An ultra-deep well drilling method based on a spiral gage PDC bit

Abstract

The application discloses a kind of ultra-deep well drilling methods based on helical gear PDC drill bit, it is related to deep well drilling technical field, comprising the following steps: receiving by multi-source decision model and concurrent generation for the parameter optimization instruction of speed-up, leakage prevention and anti-stuck target;Based on real-time downhole working condition data, calculate dynamic priority for multiple concurrent generated instructions;When conflict is identified, generate the current drilling parameter state snapshot with unique session identification, and the conflicting instruction is sorted according to dynamic priority and generates an ordered execution sequence;Execute the instruction in execution sequence in order, and monitor the working condition index after execution in the preset monitoring period;If the working condition index is monitored to deteriorate more than the safety threshold, then generate reverse adjustment instruction based on state snapshot to roll back, and adjust the priority strategy of subsequent instruction again;The application avoids the dramatic oscillation of drilling pressure, rotating speed and other parameters caused by the mutual offset of parameter adjustment effects of different instructions.

Description

Technical Field

[0001] This invention relates to the field of deep well drilling technology, specifically to an ultra-deep well drilling method based on a spiral toothed PDC drill bit. Background Technology

[0002] In oil and gas resource exploration and development, ultra-deep well drilling is a key means of exploiting deep oil and gas reservoirs. It usually refers to drilling operations with a depth of more than 6,000 meters. It is necessary to deal with complex formation conditions and extreme downhole environments. Taking Xinke 6 well as an example, its drilling depth reached 10,393 meters and encountered 12 strata from the Quaternary to the Cambrian. Among them, the Permian igneous rocks have well-developed fractures, the Silurian mudstone is prone to collapse, and the Carboniferous may have gypsum-salt rocks with narrow diameter. Moreover, the formation safety density window is narrow, which makes it easy to cause complex situations such as leakage and overflow. Therefore, it puts forward strict requirements on drilling efficiency and well control safety.

[0003] The spiral-tooth PDC drill bit is a core rock-breaking tool adapted to complex formations in ultra-deep wells. Its cutting teeth are arranged in a logarithmic spiral to form a negative rake angle structure, which can enhance the hydraulic cleaning effect and effectively improve the rock-breaking efficiency of fractured rocks and soft-hard interbedded formations. At the same time, it reduces cutting vibration. Compared with traditional PDC drill bits, it achieved an average mechanical drilling rate of 7.49 m / h in the second well section in the Xinke block application, which is significantly better than conventional drill bits.

[0004] However, existing drilling control systems are not fully adapted to their operational characteristics. When multiple commands for speed-up, leak prevention, and stuck-out prevention are generated almost simultaneously, this scheduling method will lead to two consequences: First, the settings of parameters such as drilling pressure and rotation speed of the execution system will oscillate violently. The parameter adjustment directions of different commands are contradictory, resulting in a canceling effect, which will damage the mechanical drilling speed and aggravate tool wear. Second, in order to maintain parameter stability, commands are queued, causing the system response to be unable to keep up with the rapid changes in real-time operating conditions, resulting in control lag.

[0005] When conflict commands trigger complex downhole situations, the system cannot quickly return to the safe state before the command was executed, and manual intervention is required. This causes the closed loop of automated control to fail, significantly increasing the risk and cost of drilling operations. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides an ultra-deep well drilling method based on a spiral toothed PDC drill bit.

[0007] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0008] This invention provides an ultra-deep well drilling method based on a spiral toothed PDC drill bit, comprising the following steps:

[0009] Receive parameter optimization instructions generated concurrently by a multi-source decision model for the goals of speed improvement, leakage prevention, and jamming prevention;

[0010] Based on real-time downhole operating data, a priority calculation model including basic weight, real-time operating condition adjustment coefficient and time decay coefficient is adopted to calculate dynamic priority for multiple concurrently generated instructions. At the same time, conflicting instructions are identified based on a predefined instruction conflict degree matrix.

[0011] When a conflict is identified, a snapshot of the current drilling parameter status with a unique session identifier is generated, and the conflicting instructions are sorted according to dynamic priority and an ordered execution sequence is generated.

[0012] The instructions in the execution sequence are executed sequentially, and the working condition indicators after execution are monitored within a preset monitoring period;

[0013] If the detected operating condition indicators deteriorate beyond the safety threshold, a reverse adjustment instruction is generated based on the state snapshot to roll back the operation and the priority strategy of subsequent instructions is readjusted.

