Dynamic control system for borehole stability during tidal formation drilling

Abstract

The present application relates to the technical field of foundation and foundation engineering construction, disclose the hole wall stability dynamic control system of tidal stratum drilling process, the system includes: multi-source dynamic perception module, for real-time acquisition of drilling pressure, tidal water level and hole displacement of soil layer around etc. Multidimensional data; data processing and intelligent decision module, it receives the sensing data, through the built-in dynamic mapping and self-calibration model analysis and prediction, to generate prospective hole wall stability control instruction; collaborative control and planning execution module, it receives the control instruction, and it is parsed to the collaborative regulation action of the hole wall and drilling parameter, so as to carry out accurate, closed loop dynamic control to hole wall stability. The present application realizes the prospective, self-adapting fine adjustment of hole wall stability through multi-source perception, self-calibration model prediction and collaborative planning control, improves the drilling safety.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of foundation and foundation engineering, in particular to a hole wall stability dynamic control system in the process of tidal stratum drilling. BACKGROUND

[0002] When carrying out pile foundation construction of bridge, offshore wind power and other projects in coastal, estuary and island reef areas, drilling operations are generally affected by the tide phenomenon. The periodic rise and fall of water level caused by the tide will directly cause the dynamic change of pore water pressure in the soil around the hole, thereby continuously and complexly challenging the stability of the hole wall.

[0003] In the existing drilling construction practice, in order to maintain the stability of the hole wall, the wall protection mud technology is usually used. The core of this technology is to balance the soil pressure and water pressure outside the hole wall by using the pressure generated by the mud column, through pouring mud with specific physical parameters (such as density, viscosity) into the hole. Before construction, technicians will calculate and set one or a group of mud performance parameters in advance according to the soil layer parameters provided by the geological survey report and engineering experience. During the drilling process, the operator mainly judges the general state of the hole wall by monitoring the mud loss, drilling torque and other indirect indicators, and manually adjusts the mud parameters or takes remedial grouting measures when obvious signs of instability are observed.

[0004] Although the existing technology can maintain the stability of the hole wall to some extent, there are still some deficiencies: first, the control accuracy and adaptability are insufficient. This is because the setting of the wall protection parameters is mainly based on static geological survey data and past experience, while the influence of the tide on the pore water pressure is dynamic and closely related to specific conditions such as stratum permeability. This open-loop control method based on fixed parameters cannot accurately respond to the dynamic changes of the hole pressure caused by the tide throughout the drilling process in real time, resulting in lag and deviation in the matching of the wall protection pressure. Second, its control method is passive and post. The existing technology relies on observation of instability, which means that control measures are only started when stability problems have occurred or are occurring. Due to the lack of prediction ability for risk changes in the future tide period, the construction party cannot plan the operation rhythm in advance and adjust the wall protection parameters in advance to deal with the upcoming high-risk period, so that the whole process is in a passive emergency response state. Finally, its emergency treatment measures lack pertinence. Traditional monitoring methods can only reflect the problems of the hole wall, but it is difficult to reveal the specific mode of failure, such as slow local shrinkage or sudden overall collapse. Therefore, the subsequent emergency grouting and other measures often use uniform and undifferentiated schemes, which may result in low efficiency or poor effect. SUMMARY

[0005] In view of the deficiencies of the prior art, the hole wall stability dynamic control system for the tidal stratum drilling process is provided, and the problem that the fixed parameters and the response lag are used for the hole wall control measures for the tidal stratum drilling in the prior art, the dynamic change of the hole pressure caused by the tide cannot be adapted, and thus the hole wall stability control is inaccurate and not timely is solved.

[0006] To achieve the above object, the present application is implemented by the following technical solutions:

[0007] The hole wall stability dynamic control system for the tidal stratum drilling process provided by the present application comprises:

[0008] The multi-source dynamic perception module is used for acquiring the measured pressure value in the drilling hole, the tidal water level elevation of the region and the space-time displacement field data of the hole wall soil layer in real time.

[0009] The data processing and intelligent decision module is used for receiving and processing the measured pressure value, the tidal water level elevation and the space-time displacement field data to generate the control instruction representing the hole wall stability regulation strategy.

[0010] The cooperative control and planning execution module is used for receiving and executing the control instruction to cooperatively regulate the hole wall and the drilling parameter in the drilling process, and realizing the dynamic control of the hole wall stability.

[0011] Preferably, the multi-source dynamic perception module comprises:

[0012] The in-hole pressure sensing unit is used for measuring the total pressure in the drilling hole by the pressure sensor arranged on the outer wall of the drilling rod to obtain the measured pressure value.

[0013] The regional tidal water level monitoring unit is arranged in the integrated monitoring device of the monitoring point capable of reflecting the overall hydrological dynamics of the region, and the integrated monitoring device comprises a positioning component for acquiring the geographic position information of the monitoring region, a water level measuring component for collecting the water level data of the monitoring region, and a calculation component for obtaining the tidal water level elevation of the monitoring region based on the geographic position information and the water level data.

[0014] The hole wall soil layer space-time displacement monitoring unit is used for continuously monitoring the hole wall soil layer displacement by the distributed optical fiber sensing network pre-embedded along the periphery of the drilling hole to obtain the space-time displacement field data.

[0015] Preferably, the data processing and intelligent decision module comprises:

[0016] a tidal and pore pressure dynamic mapping model unit, configured to calculate a model predicted total pressure based on the tidal water level elevation and a preset mud density parameter acquired by the multi-source dynamic perception module, through a hydrodynamic model taking into account both hydrostatic pressure and dynamic water pressure components;

[0017] a model self-calibration unit, configured to compare the measured pressure value acquired by the multi-source dynamic perception module with the model predicted total pressure to obtain a prediction error, and to correct online an internal parameter of the tidal and pore pressure dynamic mapping model unit based on the prediction error.

[0018] Preferably, the tidal and pore pressure dynamic mapping model unit calculates the model predicted total pressure through the following formula :

[0019]

[0020] wherein, is the model predicted total pressure; is the depth; is the time; is the density of mud in the borehole; is the gravitational acceleration; is the tidal water level elevation; is a dynamic water pressure coefficient in the internal parameter; is a rate of change of the tidal water level elevation.

[0021] Preferably, the model self-calibration unit corrects online the dynamic water pressure coefficient through the following formula:

[0022]

[0023] wherein, is the corrected dynamic water pressure coefficient; is the uncorrected dynamic water pressure coefficient; is the prediction error; is a learning rate; is a correction function based on the prediction error.

[0024] Preferably, the data processing and intelligent decision-making module further comprises:

[0025] a borehole wall stability real-time evaluation unit, which internally has a neural network model, configured to comprehensively process the measured pressure value, the spatiotemporal displacement field data, and the online corrected internal parameter, to real-time output a stability coefficient quantifying the stability degree of the borehole wall, and to generate the control instruction based on the stability coefficient.

[0026] ​Preferably, the data processing and intelligent decision module further comprises:

[0027] a control parameter dynamic optimization unit, configured to analyze an actual effect of the control instruction after the collaborative control and planning execution module executes the control instruction once, and optimize a strategy of the control instruction generated by the borehole wall stability real-time evaluation unit based on the actual effect by using a reinforcement learning algorithm.

[0028] Preferably, the data processing and intelligent decision module further comprises:

[0029] a control parameter dynamic optimization unit, configured to analyze an actual effect of the control instruction after the collaborative control and planning execution module executes the control instruction once, and optimize a strategy of the control instruction generated by the borehole wall stability real-time evaluation unit based on the actual effect by using a reinforcement learning algorithm.

[0030] The analysis of the actual effect by the control parameter dynamic optimization unit comprises: obtaining a change amount of a stability coefficient output by the borehole wall stability real-time evaluation unit within a preset time window after the control instruction is executed, and generating a reward signal representing a control effect according to the change amount of the stability coefficient.

[0031] The reinforcement learning algorithm updates the strategy of the borehole wall stability real-time evaluation unit by using the reward signal.

[0032] Preferably, the collaborative control and planning execution module specifically comprises:

[0033] a drilling and wall protection collaborative planning unit, configured to call the data processing and intelligent decision module to obtain a pressure prediction result for one or more future tidal cycles, and generate a time sequence operation plan including a high-speed drilling window and a stability priority window based on the pressure prediction result, and the execution of the control instruction follows the time sequence operation plan.

[0034] Preferably, the collaborative control and planning execution module further comprises:

[0035] a differentiated emergency response unit, the differentiated emergency response unit comprising:

[0036] a spatiotemporal deformation pattern recognition subunit, configured to analyze the spatiotemporal displacement field data obtained by the multi-source dynamic perception module to identify a failure mode of the borehole wall.

