this application, it should be understood that the disclosed devices can be implemented in other ways. For example, the device embodiments described above are merely schematic, such as the division of units, which is only a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some service interface, and the indirect coupling or communication connection of devices or units can be electrical or other forms.

[0139] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0140] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0141] When the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a memory 405 and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods in various embodiments of the present application. The aforementioned memory 405 includes various media that can store program codes, such as USB flash drives, mobile hard disks, magnetic disks, or optical discs.

[0142] The present application also discloses a computer-readable storage medium storing instructions. When executed by one or more processors 401, it causes the electronic device to execute the method as described in one or more of the above embodiments.

[0143] The foregoing are only exemplary embodiments of the present disclosure and should not be used to limit the scope of the present disclosure. That is, any equivalent changes and modifications made in accordance with the teachings of the present disclosure still fall within the scope covered by the present disclosure. After considering the specification and the disclosure of the practical truth, those skilled in the art will easily think of other implementation manners of the present disclosure. The present application aims to cover any variations, uses, or adaptive changes of the present disclosure, and these variations, uses, or adaptive changes follow the general principles of the present disclosure and include the common general knowledge or conventional technical means in the technical field not recorded in the present disclosure. The specification and the embodiments are only regarded as exemplary, and the scope and spirit of the present disclosure are defined by the claims.

Claims

1. A method for predicting formation pressure while drilling based on model prediction, characterized in that The method includes: Obtaining seismic data, adjacent well historical data, and design data for a target monitoring well; In the pre-drilling stage, based on the seismic data, the adjacent well historical data, and the design data, analyzing and generating prediction risk data for the target monitoring well; During the drilling process, using the drilling data collected in real time for the target monitoring well, calculating the actual formation pressure data in real time, and comparing the actual formation pressure data with the prediction risk data; In the middle-drilling stage, based on the prediction risk data, depicting a pressure safety window, calculating the circulating equivalent density of the target monitoring well in real time, comparing the circulating equivalent density with the safety window, and identifying and analyzing drilling safety risks; After the drilling is completed, summarizing the prediction risk data, the actual formation pressure data, and the drilling safety risks, and updating and iterating the model algorithm through the summarized data.

2. The method for predicting formation pressure while drilling based on model prediction according to claim 1, characterized in that Before the step of, in the pre-drilling stage, based on the seismic data, the adjacent well historical data, and the design data, analyzing and generating prediction risk data for the target monitoring well, the method further includes: Obtaining seismic data of the area where the target monitoring well is located, where the seismic data includes seismic profiles, seismic layer velocities, and seismic attribute bodies; Modeling the structural form, fault development, and formation continuity of the regional formation through the seismic data to obtain a seismic model; Analyzing the formation pressure change characteristics encountered in the drilling of each adjacent well through the adjacent well historical data using a machine learning algorithm, and establishing a formation pressure gradient model, where the adjacent well historical data includes adjacent well logging data, adjacent well mud logging data, adjacent well drilling parameter data, and adjacent well formation pressure data; Comparing the formation pressure abnormal points of different adjacent wells through the formation pressure gradient model, and summarizing the distribution data of abnormal pressures, where the distribution data includes distribution patterns and influencing factors; Based on the seismic model and the distribution data, combined with the geological structure characteristics of the area where the target monitoring well is located, adopting a well-seismic joint analysis method to predict the formation pressure distribution at the well location through seismic data to obtain predicted formation pressure data.

3. The method for predicting formation pressure while drilling based on model prediction according to claim 2, characterized in that, The step of, in the pre-drilling stage, based on the seismic data, the adjacent well historical data, and the design data, analyzing and generating prediction risk data for the target monitoring well specifically includes: Based on the predicted formation pressure data, combined with the design data of the target monitoring well, calculating the predicted formation pressure gradient of the target monitoring well, where the design data includes well trajectory design, drilling fluid parameters, bit and drill string assembly, and wellbore structure design; Calibrating and optimizing the predicted formation pressure gradient through the Eaton method, and combined with adjacent well historical accident data, comparing and analyzing the risk data of lost circulation, overflow, or well collapse occurring in the target monitoring well to generate the prediction risk data.

4. A method for predicting formation pressure while drilling based on model prediction according to claim 1, characterized in that, The step of, during the drilling process, using the drilling data collected in real time for the target monitoring well, calculating the actual formation pressure data in real time, and comparing the actual formation pressure data with the prediction risk data specifically includes: Continuously obtain the drilling data according to the real-time data acquisition system deployed in the target monitoring well, where the drilling data includes the drilling rate, rotation speed, drilling pressure, and drilling fluid density; Calculate the drilling index according to the drilling data, specifically calculated by the following formula: ； Where DC is the drilling index, R is the drilling rate, N is the rotation speed, and W is the drilling pressure; According to the real-time drilling index of the current drilling, combined with the drilling fluid density and the conventional formation pressure gradient, estimate the formation pressure gradient by the corrected drilling index method, specifically calculated by the following formula: ； Among them, ∇P n is the normal formation pressure gradient, DC obs is the real-time drilling rate of penetration during current drilling, DC n is the reference drilling rate of penetration of adjacent wells or historical formations under normal pressure, and β is the empirical adjustment coefficient; Integrate the formation pressure gradient to obtain the actual formation pressure data.