[0014] As a preferred technical solution of the present invention, a priority calculation model including basic weights, real-time operating condition adjustment coefficients, and time decay coefficients is adopted, specifically including:

[0015] Basic weights are assigned to the three types of instructions: speed-up, leak prevention, and jam prevention.

[0016] Based on the pressure window margin and the rate of change of mechanical drilling speed in the real-time operating data, the real-time operating condition adjustment coefficient is dynamically calculated.

[0017] When the pressure margin is lower than the first threshold, the real-time operating condition adjustment coefficient of the leak prevention command is increased to the first coefficient; when the mechanical drilling speed decrease rate is higher than the second threshold, the real-time operating condition adjustment coefficient of the speed-up command is increased to the second coefficient.

[0018] The time decay coefficient is calculated based on the generation timestamp of each instruction, and the delay time is positively correlated with the degree of decay.

[0019] The dynamic priority of the instruction is obtained by multiplying the basic weight, the real-time operating condition adjustment coefficient, and the time decay coefficient.

[0020] As a preferred embodiment of the present invention, identifying conflicting instructions based on a predefined instruction conflict degree matrix specifically includes:

[0021] The rows and columns of the conflict degree matrix represent the adjustment directions of the three parameters: drilling pressure, rotation speed, and displacement, respectively. The matrix element values ​​are the probabilities of the corresponding parameter adjustment combinations leading to the deterioration of the working conditions, obtained based on historical data statistics.

[0022] Calculate the conflict degree value corresponding to the conflict degree matrix for the direction of parameter adjustment between instructions;

[0023] If a conflict value exceeds the preset conflict threshold, the instructions are determined to be conflicting.

[0024] As a preferred embodiment of the present invention, generating a snapshot of the current drilling parameter status with a unique session identifier specifically includes:

[0025] When a command conflict is detected, immediately record the current drilling pressure, rotation speed, displacement, and well depth data;

[0026] The recorded data is associated with a unique session identifier and stored to form a traceable state snapshot.

[0027] As a preferred embodiment of the present invention, conflicting instructions are sorted according to dynamic priority and an ordered execution sequence is generated, specifically including:

[0028] The instruction with the highest dynamic priority is selected as the currently executed instruction.

[0029] The remaining conflicting instructions are placed into the delay queue in order of their dynamic priority.

[0030] As a preferred embodiment of the present invention, the instructions in the execution sequence are executed sequentially, and the working condition indicators after execution are monitored within a preset monitoring period, specifically including:

[0031] Execute the currently executing instruction;

[0032] Within the preset monitoring time window, assess whether the key operating condition indicators have achieved the expected goals of the instruction;

[0033] For the leak prevention target, the leakage amount is reduced to below the third threshold and remains stable; for the jam prevention target, the torque fluctuation range is reduced to below the fourth threshold; and for the speed increase target, the mechanical drilling speed increase rate reaches the fifth threshold.

[0034] If the expected goal is achieved and no new high-risk events are added, the instruction with the next priority is taken from the delay queue, and after verifying the compatibility of its parameter adjustment with the current state, it is used as the new current execution instruction.

[0035] As a preferred embodiment of the present invention, the rollback is performed by generating a reverse adjustment instruction based on the state snapshot, specifically including:

[0036] Get the most recently generated valid state snapshot;

[0037] Calculate the difference between the current drilling parameter status and the target status recorded in the status snapshot;

[0038] A parameter recovery command is generated based on the difference value. This command controls the drilling parameters to be adjusted step by step to the target state within a preset safe time window.

[0039] As a preferred embodiment of the present invention, the priority strategy for subsequent instructions is readjusted, specifically including:

[0040] If the same type of instruction triggers a rollback operation a set number of times within a preset time period, the basic weight of that type of instruction will be automatically reduced.

[0041] As a preferred embodiment of the present invention, the ultra-deep well drilling method further includes:

[0042] Periodically check the timeliness of the operating condition data on which each instruction in the delay queue is based;

[0043] If the operating condition data on which an instruction is based is outdated, the instruction is removed from the queue and an update request is sent.

[0044] As a preferred embodiment of the present invention, the multi-source decision model includes a geological model, a mechanical model, and a hydraulic model, wherein the geological model, mechanical model, and hydraulic model independently generate optimization instructions based on different data sources;

[0045] The instructions generated by the geological model, mechanical model, and hydraulic model are asynchronously published via the event bus.