[0037] a differentiated grouting execution subunit, configured to select and start a preset emergency plan matched according to the failure mode of the borehole wall from a plurality of preset emergency plans including different grouting pressures and ranges.

[0038] The application provides a borehole wall stability dynamic control system for a tidal stratum drilling process, and has the following beneficial effects:

[0039] 1. The borehole wall stability dynamic control system for the tidal stratum drilling process can collect real-time measured pressure values, regional tidal water level elevations and hole wall soil space-time displacement field data in the borehole through a multi-source dynamic sensing module, thereby providing comprehensive and real-time basic data for the system; a data processing and intelligent decision module can perform in-depth processing and analysis based on the data, accurately identify the borehole wall stability state and potential risks, generate targeted control instructions, and avoid errors caused by experience-based judgment; and a collaborative control and planning execution module can receive the instructions, collaboratively control the wall protection and drilling parameters in the drilling process, realize dynamic control of the borehole wall stability, effectively deal with the borehole wall instability problems caused by the periodic changes of the water level and the complex stress of the stratum in the tidal stratum, improve the timeliness and accuracy of monitoring compared with the traditional manual monitoring and control mode, reduce the probability of accidents such as borehole wall collapse and diameter reduction, reduce the construction risk and cost, improve the construction efficiency and engineering quality, adapt to the complex and changeable tidal stratum environment, and ensure the safety, efficiency and stability of the drilling engineering.

[0040] 2. The data processing and intelligent decision module, especially the tidal and pore pressure dynamic mapping model unit and the model self-calibration unit, can continuously compare the measured pressure values obtained by the multi-source dynamic sensing module with the model predicted total pressure, and correct the dynamic water pressure coefficient and other parameters in the model online by using the prediction error. This makes the system's prediction of the pressure in the borehole dynamically adapt to the specific geological and hydrological conditions on site, and eliminates the dependence on fixed experience parameters, thereby improving the accuracy and adaptability of the borehole wall stability evaluation and prediction.

[0041] 3. The drilling and wall protection collaborative planning unit in the collaborative control and planning execution module can plan efficient drilling windows and stable priority windows based on the pressure prediction of the future tidal cycle. This proactive planning based on prediction enables the system to adjust the wall protection and drilling parameters in advance and gently before the high-risk tidal rise and fall period, changes the passive emergency response to active risk management, and improves the planning and safety of the entire drilling process.

[0042] 4. The differential emergency response unit in the collaborative control and planning execution module can accurately identify the failure mode of the borehole wall in combination with the space-time displacement field data provided by the multi-source dynamic sensing module. For example, the system can distinguish between local diameter reduction and overall creep, and accordingly start a matching emergency plan with different grouting pressures and ranges, thereby avoiding the blindness of traditional single emergency measures and improving the pertinence, timeliness and effectiveness of emergency disposal. BRIEF DESCRIPTION OF DRAWINGS

[0043] Figure 1 is a system structure schematic diagram of an embodiment of the present application.

[0044] Figure 2 is a method flow schematic diagram of an embodiment of the present application.

[0045] Figure 3 is an internal structure schematic diagram of a data processing and intelligent decision module of an embodiment of the present application.

[0046] Figure 4 is an internal structure schematic diagram of a cooperative control and planning execution module of an embodiment of the present application.

[0047] Figure 5 is an internal structure schematic diagram of a multi-source dynamic perception module of an embodiment of the present application.

[0048] Figure 6 is an internal structure schematic diagram of a remote monitoring and man-machine interaction module of an embodiment of the present application.

[0049] Figure 7 is a measured pressure and tidal water level dynamic curve diagram of an embodiment of the present application.

[0050] Figure 8 is a model predicted pressure curve diagram of an embodiment of the present application.

[0051] Figure 9 is a prediction error and tidal water level dynamic curve diagram of an embodiment of the present application. DETAILED DESCRIPTION

[0052] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the specification of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the scope of protection of the present application.

[0053] Figure 1 is a system structure schematic diagram according to an embodiment of the present application. As shown in Figure 1 , the borehole wall stability dynamic control system for the drilling process of the tidal stratum provided by the embodiment of the present application can include:

[0054] A multi-source dynamic perception module 10 is configured to acquire multi-dimensional physical quantity data of a drilling site in real time. The multi-dimensional physical quantity data can include: a measured pressure value provided by a borehole pressure sensor; a tidal water level elevation provided by a regional tidal water level monitoring unit 12; and a time-space displacement field data of a soil layer around the borehole provided by a distributed optical fiber sensing network. The data are sent to a data processing and intelligent decision-making module 20 as original state data;

[0055] The data processing and intelligent decision-making module 20 is electrically connected with the multi-source dynamic perception module 10, and is configured to process, operate and make decisions on the multi-dimensional physical quantity data to generate a control instruction. The control instruction refers to a set of machine executable instructions associated with coordinated regulation of a protection wall and drilling parameters. The control instruction is finally used to control core engineering parameters such as drilling rate, pump pressure and displacement of a mud circulating pump, and mud density which directly affect the stability of the borehole wall;

[0056] The coordinated control and planning execution module 30 is electrically connected with the data processing and intelligent decision-making module 20, and is configured to receive and execute the control instruction, and convert the control instruction into specific control actions on field devices. In addition, the coordinated control and planning execution module 30 feeds back its own execution state (for example, actual rotating speed of the mud pump, actual drilling speed of the drilling rig, etc.) to the data processing and intelligent decision-making module 20, to provide factual basis for higher-order control optimization (for example, reinforcement learning);

[0057] The remote monitoring and man-machine interaction module 40 is electrically connected with the aforementioned modules, and is configured to visually present data and states and provide a manual intervention interface. The visual presentation of data and states includes:

[0058] The visualization of data can include: drawing the tidal water level elevation into a tidal curve graph changing with time; comparing and displaying the measured pressure value and the model predicted pressure value in the same coordinate system; rendering the time-space displacement field data into a two-dimensional cloud chart or a heat map along the drilling depth and time distribution; and displaying the stability coefficient and the hydrodynamic pressure coefficient and other key internal parameters in the form of a dashboard or a number in real time.

[0059] The visualization of states can include: displaying the current borehole wall stability state (for example, stable state, early warning state, dangerous state) analyzed by the data processing and intelligent decision-making module 20 in real time in the form of a prominent traffic light color (green, yellow, red) or a text label; and displaying the field device running state (for example, normal drilling, speed reduction operation, and drilling circulation) fed back by the coordinated control and planning execution module 30. The visualized decision-making process and state enable an operator to intuitively and comprehensively master the overall situation of system operation;

[0060] The providing of the manual intervention interface means that the remote monitoring and human-computer interaction module 40 provides a graphical user interface (GUI) to allow authorized operators to bypass the automatic decision of the system in an emergency or special working condition, and directly send a manual intervention instruction to the data processing and intelligent decision module 20 or the collaborative control and planning execution module 30, such as forced shutdown, forced lifting of mud density, etc.

[0061] Figure 2 is a method flowchart according to an embodiment of the present application. As shown in the figure, the embodiment of the present application provides a borehole wall stability dynamic control method for a tidal formation drilling process, which can include the following steps: Figure 2

[0062] Step S1, performing multi-source dynamic perception, and uninterruptedly collecting key parameters affecting borehole wall stability;

[0063] Step S2, performing data processing and intelligent decision, analyzing, predicting and evaluating the collected key parameters, and generating a regulation and control strategy accordingly;

[0064] Step S3, performing collaborative control and planning execution, and collaboratively adjusting related equipment in the drilling process according to the regulation and control strategy.

[0065] In a specific workflow, the system of the embodiment of the present application first performs step S1 through the multi-source dynamic perception module 10. The multi-source dynamic perception module 10 uninterruptedly collects the measured pressure value in the borehole, the tidal water level elevation of the drilling site area, and the time-space displacement field data of the borehole wall soil layer distributed along the borehole periphery. These collected data are transmitted to the data processing and intelligent decision module 20 in real time. The area refers to the entire engineering site area where the drilling operation is located or corresponds to a complete hydrogeological unit.

[0066] After receiving the data, the data processing and intelligent decision module 20 performs step S2. In this step, the data processing and intelligent decision module 20 first utilizes the internal tidal and pore pressure dynamic mapping model unit to calculate a model predicted total pressure through a hydrodynamic model that takes into account both static water pressure and dynamic water pressure components, based on the received tidal water level elevation and preset mud density parameters.