5. A method for predicting formation pressure while drilling based on model prediction according to claim 4, characterized in that, During the drilling process, use the drilling data collected in real time for the target monitoring well to calculate the actual formation pressure data in real time, and compare the actual formation pressure data with the predicted risk data. Specifically, it further includes: Align the actual formation pressure data with the depth interval corresponding to the predicted formation pressure data of the predicted risk data according to the depth sequence, and perform horizon calibration; Use the residual calculation method to perform point-to-point calculation on the aligned predicted risk data and the actual formation pressure data, and calculate the absolute value of the residual; Judge the magnitude relationship between the absolute value of the residual and the preset threshold. If it is judged that the absolute value of the residual is greater than or equal to the preset threshold, analyze the pressure change of the target monitoring well in combination with the predicted formation pressure gradient of the target monitoring well corresponding to the predicted risk data.

6. The method for predicting formation pressure while drilling based on model prediction according to claim 1, wherein, During the drilling process, based on the predicted risk data, depict the pressure safety window, and calculate the circulating equivalent density of the target monitoring well in real time. Compare the circulating equivalent density with the safety window to identify and analyze the drilling safety risk. Specifically, it includes: Construct a pressure safety window for the well section distribution based on the formation pore pressure, fracture pressure, and collapse pressure data included in the predicted risk data; Calculate the equivalent static pressure at the bottom of the drill string in real time according to the drilling fluid density and the current well depth; Make a preliminary comparison and judgment between the equivalent static pressure and the pressure safety window. If it is determined that the equivalent static pressure is within the pressure safety window, calculate the circulating equivalent density using a dynamic hydraulics model; Align the continuous curve constructed according to the circulating equivalent density with the pressure safety window in depth, and determine the drilling safety risk according to the relative size relationship between the continuous curve and the pressure safety window.

7. A method for predicting formation pressure while drilling based on model prediction according to claim 1, characterized in that, After the drilling is completed, summarize the predicted risk data, the actual formation pressure data, and the drilling safety risk, and update and iterate the model algorithm through the summarized data. Specifically, it includes: Perform well depth consistency correction on the predicted risk data, the actual formation pressure data, and the drilling safety risk through a depth registration algorithm, organize them into a standardized format according to the horizon structure, and archive the predicted risk data, the actual formation pressure data, and the drilling safety risk of each well section to form a complete risk assessment data set; Extract the ECD curve deformation characteristics, drilling parameter fluctuation patterns, and corresponding response feedback data in the abnormal section, and input them as the feature vectors of the training samples; Take the key parameters adopted before the end of drilling as the variables to be optimized, construct a loss function, and aim to minimize the weighted mean square error of the whole well section between the predicted pressure and the measured pressure. The key parameters include the drilling index gradient model coefficient, the Eaton method index value, and the adjacent well reference curve weight factor; For the prediction model constructed based on machine learning, use the risk assessment data set as the extended sample for incremental learning to absorb the new feature changes in the actual drilling process.

8. A formation pressure early warning device while drilling based on model prediction, characterized in that The device is used to execute a model-prediction-based formation pressure early warning method during drilling as described in any one of claims 1-7. The device includes an acquisition module (301), a processing module (302), and an output module (303), where: The acquisition module (301) is used to acquire seismic data, adjacent well historical data, and design data for the target monitoring well; The processing module (302) is used to analyze and generate the predicted risk data of the target monitoring well based on the seismic data, the adjacent well historical data, and the design data during the pre-drilling stage; The processing module (302) is used to calculate the actual formation pressure data in real time using the drilling data collected in real time for the target monitoring well during the drilling process, and compare the actual formation pressure data with the predicted risk data; The processing module (302) is used to depict the pressure safety window based on the predicted risk data during the drilling stage, calculate the circulating equivalent density of the target monitoring well in real time, and compare the circulating equivalent density with the safety window to identify and analyze the drilling safety risks; The output module (303) is used to summarize the predicted risk data, the actual formation pressure data, and the drilling safety risks after the drilling is completed, and update and iterate the model algorithm through the summarized data.

9. An electronic device, characterized in that, It includes a processor (401), a communication bus (402), a user interface (403), a network interface (404), and a memory (405). The memory (405) is used to store instructions. The user interface (403) and the network interface (404) are both used to communicate with other devices. The communication bus (402) is used to realize the connection and communication between components in the electronic device. The processor (401) is used to execute the instructions stored in the memory (405) to enable the electronic device to execute the method described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions, and when the instructions are executed, the method described in any one of claims 1-7 is executed.

Images