[0046] The beneficial effects of this invention are:

[0047] 1. In this invention, the conflict degree matrix based on historical statistics is used to quantitatively evaluate the conflict between instructions. After the conflict is confirmed, the instruction with the highest dynamic priority is used as the current execution instruction, and the remaining conflicting instructions are placed into the delay queue according to the dynamic priority order. This mechanism transforms the parallel instruction stream that originally overlapped in time and conflicted in parameter space into an ordered execution sequence in the time dimension. This avoids the violent oscillation of parameters such as drilling pressure and rotation speed caused by the mutual cancellation of the parameter adjustment effects of different instructions. At the same time, since the instructions in the sequence are sorted and connected based on the dynamic priority calculated based on real-time working conditions, rather than simply queued, the system can respond to changes in real-time working conditions and effectively solve the control lag problem caused by fixed queuing.

[0048] 2. In this invention, upon identifying the critical decision point of command conflict, the system immediately records complete drilling parameters and generates a traceable state snapshot. When a deterioration in the working condition is detected within a preset monitoring period, the system calculates the state difference based on this state snapshot and generates step-by-step parameter recovery commands. This provides the system with precise rollback points and recovery paths, enabling it to quickly and safely roll back to the previous stable state without relying on manual intervention when the execution of conflict commands causes complex downhole situations, thereby maintaining the effectiveness of the automated control closed loop. Attached Figure Description

[0049] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0050] Figure 1 This is a schematic diagram of the workflow of the ultra-deep well drilling method based on the spiral toothed PDC drill bit of the present invention. Detailed Implementation

[0051] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0052] like Figure 1 As shown, an ultra-deep well drilling method based on a spiral toothed PDC drill bit includes the following steps:

[0053] It receives parameter optimization instructions generated concurrently by the multi-source decision model for the goals of speed improvement, leakage prevention, and sticking prevention. The multi-source decision model encapsulates the generated instructions into structured instruction events and publishes them asynchronously through the event bus. At the same time, real-time collected downhole sensor data (such as pressure, torque, well depth, etc.) are also encapsulated into data events and published on the bus.

[0054] Based on real-time downhole operating data, a priority calculation model including basic weight, real-time operating condition adjustment coefficient and time decay coefficient is adopted to calculate dynamic priority for multiple concurrently generated instructions. At the same time, conflicting instructions are identified based on a predefined instruction conflict degree matrix.

[0055] When a conflict is identified, a snapshot of the current drilling parameter status with a unique session identifier is generated, and the conflicting instructions are sorted according to dynamic priority and an ordered execution sequence is generated.

[0056] The instructions in the execution sequence are executed sequentially, and the working condition indicators after execution are monitored within a preset monitoring period;

[0057] If the detected operating condition indicators deteriorate beyond the safety threshold, a reverse adjustment instruction is generated based on the state snapshot to roll back the operation and the priority strategy of subsequent instructions is readjusted.

[0058] Through the above technical solution, the present invention effectively solves the problem of parameter oscillation and control lag caused by multi-target command conflicts during drilling in complex formations of ultra-deep wells, establishes a complete state traceability and safety backtracking mechanism, and significantly improves the automation level and safety of the drilling process.

[0059] It should be noted that this application also includes a dynamic priority arbitrator, which, as the core subscriber of the event bus, listens to all instruction events and data events in real time. Its core responsibility is to perform conflict detection, dynamic priority calculation and instruction serialization when multiple instruction events arrive at almost the same time or conflict with the currently executed instruction.

[0060] The State Snapshot Manager is responsible for capturing and storing a complete data snapshot of the system's current state at critical decision points (such as when arbitration begins). This snapshot contains instantaneous values ​​of key drilling parameters such as drill pressure, rotation speed, displacement, and well depth, and is associated with a unique session identifier.

[0061] And the intelligent driller system, which, as the final execution unit of the instructions, receives a serialized single control instruction output by the dynamic priority arbitrator and drives the drilling rig to perform corresponding parameter adjustments.

[0062] Furthermore, a priority calculation model is adopted, which includes basic weights, real-time operating condition adjustment coefficients, and time decay coefficients, specifically including:

[0063] When initializing the system or configuring it for a specific drilling stage, set basic weights for the three types of commands: speed-up, leakage prevention, and jamming prevention.

[0064] The base weight is a preset static coefficient used to reflect the inherent relative importance of different optimization objectives under a specific geological background. Its value comes from drilling engineering experience and the geological characteristics of the block. For example, in a block where fractures are known to be developed, the base weight of the leak prevention command can be set to a higher value (such as 0.7) to prioritize well control safety; while in a relatively stable formation, the base weight of the acceleration command can be appropriately increased (such as 0.5) to optimize efficiency. These weight values ​​are stored in the system's configuration table for use during calculation.