[0067] Subsequently, the model self-calibration unit in the data processing and intelligent decision module 20 compares the model predicted total pressure with the measured pressure value provided by the multi-source dynamic perception module 10 to generate a prediction error. Based on the prediction error, the online learning algorithm is used to correct the internal parameters such as the dynamic water pressure coefficient in the hydrodynamic model, so as to realize the continuous self-adaptation of the hydrodynamic model. The online learning algorithm is embodied as an iterative formula for correcting the dynamic water pressure coefficient. ​

[0068] The data processing and intelligent decision module 20 includes a real-time evaluation unit for hole wall stability, which has a neural network model built in. The real-time evaluation unit for hole wall stability uses the measured pressure values, the spatiotemporal displacement field data, and the corrected model parameters to output a quantitative stability coefficient and finally form a control instruction representing a specific regulation strategy.

[0069] The neural network model built in the real-time evaluation unit for hole wall stability uses a multi-layer feedforward neural network in a preferred embodiment, and the specific structure is as follows:

[0070] The input layer is used to receive the preprocessed and normalized multidimensional physical quantity data. The number of neurons in the input layer corresponds to the number of input features, and specifically includes: a measured pressure value (1 neuron), which is the normalized measured pressure in the hole . Spatiotemporal displacement field data features (N neurons), since the original spatiotemporal displacement field data has a high dimension, key feature vectors (for example, maximum displacement value, mean and variance of displacement gradient, etc.) are extracted through principal component analysis (PCA) or convolutional autoencoder technology, forming N input features. Corrected model parameters (1 neuron), which are the hydrodynamic pressure coefficients corrected in real time by an online learning algorithm.

[0071] The hidden layer is set between the input layer and the output layer, and one or more hidden layers (for example, 2 hidden layers, each containing 64 neurons) are provided. The hidden layer uses its internal weight matrix and nonlinear activation function (for example, ReLU function) to perform complex cross, combination and transformation on the input features, so as to learn and mine the deep and nonlinear relationships between these different physical quantities.

[0072] The output layer includes 1 neuron and uses a Sigmoid activation function, and its only output is the quantitative stability coefficient.

[0073] The neural network model needs an offline training process to determine the optimal weight and bias parameters of its internal hidden layer. This training process is completed before system deployment, and the specific steps are as follows:

[0074] First, a training dataset needs to be prepared. A large amount of historical drilling data is collected and organized as training samples, which can come from actual engineering cases under similar geological conditions or from the results of numerical simulation. In historical engineering cases, drilling parameters, environmental monitoring data, and final engineering result records, such as whether a hole collapse or a pipe sticking occurs, are included. In terms of numerical simulation, finite element method or discrete element method can be used to simulate the drilling process in tidal strata under different working conditions, thereby generating a large amount of training data.

[0075] After the training dataset is prepared, each sample needs to be labeled by a geology and engineering expert. The labeling process is to assign a real stability coefficient value to the sample according to its corresponding engineering result. For example, for the working condition data that records a serious hole collapse, its label value is set to zero; for the working condition data that the drilling process is completely smooth, its label value is set to one; and for the working condition that appears a slight block or shrinkage, its label value may be set to zero point five.

[0076] After the data labeling is completed, the training dataset is input into the built neural network structure. A loss function, such as mean square error loss function, is set during the training process to measure the difference between the model's predicted stability coefficient and the expert's labeled real label. Through the back propagation algorithm and gradient descent-based optimization method, such as Adam optimizer, the model can adjust the connection weights of all neurons in the network from the output layer to the input layer iteratively to continuously improve the prediction accuracy.

[0077] After the training phase is completed, the model needs to be verified and deployed. The training and verification process will be repeated until the model's prediction error on the independent verification set converges to a small enough range. At this time, the model training is completed, and its internal weight and bias parameters are fixed and saved. Finally, this trained neural network model is deployed in the borehole wall stability real-time evaluation unit, thereby having the ability to perform online real-time evaluation of borehole wall stability in the actual drilling process.

[0078] In the real-time running phase of the system, from the output quantitative stability coefficient S to the formation of specific control instructions, the detailed working process is as follows:

[0079] In the real-time inference step, at each time step, the borehole wall stability real-time evaluation unit inputs the latest measured pressure, displacement field features, and corrected model parameters into the already trained neural network model for a forward propagation (inference), and instantaneously calculates the current stability coefficient S.

[0080] In the threshold judgment and state division step, a control instruction generation logic module is internally arranged in the unit. The module compares the stability coefficient S calculated in real time with a set of preset threshold values that can be adjusted by the user, thereby dividing the state of the current hole wall into different safety levels. For example: a safety threshold (e.g. 0.85); a danger threshold (e.g. 0.6);

[0081] In the instruction mapping and generation step, according to the interval in which the stability coefficient S is located, the control instruction generation logic module selects and generates the corresponding control instruction from a predefined state and instruction mapping table: if S , it is determined to be a stable state, and instruction code 0x01 is generated, which maintains the current drilling and wall protection parameters. If S , it is determined to be a warning state, and instruction code 0x02 is generated with specific adjustment parameters, such as the instruction: mud density is increased by 2%; drilling speed is reduced by 15%. If S , it is determined to be a dangerous state, and the highest priority instruction code 0x03 is generated, which is the instruction: immediately stop drilling and circulate the mud with maximum displacement.

[0082] In the instruction issuing step, the control instruction formed finally, which contains the instruction code and specific parameters, representing the specific control strategy, is sent to the collaborative control and planning execution module 30, which parses and executes it.

[0083] Through the above series of detailed and coherent work processes, the embodiment of the present application realizes a complete technical link from multi-source data sensing, to model adaptive correction, to deep learning intelligent evaluation, and finally to specific equipment control.

[0084] After the collaborative control and planning execution module 30 receives the control instruction generated by the data processing and intelligent decision-making module 20, step S3 is performed. In this step S3, the collaborative control and planning execution module 30 parses the control instruction into adjustment actions for specific equipment on site, such as adjusting the output pressure of the mud pump to change the wall protection effect, or adjusting the rotation speed and feed speed of the drilling machine to change the drilling parameters;

[0085] In addition, these adjustment actions can also include:

[0086] Direct regulation of mud density, which includes: issuing instructions to the mud mixing system on site through the weighting agent automatic adding unit (e.g. controlling the valve or screw feeder connected to the barite tank), adding weighting materials to the circulating mud in a predetermined amount to accurately increase the overall mud density, thereby directly enhancing the hydrostatic pressure support to the hole wall.

[0087] Fine adjustment of circulating rate, which includes: issuing instructions to the frequency drive of the mud pump to increase or decrease the circulating rate. Increasing the rate can increase the annular return velocity and the equivalent circulating density (ECD), thereby enhancing the dynamic pressure support of the borehole wall and the cleaning efficiency of the cuttings; while in the case of low formation pressure capacity, the rate can be appropriately reduced to prevent the formation from leaking.

[0088] Intelligent control of the weight on bit (WOB), which includes: issuing instructions to the automatic drilling system of the drilling rig or the driller's console to reduce or increase the weight on bit applied to the drill bit. In the early warning state of the borehole wall stability, actively reducing the weight on bit can reduce the mechanical impact and disturbance of the drill bit to the bottom and wall of the well.

[0089] Intervention on the rheological properties of the mud, which includes: when the downhole cleaning efficiency is detected to be insufficient, issuing instructions to the mud maintenance system to add viscosity reducers or viscosity enhancers to adjust the viscosity and yield value of the mud, thereby optimizing its ability to carry cuttings and preventing cuttings from depositing around the well wall to cause additional instability risks.

[0090] Combined action of wellbore treatment, which includes: when the system determines that wellbore treatment is needed, a combined action instruction is generated: first instruct the drilling rig to pause the feed of the drilling tool, then instruct the mud pump to circulate at a specific rate, and at the same time instruct the drilling rig drive system to start the short-distance, low-speed reciprocating motion (reaming) program of the drill string. This series of combined actions is beneficial to the treatment of unstable floating cuttings on the well wall and the improvement of the quality of the well wall mud cake.

[0091] By converting the intelligent decisions of the upper modules into coordinated and precise control of the above-mentioned various field devices, the embodiments of the present application realize dynamic, closed-loop and fine management of the borehole wall stability.

[0092] The implementation of the adjustment action will change the physical state of the borehole and its surroundings, and this change will be re-collected by the multi-source dynamic perception module 10 in the next step S1, thereby forming a complete closed loop of data collection, analysis and decision, control execution, and effect feedback.

[0093] Throughout the entire workflow, the remote monitoring and human-computer interaction module 40 continuously receives data and state information from other modules and visualizes it on the user interface, while the remote monitoring and human-computer interaction module 40 also provides a channel for authorized operators to issue manual intervention instructions.

[0094] As Figure 5As shown, the multi-source dynamic perception module 10 is the data input part of the system of the embodiment of the present application, which is configured to synchronously and continuously collect multiple physical quantities of the drilling site, to provide necessary and real-time input data for the subsequent data processing and intelligent decision module 20. In a specific embodiment, the multi-source dynamic perception module 10 specifically includes an in-hole pressure sensing unit 11, a regional tidal water level monitoring unit 12, and a hole wall soil space-time displacement monitoring unit 13.

[0095] In Figure 5 In order to more intuitively represent the monitoring objects of each sensing unit, the tides, wellbore pressure, and well wall displacement are schematically drawn. They respectively represent the physical phenomena or physical quantities to be perceived by each unit, wherein the water level change caused by tides is measured by the regional tidal water level monitoring unit 12, the wellbore pressure is measured by the in-hole pressure sensing unit 11, and the well wall displacement is measured by the hole wall soil space-time displacement monitoring unit 13. In addition, Figure 5 The data preprocessing and transmission unit 14 is also schematically included, which is used to preliminarily process (such as filtering and analog-digital conversion) the raw signals collected by the in-hole pressure sensing unit 11, the regional tidal water level monitoring unit 12, and the hole wall soil space-time displacement monitoring unit 13, and then uniformly package and transmit to the data processing and intelligent decision module 20.

[0096] The in-hole pressure sensing unit 11 is used to measure the wellbore pressure, which includes at least one high sampling frequency pressure sensor. The pressure sensor is installed at a predetermined depth position of the outer wall of the drill pipe, which is used to directly measure the total pressure inside the drilling hole at the position, which is a comprehensive embodiment of the in-hole hydrostatic pressure and the dynamic water pressure caused by tidal fluctuations. The in-hole pressure sensing unit 11 continuously outputs the measurement results as measured pressure values to the data processing and intelligent decision module 20.

[0097] The regional tidal water level monitoring unit 12 is used to measure the tidal water level, which includes a monitoring device integrating Beidou high-precision positioning technology and an electronic water level gauge. The monitoring device is arranged at a stable monitoring point that can represent the hydrogeological conditions of the entire engineering site. For example, the monitoring point can be located on a fixed trestle in the engineering site, or on a stable bedrock point of the nearby coast in hydraulic communication with the drilling area, which is used to continuously monitor the absolute elevation of the regional tidal water level. In order to ensure the consistency and comparability of the data, all water level measurements are referenced to a unified engineering reference surface. The regional tidal water level monitoring unit 12 outputs the water level elevation values in the continuous time series as the tidal water level elevation to the data processing and intelligent decision module 20.

[0098] The borehole wall soil space-time displacement monitoring unit 13 is used for measuring the displacement of the wall of the borehole, and comprises a distributed optical fiber sensing network arranged along the periphery of the borehole. The distributed optical fiber sensing network can sense the strain caused by the deformation of the soil at any point on the optical fiber, and calculate the radial displacement of the soil relative to the borehole.

[0099] By processing the data of the entire sensing network, the borehole wall soil space-time displacement monitoring unit 13 can obtain a two-dimensional displacement field with high spatial resolution distributed along the depth of the borehole and time . The borehole wall soil space-time displacement monitoring unit 13 outputs the two-dimensional displacement field as space-time displacement field data to the data processing and intelligent decision-making module 20, which is used to represent the deformation process and morphology of the soil around the wall of the borehole. The space-time displacement field data is a two-dimensional data set representing the distribution of the displacement of the soil around the wall of the borehole in space and time. The data set can be specifically understood as a data matrix.

[0100] For example, each row of the data matrix corresponds to a specific depth (spatial dimension) of a sensor arranged along the borehole, and each column corresponds to a specific data acquisition time point (time dimension). The value of the cell located at the i-th row and the j-th column in the matrix represents the radial displacement of the soil at the i-th depth at the j-th time point.

[0101] For example, the space-time displacement field data can be represented as: at time T1, the displacements at depths Z1, Z2, and Z3 are D11, D21, and D31, respectively; at the next time T2, the displacements at the same depths are updated to D12, D22, and D32. By analyzing this complete set of data field, the system can not only know whether the wall of the borehole has deformed, but also accurately identify the specific depth of deformation, the size of the deformation, and the rate of deformation development.

[0102] The data processing and intelligent decision-making module 20 is electrically connected to the multi-source dynamic perception module 10 at the input end and electrically connected to the collaborative control and planning execution module 30 at the output end. The data processing and intelligent decision-making module 20 is configured to receive and process the real-time data provided by the multi-source dynamic perception module 10, and generate control instructions representing specific control strategies through internal operation and evaluation.

[0103] Figure 3 is a schematic diagram of the internal structure of the data processing and intelligent decision-making module according to an embodiment of the present application. As Figure 3 shown, the data processing and intelligent decision-making module 20 comprises a tidal and pore pressure dynamic mapping model unit 21, a model self-calibration unit 22, a borehole wall stability real-time evaluation unit 23, and a control parameter dynamic optimization unit 24.

[0104] The tidal and pore pressure dynamic mapping model unit 21 contains a mathematical model based on hydraulic principles. This unit receives tidal level elevations from the regional tidal level monitoring unit 12 and calculates using the following formula to generate a model that predicts total pressure independently of measured pressure. The following formula decomposes the model's predicted total pressure into the sum of hydrostatic pressure and hydrodynamic pressure terms:

[0105]

[0106] In the formula, In depth and time The model predicts total pressure; The real-time density of the drilling mud inside the borehole can be input from an external system or preset. Let gravitational acceleration be , which is a physical constant. The tidal water level elevation is obtained by the regional tidal water level monitoring unit 12; To calculate the depth of the point; The dynamic water pressure coefficient is an internal parameter of the model that reflects the comprehensive hydrogeological conditions such as formation permeability and porosity. The rate of change of tidal water level with respect to time can be derived from... Obtained through difference calculation.

[0107] Regarding the borehole mud density in the formula It can be obtained in a variety of ways to adapt to construction sites with different levels of automation: In a preferred embodiment, The density is obtained through automated real-time measurement. The system integrates with the on-site mud circulation system, and online mud densitometers are installed at specific locations in the mud circulation pipeline, such as the inlet of the mud pump or the return tank after the vibrating screen. This densitometer can continuously and in real-time measure the density of the mud and use this real-time measurement as... The data is directly input into the tidal and pore pressure dynamic mapping model unit 21. This method can most accurately reflect the actual state of the drilling mud inside the pore. In another embodiment, The data is obtained through periodic manual measurement and input. At construction sites without online densitometers, on-site operators can use conventional mud hydrometers (e.g., mud balances) to periodically measure the circulated mud according to construction procedures. After measurement, the operator manually inputs the measured value into the system via the remote monitoring and human-machine interface module 40 of this invention. In a more simplified embodiment, A preset value can be set. The preset value corresponds to the target density value of the mud configured before the drilling operation starts. The value is manually updated when the mud batch is replaced or the mud proportioning needs to be adjusted according to the engineering instructions. In addition, the change rate of the tidal water level elevation in the formula , which is physically the change rate of the tidal water level elevation with respect to time .

[0108] In a specific embodiment, since the system collects a series of measurement values at discrete time points, the change rate is calculated by numerical differentiation. The calculation formula can be:

[0109]

[0110] In the formula, is the tidal water level elevation measured by the regional tidal water level monitoring unit 12 at the current time ; is the tidal water level elevation measured at the previous sampling time ; is the time interval between two consecutive samplings, which is a known system parameter (for example, 1 minute or 5 minutes) determined by the data collection frequency of the regional tidal water level monitoring unit 12. Through the formula, the system can calculate the rising or falling rate of the tidal water level in real time, which is the direct cause of the hydrodynamic pressure.

[0111] The model self-calibration unit 22 is electrically connected to the tidal and pore pressure dynamic mapping model unit 21. The model self-calibration unit 22 receives the measured pressure value provided by the pore pressure sensing unit 11 and compares the measured pressure value with the model predicted total pressure generated by the tidal and pore pressure dynamic mapping model unit 21 to obtain the prediction error between the two.