[0065] The dynamic priority arbiter continuously subscribes to the latest real-time downhole condition data via the event bus. Based on the pressure window margin and rate of change of mechanical drilling speed in the real-time condition data, it dynamically calculates the real-time condition adjustment coefficient. Its calculation logic is based on a set of predefined condition rules.

[0066] When the pressure margin is lower than the first threshold, the real-time operating condition adjustment coefficient of the leak prevention command is increased to the first coefficient. For example, if the first threshold is set to 10% and the first coefficient is 2.0, this means that when the real-time monitored formation pressure safety window margin is less than 10%, the system determines that the risk of leakage increases sharply. At this time, the real-time operating condition adjustment coefficient of any leak prevention command will be dynamically adjusted to 2.0, thereby significantly increasing its urgency.

[0067] When the rate of decrease in mechanical drilling speed exceeds the second threshold, the real-time operating condition adjustment coefficient of the speed-up command is increased to the second coefficient. For example, if the second threshold is set to 30% and the second coefficient is 1.8, this means that when the mechanical drilling speed decreases rapidly (e.g., exceeding 30%), the system judges that the drilling efficiency is severely impaired. At this time, the urgency of speed-up commands increases, and the real-time operating condition adjustment coefficient is adjusted to 1.8.

[0068] This dynamic adjustment mechanism ensures that the system can automatically adjust its decision priorities based on real-time risk conditions, prioritizing the allocation of limited control resources to the most pressing safety or efficiency objectives. The first threshold, second threshold, first coefficient, and second coefficient can all be configured and adjusted according to actual drilling process requirements.

[0069] To ensure that the system prioritizes the processing of the latest decisions, the priority calculation model introduces a time decay coefficient based on the generation timestamp of each instruction, and the delay time is positively correlated with the degree of decay.

[0070] The time decay coefficient is a timeliness adjustment factor calculated based on the generation timestamp of each instruction. It is used to reflect the timeliness value of instruction decisions. Specifically, each instruction event has its generation timestamp. When calculating, the dynamic priority arbitrator will obtain the current system time and calculate the delay time since the instruction was generated.

[0071] The design principle of the time decay coefficient is that the delay time is positively correlated with the degree of decay. That is, the longer the time it takes to wait for the instruction to be executed after it is generated, the more its priority score should be decayed. Optionally, a decay rate (e.g., 5%) per unit time (e.g., 1 second) can be set. Then the time decay coefficient = max(0, 1 - decay rate × delay time).

[0072] Finally, the basic weight, real-time operating condition adjustment coefficient, and time decay coefficient are multiplied together to obtain the dynamic priority of the instruction;

[0073] This calculation model comprehensively considers three dimensions: the inherent importance of the target, the urgency of the current working condition, and the timeliness of the decision-making, forming an adaptive priority evaluation system. In practical applications, when the system receives multiple conflicting instructions at the same time, this priority calculation model can accurately identify the instruction that should be executed first, thereby optimizing drilling efficiency while ensuring well control safety.

[0074] Taking drilling into a high-pressure fracture zone as an example, assuming the basic weights of the system configuration are: leakage prevention 0.6, sticking prevention 0.3, and speed increase 0.1, at this time, the drilling platform generates three commands almost simultaneously: leakage prevention command A (reduce displacement), sticking prevention command B (increase rotation speed), and speed increase command C (increase drilling pressure). When the dynamic priority arbitrator receives these commands, the real-time operating condition data obtained from the event bus shows that the pressure window margin is 8% (lower than the first threshold of 10%), and the mechanical drilling rate decrease rate is 25% (not reaching the second threshold of 30%).

[0075] For instruction A (leak prevention): base weight 0.6 × real-time operating condition adjustment coefficient 2.0 (due to pressure margin < 10%) × time decay coefficient 1.0 (assuming no delay) = dynamic priority 1.2;

[0076] For instruction B (anti-blocking): base weight 0.3 × real-time operating condition adjustment coefficient 1.0 (operating condition not triggered adjustment) × time decay coefficient 0.95 (assuming a delay of 1 second) = dynamic priority 0.285;

[0077] For instruction C (speed up): base weight 0.1 × real-time operating condition adjustment coefficient 1.0 (operating condition not triggered adjustment) × time decay coefficient 1.0 = dynamic priority 0.1.

[0078] Based on the calculations of this model, instruction A obtained the highest dynamic priority. This accurately reflects that under the current working conditions of high-pressure crack development and prominent leakage risk, leakage prevention should be the top priority. This calculation result provides an accurate basis for subsequent conflict arbitration and instruction serialization.