[0112] Subsequently, the model self-calibration unit 22 iteratively corrects the internal parameters of the tidal and pore pressure dynamic mapping model unit 21, in particular the hydrodynamic pressure coefficient , based on the prediction error using a preset online learning algorithm. In a specific embodiment, the correction is performed by the following formula:

[0113]

[0114] In the formula, is the corrected hydrodynamic pressure coefficient at time ; is the uncorrected hydrodynamic pressure coefficient at time .​ is a learning rate used to control the size of the correction step, in one embodiment, is a correction function based on the prediction error; is a prediction error, whose value is:

[0115]

[0116] wherein, is the measured pressure value provided at depth and time ; is the model predicted total pressure at depth and time ;

[0117] The learning rate can be obtained in one of the following two ways, depending on the complexity of the system design and the performance requirements:

[0118] In the first embodiment, the learning rate a is set as a fixed preset constant, whose value is not dynamically calculated during the system operation, but determined offline before the system is deployed. Specifically, an initial empirical value can be directly set by experts in the field according to the experience of similar formations or projects, for example, a is equal to 0.01; or parameter optimization can be performed based on historical data, that is, historical drilling data of the formation or similar formations is played back for testing, and different a values are tried within a certain range, for example, from 0.0001 to 0.1, by using grid search method, and finally the a value that makes the model prediction error converge fastest and has the best stability is selected as the constant used in the system. The advantage of this method is that the implementation process is simple, and the system has less computational overhead.

[0119] In the second embodiment, is a dynamic variable, and in this more optimal embodiment, the learning rate will dynamically change as the correction process proceeds, to achieve better convergence performance. For example, a larger learning rate is used to speed up convergence when the error is large at the beginning of the correction; while a smaller learning rate is used for fine adjustment when the model tends to be stable at the later stage of the correction, to prevent oscillation around the optimal value. At this time, is obtained through a calculation formula. In one specific embodiment, a time-based learning rate decay strategy can be used, whose calculation formula is: ; wherein, is the dynamic learning rate at the current time , which will replace the fixed in the foregoing correction formula; is a preset initial learning rate (for example, = 0.1), whose value can be determined by the method in the first embodiment; is a preset constant used to control the speed of decay (e.g., = 0.001); is the number of correction iterations or time steps that have been performed. In this way, the learning rate can start from a larger initial value and be smoothly reduced over time, thus achieving a more robust and high-speed model self-calibration process.

[0120] In a preferred embodiment, the correction function is an identity function. In this case, the correction function directly uses the prediction error itself for correction, i.e.,

[0121]

[0122] At this time, the online correction formula is specified as:

[0123]

[0124] wherein, is the corrected dynamic water pressure coefficient, which will be used as the new parameter value for model prediction of total pressure at the next time ); is the uncorrected dynamic water pressure coefficient, which is the parameter value used for model prediction of total pressure at the current time t; is the learning rate, which is a preset hyperparameter used to control the correction step size and convergence speed; is the correction function based on the prediction error, which can be an identity function or a sign function; is the prediction error, which refers to the difference between the model-predicted total pressure and the measured pressure value at time . The physical meaning of this approach is that the size of the correction is proportional to the size of the prediction error, and the direction is consistent with the direction of the prediction error. For example, when the measured pressure is greater than the predicted pressure, the error is positive, and the system will automatically increase the dynamic water pressure coefficient, and vice versa.

[0125] In another embodiment, the correction function can also be a sign function. In this case, the correction function only takes the sign (direction) of the prediction error, and ignores its specific size, i.e.,

[0126]

[0127] At this time, the online correction formula is specified as:

[0128]

[0129] wherein, is the prediction error; is a correction function based on the prediction error; is a sign function, whose operation rule is: when its input value is positive, the function output is +1; when the input value is negative, the output is -1; when the input value is zero, the output is 0; is the updated dynamic water pressure coefficient at time is the output of this round of calculation; is the dynamic water pressure coefficient at time is the input of this round of calculation; is the learning rate, which is a preset hyperparameter used to control the fixed step of each correction; is the sign of the prediction error, which represents the direction of correction. The feature of this way is that the step of each correction is fixed (determined by the learning rate ), and only the direction of the error is adjusted, which helps to avoid over-correction caused by sudden changes in error in some cases.

[0130] In the above manner, the model self-calibration unit 22 can continuously and automatically iteratively optimize the dynamic water pressure coefficient according to the deviation between the measured data and the model prediction, so that the tidal and pore pressure dynamic mapping model can be self-adapted to the specific and changing working conditions on site.

[0131] The measured pressure value is a physical quantity directly obtained by physical measurement means and dynamically changes with time and working conditions. In a specific embodiment, the measured pressure value is obtained by the borehole pressure sensing unit 11 in the multi-source dynamic perception module 10. In the first implementation, the borehole pressure sensing unit 11 adopts a pressure while drilling (PWD) system, the sensor of which is integrated in the bottom hole assembly (BHA) and installed at the position of the drill collar close to the drill bit. The system can measure the fluid pressure in the wellbore near the bottom hole in real time and continuously, and the pressure comprehensively reflects the combined action of the mud hydrostatic pressure, the equivalent circulating density (ECD), and the formation pore pressure. The measured pressure signal is encoded underground, transmitted to the surface receiver in the form of pressure waves along the mud column in the wellbore through the mud pulse telemetry technology, and decoded to obtain the real-time measured pressure value at the bottom hole. In the second implementation, the borehole pressure sensing unit 11 adopts a distributed optical fiber sensing network in the form of one or more special optical cables deployed along the drill pipe or casing. The optical cable can be attached to the outer wall of the drill pipe or pre-installed and fixed before the casing is lowered. The system is based on distributed acoustic sensing (DAS) or distributed strain sensing (DSS) technology, which inverts the external pressure at each position on the optical cable by emitting laser pulses into the optical fiber and analyzing the backscattered signals. Compared with point sensors, this method not only obtains single-point pressure information but also provides a continuous pressure profile along the entire well depth. The measured optical signal is directly transmitted to the demodulation equipment on the ground through the optical cable itself, and finally the high-resolution and high-frequency measured pressure value is obtained. In the third implementation, the borehole pressure sensing unit 11 adopts a wireline logging sensor, and the typical application scenario is when drilling is paused and intermediate logging is performed. The wireline logging sensor is accurately lowered to the target depth underground through the cable to achieve focused monitoring at a specific position. The sensor can directly measure the fluid pressure at the position and transmit the measured signal to the ground in real time through the cable, thereby obtaining the local measured pressure value in the wellbore. In summary, regardless of the implementation, the borehole pressure sensing unit 11 is used to convert the real total fluid pressure at a specific depth in the wellbore into a digitized signal that changes with time and can be received by the computer system, and transmit it to the data processing and intelligent decision-making module 20 as the most critical basis for subsequent model calibration and stability evaluation.

[0132] ​​​a BP (Back-Propagation) neural network model pre-trained by historical data. The input end of the borehole wall stability real-time evaluation unit 23 receives multiple dimensions of data, specifically including: the measured pressure value provided by the borehole pressure sensing unit 11, the time-space displacement field data (or its main feature vector) provided by the borehole wall soil layer time-space displacement monitoring unit 13, and the real-time corrected hydrodynamic pressure coefficient of the model self-calibration unit 22 .

[0133] The borehole wall stability real-time evaluation unit 23 performs nonlinear mapping and comprehensive processing on these input data, and the real-time output is a single stability coefficient quantifying the stability degree of the borehole wall In a preferred embodiment, the stability coefficient is normalized to the interval [0, 1], where =1 indicates that the borehole wall is in a completely stable state, =0 indicates that the borehole wall has lost stability or collapsed.

[0134] The measured pressure value provided by the borehole pressure sensing unit 11, the time-space displacement field data provided by the borehole wall soil layer time-space displacement monitoring unit 13, and the real-time corrected hydrodynamic pressure coefficient of the model self-calibration unit 22 and other model internal parameters.

[0135] The stability coefficient is then compared with one or more preset safety thresholds (such as a warning threshold and an alarm threshold), and the comparison result will be directly used as the basis for generating specific control instructions.

[0136] The borehole wall stability real-time evaluation unit 23 also has a control instruction generation logic module inside, which maps the value of the real-time output stability coefficient to specific and executable control instructions according to the preset threshold and rules. In a specific embodiment, the mapping rule can be divided into the following three levels:

[0137] stable state ( >0.85), when the stability coefficient is greater than the preset safety threshold (such as 0.85), the control instruction generation logic module generates a control instruction for normal drilling or maintaining the current parameters. After the instruction is sent to the cooperative control and planning execution module 30, the system will maintain the current borehole protection mud density and drilling rate parameters unchanged.

[0138] warning state (0.6≤ ≤0.85), when the stability coefficient When the drilling speed drops to the warning range, the control command generation logic module generates a preventative adjustment control command. This command includes specific parameter adjustment suggestions, such as increasing mud density by 2% or reducing drilling rate by 15%. Upon receiving this command, the collaborative control and planning execution module 30 will automatically fine-tune the mud pump or drilling rig drive system to implement this preventative measure, thereby eliminating potential instability risks at the outset.

[0139] Dangerous conditions ( <0.6), when the stability coefficient When the borehole wall temperature falls below a preset danger threshold (e.g., 0.6), it indicates that significant instability has occurred. At this point, the control command generation logic module will generate emergency control commands. These commands may include multiple coordinated operations such as immediately stopping drilling, circulating and weighting the drilling mud, and raising the drill string. The collaborative control and planning execution module 30 will execute this command with the highest priority to ensure construction safety and prevent major accidents such as borehole collapse.

[0140] Through the above methods, the embodiments of the present invention transform the complex and nonlinear evaluation results of hole wall stability into clear and hierarchical machine-executable instructions, realizing full-process automation from intelligent sensing and evaluation to closed-loop control.

[0141] The control parameter dynamic optimization unit 24 is used to perform high-order optimization of the control strategy of the entire system. After a control command is executed by the collaborative control and planning execution module 30, the control parameter dynamic optimization unit 24 will receive performance evaluation data about the control action, such as the trend of stability coefficient changes or the convergence of spatiotemporal displacement field data over a subsequent period of time.

[0142] The dynamic optimization unit 24 for control parameters uses a reinforcement learning algorithm, taking performance evaluation data as a reward signal, to iteratively adjust the strategy for generating control commands by the real-time evaluation unit 23 for borehole wall stability. Through this process, the system can autonomously learn which combination of control parameters can achieve the optimal control effect under different operating conditions, thereby realizing the self-evolution of the control logic.

[0143] The specific logic of the control parameter dynamic optimization unit 24 for implementing strategy optimization is as follows:

[0144] In defining the reward function, the control parameter dynamic optimization unit 24 first defines a reward function to quantify the actual effect of any control command generated by the hole wall stability real-time evaluation unit 23. In a specific embodiment, the reward function is defined as: the change in the stability coefficient S within a preset time window (e.g., the next 10 minutes) after the collaborative control and planning execution module 30 executes a control command. That is: wherein, R is the reward signal obtained at the moment when the control instruction is executed; K is the stability coefficient at the moment when the control instruction is executed; K is the stability coefficient at the moment before the control instruction is executed. The reward signal directly and quantitatively reflects whether the control action brings positive effect (R > 0) or negative effect (R < 0).

[0145] In the step of executing strategy optimization, the control parameter dynamic optimization unit 24 receives the reward signal R calculated above as the core input of the reinforcement learning algorithm (such as Q-learning or DeepQ-Network, etc.). The reinforcement learning algorithm updates the state-action mapping strategy inside the borehole wall stability real-time evaluation unit 23 according to the reward signal. If a control instruction (action) brings positive reward, the reinforcement learning algorithm will enhance the probability of selecting the instruction again under similar working conditions (state). Conversely, if a control instruction brings negative reward (punishment), the reinforcement learning algorithm will reduce the probability of selecting the instruction again under similar working conditions.

[0146] Through the iterative cycle of continuously executing instructions, analyzing effects (calculating rewards), and optimizing strategies, the control parameter dynamic optimization unit 24 can make the control strategy of the entire system evolve from initially relying on simple rules set by experts to the optimal control strategy that can adapt to specific formations and specific working conditions, thereby maximizing long-term borehole wall stability and achieving higher level of intelligent control.

[0147] The collaborative control and planning execution module 30 has its input end electrically connected with the data processing and intelligent decision-making module 20, and its output end connected with the control interface of the drilling and wall protection equipment (such as mud pump, grouting system, drilling machine PLC, etc.) in the field. The collaborative control and planning execution module 30 is configured to receive and execute the control instructions generated by the data processing and intelligent decision-making module 20, and to convert the abstract decision results into precise and collaborative adjustment actions for physical equipment.

[0148] As shown in Figure 4 , in one specific embodiment, the collaborative control and planning execution module 30 includes a drilling and wall protection collaborative planning unit 301, a multi-parameter collaborative adaptive adjustment unit 302, and a differentiated emergency response unit 303.

[0149] ​​​​The drilling and shoring co-planning unit 301 is configured to perform forward-looking job planning. The drilling and shoring co-planning unit 301 invokes the prediction function within the data processing and intelligent decision-making module 20 to obtain a predicted curve of the borehole pressure variation over one or more future complete tidal cycles. Based on the predicted curve, the drilling and shoring co-planning unit 301 automatically divides the future time period into high-speed drilling windows, in which the pressure fluctuation is gentle and the borehole wall stability risk is low, and stability-first windows, in which the pressure variation is dramatic and the borehole wall stability risk is high.

[0150] Subsequently, the drilling and shoring co-planning unit 301 generates a time-sequenced job plan that clearly labels the two types of windows. In the time-sequenced job plan, different macro job strategies are preset for different windows, for example, a higher drilling speed is allowed in the high-speed drilling window, while the drilling speed is reduced and the safety margin of the shoring parameters is increased in the stability-first window. That is, the drilling speed in the high-speed drilling window is higher than that in the stability-first window.

[0151] The high-speed drilling window refers to one or more future time periods in the time-sequenced job plan, in which the borehole wall hydrodynamic pressure caused by the tide is in the trough region or gentle variation region of its variation cycle, as predicted by the data processing and intelligent decision-making module 20. This corresponds to the time of slack tide or slow tide variation rate. In this window period, the formation pore pressure is least disturbed by the tide, the inherent stability of the borehole wall is higher, and the instability risk is lower. Therefore, the system plans to execute job parameters that aim to maximize drilling efficiency in this period, for example, higher drilling speed and drilling pressure are used.

[0152] The stability-first window refers to one or more future time periods in the time-sequenced job plan, in which the predicted borehole wall hydrodynamic pressure will experience dramatic fluctuations or be in the peak region of its variation cycle. This corresponds to the time of fastest tide rise or fall. In this stability-first window period, the formation pore pressure is most dramatically disturbed by the tide, and the risk of borehole wall instability is significantly increased. Therefore, the system plans to execute job parameters that prioritize absolute borehole wall stability in this period. This includes reducing the drilling speed and drilling pressure, or temporarily stopping the drilling operation and performing special shoring operations such as circulating weighted mud.

[0153] Through this long-term prediction-based planning, the embodiment of the present application realizes an upgrade from passive response to active planning, and can take a conservative strategy in advance before the arrival of a high-risk period, and fully release the drilling potential in a safe window period, thereby maximizing the overall operation efficiency under the premise of safety.

[0154] The higher drilling speed and the reduced drilling speed determination criterion and comparison basis are as follows:

[0155] The system first calculates and determines a baseline ROP (Rate of Penetration) in the current formation and the current equipment configuration according to a series of input parameters. The baseline ROP is the anchor point and comparison basis for all subsequent speed planning. In specific embodiments, the determination of the baseline ROP can comprehensively refer to multiple key information: it not only relies on the mechanical performance parameters of the drilling rig and the drilling tool, covering the maximum torque, speed of the drilling rig, and the maximum displacement of the mud pump and other hardware indicators, to provide equipment capability level basis constraints for drilling speed setting; it also needs to combine the design parameters of the drill bit, including the drill bit type, size, and the best operating parameter range recommended by the manufacturer, to ensure that the drilling speed is adapted to the working characteristics of the drill bit; at the same time, it can also be based on adjacent well data or geosteering models, and with the known geological information such as formation lithology and drillability grade, to calculate a drilling speed that is reasonable and economical, and finally determine the baseline ROP by integrating the above multi-dimensional information.

[0156] Based on the calculated baseline ROP, the speed strategy in different windows is defined as follows:

[0157] For the high-speed drilling window, allowing a higher drilling speed in this window means that the cooperative control and planning execution module 30 will set a target drilling speed that is definitely higher than the baseline ROP. For example, the target drilling speed is set to 115% to 140% of the baseline ROP. The goal of this strategy is to maximize footage efficiency at the expense of a certain equipment wear allowance during the period of good inherent stability of the borehole wall.

[0158] For the stability priority window, requiring a reduced drilling speed in this window means that the cooperative control and planning execution module 30 will set a target drilling speed that is lower than the baseline ROP. For example, the target drilling speed is set to 40% to 70% of the baseline ROP. In the case of predicting a very high risk of borehole wall instability, the target drilling speed can even be set to 0, that is, to perform a pure wall protection operation of pausing drilling and circulating reaming. The goal of this strategy is to sacrifice drilling efficiency to maximize the safety redundancy of borehole wall stability.

[0159] In this way, higher and lower are quantitative strategies with clear numerical intervals based on a dynamically calculated baseline value. This makes the entire time sequence operation plan precise, executable, and repeatable.

[0160] A multi-parameter coordinated adaptive regulation unit 302, which is used to execute the specific control instructions generated by the data processing and intelligent decision module 20 under the guidance of the above-mentioned time sequence operation plan. The multi-parameter coordinated adaptive regulation unit 302 communicates with the control system interface of multiple devices on site, parses the control instructions into synchronous regulation of a group of wall protection and drilling parameters. This group of wall protection and drilling parameters at least includes: mud density and circulating pressure adjusted by controlling the mud pump, grouting pressure adjusted by controlling the grouting system, and drilling speed and bit rotation speed adjusted by controlling the drilling rig frequency converter. The unit ensures that the regulation of these parameters is coordinated to achieve the optimal integrated control effect.

[0161] The differentiated emergency response unit 303 is enabled when the system determines that there is an immediate risk of instability of the borehole wall. The differentiated emergency response unit 303 internally includes a spatiotemporal deformation pattern recognition subunit and a differentiated grouting execution subunit.

[0162] The spatiotemporal deformation pattern recognition subunit is used to receive the spatiotemporal displacement field data provided by the borehole wall soil spatiotemporal displacement monitoring unit 13 of the multi-source dynamic perception module 10. The spatiotemporal deformation pattern recognition subunit analyzes the time evolution characteristics and spatial distribution patterns of the spatiotemporal displacement field data using built-in pattern recognition algorithms.

[0163] The time evolution feature specifically refers to the dynamic law and key trend presented by the displacement amount (i.e. deformation amount) of the monitoring point around the hole wall changing with time, which is used to describe the speed and acceleration characteristics in the deformation development process of the hole wall; in a specific embodiment, the analysis of the time evolution feature specifically includes the identification of one or more of the following modes: first, displacement rate (Displacement Rate), that is, the displacement increment of the monitoring point per unit time is obtained by calculation (for example, the unit can be millimeters / hour), wherein if the displacement rate is stable and at a very low level, it indicates that the formation where the hole wall is located is in a stable state, and if the displacement rate presents a continuous growth trend, it can be determined as a precursor signal of formation instability; second, displacement acceleration (Displacement Acceleration), which focuses on analyzing the change trend of the displacement rate. Practice shows that the continuous acceleration of the displacement amount is a typical and dangerous characteristic signal of the progressive destruction of the hole wall and eventually leads to the overall collapse, therefore, accurately identifying the inflection point of the hole wall deformation from the uniform speed stage to the acceleration stage is an important link to realize the early warning of hole wall instability; third, creep characteristics (Creep Characteristics), which are used to identify the deformation mode of the monitoring point displacement slowly and continuously increasing with time when the external load (mud pressure inside the hole) remains basically unchanged. This feature directly corresponds to the typical creep behavior of soft or plastic formation, and can be used as an important basis for judging the formation properties; fourth, periodic fluctuation characteristics (Periodic Fluctuation Characteristics), which are used to identify the regular and reciprocal hole wall deformation synchronized with the tidal period. With the help of this feature, the system can effectively distinguish between normal elastic deformation caused by tidal action and irreversible plastic deformation of the hole wall itself, and avoid the deviation of the warning caused by misjudgment of normal deformation; fifth, convergence trend (Convergence Trend), which is mainly applied after taking hole wall protection measures (such as increasing the mud density inside the hole), to judge the effectiveness of the hole wall protection measures by analyzing whether the displacement rate of the monitoring point gradually decreases and eventually tends to zero. If the displacement rate presents the above change trend, it indicates that the hole wall protection measures are effective and the hole wall is transitioning to a stable state.

[0164] The spatial distribution pattern refers to the distribution mode and shape of the displacement field around the entire borehole wall on a specific time section in the spatial geometric layer, which is mainly used to describe the specific position and form of borehole wall deformation; in a specific embodiment, the analysis of the spatial distribution pattern specifically covers the identification of one or more of the following modes: first, uniform convergence, i.e., the borehole wall at the same depth uniformly shrinks from all directions to the center, which indicates that the external pressure (e.g., ground stress) is greater than the internal support pressure, thereby causing the overall shrinkage of the borehole wall; second, asymmetric convergence, which is characterized by the fact that the borehole wall deformation is mainly concentrated in one or several directions, while the deformation in the remaining directions is relatively small, which strongly suggests that there is a dominant structural plane (e.g., joint, fracture, or weak interlayer) in the borehole wall, and the deformation of the borehole wall is developing along the structural plane; third, localized extrusion or bulging, which refers to the fact that at a specific depth, the borehole wall exhibits obvious and concentrated bulging deformation into the well, which corresponds to the existence of a high-plasticity weak interlayer (e.g., mudstone or shale layer) that is extruding into the well; fourth, shear slip surface, which requires the identification of one or more continuous regions that penetrate and have a sudden change in displacement, presenting a dislocation of one part of the borehole wall relative to another part, which is a characteristic of the formation of a shear failure surface and belongs to a more serious borehole wall instability mode; and fifth, radius of influence, which is determined by analyzing the range of borehole wall deformation extending into the formation, and if the radius of influence continues to expand, it indicates that the borehole wall instability region is expanding from the well wall to the deep formation.

[0165] Through comprehensive pattern recognition of the above-mentioned time evolution characteristics and spatial distribution patterns, the spatiotemporal deformation pattern recognition subunit can output higher-order judgment results with clear geomechanical significance (e.g., identifying that a local soft rock interlayer is in an accelerated creep stage), thereby providing a more profound and accurate basis for subsequent decision-making. For example, by analyzing the deformation rate, the extent of the influence range, and the geometric shape of the deformation region, and comparing them with a preset failure mode library, the current instability precursor can be identified as one of local shrinkage, overall creep, or sudden collapse.

[0166] The differential grouting execution subunit is used to receive the specific failure mode identified by the spatiotemporal deformation pattern recognition subunit. The differential grouting execution subunit internally stores an emergency plan library, where each emergency plan corresponds to a failure mode and specifies specific parameters such as grouting pressure, grouting range, and grout type.

[0167] Based on the received failure mode, the differentiated grouting execution subunit automatically selects and activates a matching one from the emergency plan library. For example, if a local narrowing is identified, a low-pressure permeable grouting plan for a specific depth is activated; if a sudden collapse is identified, a high-pressure jet grouting plan covering the entire risk section is activated, so as to achieve targeted handling of different risk scenarios.

[0168] The differentiated grouting execution subunit can automatically select and activate appropriate emergency response plans from the emergency plan library based on the identified failure modes, enabling targeted handling of various risk scenarios. Specifically, when a localized narrowing mode is identified, given its small deformation area and the fact that the stratum still possesses a certain degree of self-stabilization, the system will activate a low-pressure permeable grouting plan for a specific depth. If an overall creep mode is identified, considering its wide deformation range, long duration, and common occurrence in large sections of soft plastic strata, a segmented reinforcement grouting plan using long-sleeved valve pipes will be activated. When a sudden collapse mode is detected, due to its critical situation, rapid instability, and the formation of cavities, the system will quickly activate a high-pressure jet grouting plan covering the entire risk section. Through this differentiated plan execution mechanism based on precise pattern recognition, this invention overcomes the limitations of traditional uniform response methods, changes the drawbacks of passive emergency response, and enables precise, efficient, and economical management of complex borehole wall disasters.

[0169] The remote monitoring and human-machine interaction module 40 is electrically connected to the multi-source dynamic sensing module 10, the data processing and intelligent decision-making module 20, and the collaborative control and planning execution module 30. This remote monitoring and human-machine interaction module 40 is configured to provide a centralized visual presentation of the entire system's operating status and data flow, and to provide authorized operators with a monitoring interface and intervention point.

[0170] like Figure 6 As shown, in one specific embodiment, the remote monitoring and human-computer interaction module 40 integrates and displays the following content on the display device in the monitoring center through its internal panoramic visualization integration unit:

[0171] Real-time video signals from the work site;

[0172] Dynamic curves designed with dual vertical axes (dual Y-axis) Figure 7 This is used to visually compare the measured pressure and tidal water level collected by the multi-source dynamic sensing module 10 under the same time coordinate system.

[0173] The model-predicted stress map generated by the data processing and intelligent decision-making module 20 ( Figure 8 );

[0174] The prediction error analysis diagram also uses a dual Y-axis design ( Figure 9) for diagnosing the correlation between the model prediction error and the tidal water level;

[0175] The time-space displacement field data provided by the hole-surrounding soil layer time-space displacement monitoring unit 13 is presented in the form of a color-coded cloud chart, thereby realizing the conversion of massive and abstract displacement data into intuitive image information, reducing the cognitive load of the operator, enabling him to instantly locate the risk area through color and accurately diagnose specific failure modes such as overall convergence or local extrusion through the geometric morphology of the cloud chart, thereby providing technical support for the subsequent realization of fast and accurate differentiated emergency disposal decisions by the system.

[0176] and the graphical representation of the time-sequential operation plan generated by the collaborative control and planning execution module 30.

[0177] Figure 7 The horizontal axis unified by time, together with the left and right two coordinate axes with completely different dimensions and numerical ranges, respectively represents the tidal water level elevation (m) and the measured pressure value (kPa), thereby enabling precise and intuitive comparison and display of the dynamic change relationship between the two key physical quantities within the same view.

[0178] Figure 7 The dynamic characteristics of the curve chart are embodied in the time attribute of the data, with the horizontal axis of the curve chart set as the time axis, showing the continuous time flow from a certain time in the past (for example, 48 hours ago) to the current real-time time, and each data point on the curve corresponding to the measurement value at a specific time point. On the other hand, it is embodied in the real-time updating of the view. The curve chart is not a static historical record picture, but a dynamic view generated on the screen in real time. As the multi-source dynamic perception module 10 uploads new measurement data at a fixed period (for example, every minute), the curve chart will continuously plot new data points and push the curve to the right, thereby enabling the remote monitoring personnel to intuitively and in real time observe the complete process, current state and change trend (including the rise and fall of the numerical value, the speed of change, etc.) of the physical quantities such as pressure and water level over time.

[0179] Consistent with the design idea of Figure 7 Figure 9 ​The shown prediction error graph is also designed with double Y-axes, whose core purpose is to deeply diagnose the prediction performance of the model. This graph displays the prediction error (kPa) (usually corresponding to the right Y-axis) and the tidal water level elevation (m) (usually corresponding to the left Y-axis) in the same time coordinate, through which it can be observed that although the tidal water level presents obvious periodic fluctuations (as shown by the dashed line), the prediction error curve of the model (as shown by the solid line) only presents random, non-trend fluctuations around the zero-value baseline, and the fluctuation form thereof does not have any fixed correlation with the high-low changes of the tidal water level, proving that the self-calibration function of the data processing and intelligent decision module 20 in the present application is efficient and successful, which can effectively identify and compensate the systematic influence caused by external environmental factors such as tides, so as to control the final prediction error of the model within the range of pure random noise, ensuring the high precision and high reliability of the prediction result.

[0180] As shown in Figure 6 The remote monitoring and man-machine interaction module 40 directly presents the internal judgment result of the data processing and intelligent decision module 20 through the intelligent decision transparency unit therein. For example, the stability coefficient value output by the borehole wall stability real-time evaluation unit 23 is displayed in the form of a dashboard or a numerical value, and the specific failure mode identified by the spatiotemporal deformation mode identification subunit is highlighted in the form of text or icons.

[0181] As shown in Figure 6 The remote monitoring and man-machine interaction module 40 also includes a remote intervention and permission management unit, which functions as an instruction arbitration and formatting center. It receives intervention requests from remote operators or directly receives operation instructions from the on-site machine-side operation box, and according to the internal permission and priority logic, finally generates a standardized control instruction that can be directly executed by the collaborative control and planning execution module 30.

[0182] To ensure the final safety of the operation, the system follows the principle of local control priority. This principle is realized through hardware or software interlocking. The on-site operator can issue operation instructions through the machine-side operation box, and the execution priority thereof is higher than any instruction issued by the remote monitoring and man-machine interaction module 40. The communication of the entire system is established on the redundant industrial Ethernet and executed by the redundant programmable logic controller (PLC), so as to ensure the reliability and safety of the control instruction transmission.

[0183] While embodiments of the application have been shown and described, it is to be understood that the embodiments described are merely exemplary of the principles and application of the present application. Numerous modifications and adaptions can be effected without departing from the spirit and scope of the present application, which is not limited to the exact construction and arrangement described. It is intended, therefore, to cover all modifications and adaptions that fall within the scope of the claims and their equivalents.

Claims

1. A system for dynamic control of borehole stability during drilling in tidal strata, characterized in that The system comprises: A multi-source dynamic perception module for acquiring real-time measured pressure values in the borehole, tidal water level elevations of the region, and spatiotemporal displacement field data of the soil layer around the borehole; A data processing and intelligent decision-making module for receiving and processing the measured pressure values, the tidal water level elevations, and the spatiotemporal displacement field data to generate control instructions representing a borehole wall stability regulation strategy; A collaborative control and planning execution module for receiving and executing the control instructions to collaboratively regulate borehole wall protection and drilling parameters during drilling to achieve dynamic control of borehole wall stability. The multi-source dynamic perception module comprises: A borehole pressure sensing unit for measuring the total pressure in the borehole in real time by deploying pressure sensors on the outer wall of the drill pipe to obtain the measured pressure values; A regional tidal water level monitoring unit arranged in an integrated monitoring device at a monitoring point that can reflect the overall hydrological dynamics of the region, the integrated monitoring device comprising a positioning component for acquiring geographic location information of the monitoring region, a water level measuring component for collecting water level data of the monitoring region, and a calculation component for obtaining the tidal water level elevation of the monitoring region based on the geographic location information and the water level data; A borehole wall soil layer spatiotemporal displacement monitoring unit for continuously monitoring the displacement of the soil layer around the borehole by a distributed optical fiber sensing network pre-embedded along the borehole periphery to obtain the spatiotemporal displacement field data; The data processing and intelligent decision-making module comprises: A tidal and borehole pressure dynamic mapping model unit for obtaining a model predicted total pressure by a hydrodynamic model that simultaneously accounts for static water pressure and dynamic water pressure components based on the tidal water level elevations and preset mud density parameters; A model self-calibration unit for comparing the measured pressure values obtained by the multi-source dynamic perception module with the model predicted total pressure to obtain a prediction error, and online correcting internal parameters of the tidal and borehole pressure dynamic mapping model unit based on the prediction error; The tidal and pore pressure dynamic mapping model unit calculates the model predicted total pressure by the following equation : ； wherein is the model predicted total pressure; is the depth; is the time; is the density of the mud in the borehole; is the acceleration due to gravity; is the tidal water level elevation; is the internal parameter of the hydrodynamic pressure coefficient; is the rate of change of the tidal water level elevation; The model self-calibration unit corrects the hydrodynamic pressure coefficient online by the following formula online ； wherein is the corrected dynamic water pressure coefficient; is the uncorrected dynamic water pressure coefficient; is the prediction error; is the learning rate; is a correction function based on the prediction error; The collaborative control and planning execution module specifically comprises: A drilling and wall protection collaborative planning unit for calling the data processing and intelligent decision-making module to obtain pressure prediction results for one or more future tidal cycles, and generating a timing operation plan containing a high-speed drilling window and a stability priority window based on the pressure prediction results, the execution of the control instructions following the timing operation plan; the drilling speed in the high-speed drilling window is higher than that in the stability priority window.

2. The system for dynamic control of borehole wall stability for tidal strata drilling operations of claim 1, wherein, The data processing and intelligent decision-making module further comprises: A borehole wall stability real-time evaluation unit with a neural network model for comprehensively processing the measured pressure values, the spatiotemporal displacement field data, and the online corrected internal parameters, real-time outputting a stability coefficient quantifying the stability degree of the borehole wall, and generating the control instructions based on the stability coefficient.

3. The system for dynamic control of borehole wall stability for tidal strata drilling operations of claim 2, wherein, The data processing and intelligent decision-making module further comprises: The control parameter dynamic optimization unit is configured to analyze actual effects of the control instruction after the collaborative control and planning execution module executes the control instruction once, and optimize a strategy of the control instruction generated by the hole wall stability real-time evaluation unit based on the actual effects by using a reinforcement learning algorithm.

4. The system for dynamic control of borehole wall stability for tidal strata drilling operations of claim 3, wherein, The analysis of the actual effects by the control parameter dynamic optimization unit includes: obtaining a variation of a stability coefficient output by the hole wall stability real-time evaluation unit within a preset time window after the control instruction is executed, and generating a reward signal representing a control effect according to the variation of the stability coefficient. The reinforcement learning algorithm is configured to update the strategy of the hole wall stability real-time evaluation unit by using the reward signal.

5. The system for dynamic control of borehole wall stability for tidal strata drilling operations of claim 1, wherein, The collaborative control and planning execution module further includes: The differentiated emergency response unit includes: A spatiotemporal deformation pattern recognition subunit is configured to analyze the spatiotemporal displacement field data obtained by the multi-source dynamic perception module to identify a failure mode of the hole wall. A differentiated grouting execution subunit is configured to select and start a preset emergency plan matched with the failure mode of the hole wall from a plurality of preset emergency plans containing different grouting pressures and ranges.

Images