[0079] This priority calculation model organically combines the basic weights of static configuration, the real-time adjustment coefficients of real-time sensing operating conditions, and the time decay coefficients that reflect the timeliness of decision-making. This enables the calculation results of dynamic priorities to sensitively and reasonably reflect the true urgency and importance of each optimization objective in the real-time drilling environment, laying the core decision-making foundation for intelligent arbitration under multiple command conflicts.

[0080] Furthermore, conflicting instructions are identified based on a predefined instruction conflict degree matrix, specifically including:

[0081] The rows and columns of the conflict degree matrix represent the adjustment directions of the three parameters: drilling pressure, rotation speed, and displacement, respectively. The matrix element values ​​are the probabilities of the corresponding parameter adjustment combinations leading to the deterioration of the working conditions, obtained based on historical data statistics.

[0082] The conflict degree matrix is ​​a predefined two-dimensional data structure, whose rows and columns represent the adjustment direction of three key drilling parameters: drilling pressure, rotation speed, and displacement, respectively.

[0083] In this matrix, the adjustment direction of each parameter is quantified into three states: +1 represents increasing the parameter, -1 represents decreasing the parameter, and 0 represents keeping the parameter unchanged.

[0084] Each element in the matrix is ​​derived from statistical analysis of historical drilling data, specifically representing the probability of deterioration of downhole conditions when two parameters are adjusted simultaneously in a specific direction.

[0085] For example, when the drilling pressure parameter is adjusted in the +1 direction (i.e., increasing drilling pressure) while the displacement parameter is adjusted in the -1 direction (i.e., decreasing displacement), according to historical data, the probability of this combination leading to a deterioration in working conditions is 0.8. Therefore, the element value at the corresponding position in the command conflict degree matrix is ​​set to 0.8. Similarly, when the rotation speed parameter is adjusted in the +1 direction (i.e., increasing rotation speed) while the displacement parameter is also adjusted in the +1 direction (i.e., increasing displacement), according to historical data, the probability of this combination leading to a deterioration in working conditions is 0.4. Therefore, the element value at the corresponding position in the command conflict degree matrix is ​​set to 0.4.

[0086] This quantitative method based on historical data ensures the objectivity and accuracy of conflict assessment.

[0087] In practical applications, when the system needs to determine whether two instructions conflict with each other, it first parses the parameter adjustment direction information contained in each instruction. For any two instructions, the system extracts their adjustment directions for the three parameters of drilling pressure, rotation speed, and displacement, respectively. Then, it searches for the corresponding conflict degree value in the instruction conflict degree matrix. Specifically, if the first instruction requires increasing drilling pressure (drilling pressure adjustment direction is +1) and decreasing displacement (displacement adjustment direction is -1), and the second instruction requires decreasing drilling pressure (drilling pressure adjustment direction is -1) and increasing displacement (displacement adjustment direction is +1), then the system queries the conflict degree value corresponding to these two sets of parameter adjustment combinations in the instruction conflict degree matrix and takes the maximum value as the overall conflict degree value between the two instructions.

[0088] The conflict threshold is a preset critical value used to determine whether there is a substantial conflict between commands. In a preferred embodiment of the present invention, the conflict threshold is set to 0.7. The value is set based on the statistical analysis results of a large amount of historical drilling data: when the probability of the combination of parameter adjustment directions of two commands leading to the deterioration of working conditions exceeds 70%, executing these two commands at the same time will significantly increase the risk of complex downhole situations.

[0089] In actual operation, the system compares the calculated conflict degree value with the preset conflict threshold. If there is a conflict degree value greater than the preset conflict threshold, the two instructions are determined to be in conflict and arbitration is required. This means that historical experience shows that the simultaneous execution of these two instructions is very likely to lead to adverse consequences. Conversely, if all conflict degree values ​​are less than or equal to the preset conflict threshold, the instructions are determined to be safe to be executed simultaneously without serialization.

[0090] This conflict identification mechanism, based on a predefined command conflict degree matrix, can accurately quantify the degree of conflict between different commands, providing an objective basis for subsequent dynamic priority arbitration. During the drilling process of ultra-deep wells in complex formations, when the geological model, mechanical model, and hydraulic model simultaneously generate multiple optimized commands, this mechanism can identify potential conflict combinations before the commands are executed, thereby avoiding parameter oscillations and downhole risks caused by the simultaneous execution of conflicting commands. By transforming historical experience data into a quantitative conflict degree evaluation standard, this invention achieves accurate identification and effective prevention of multi-source command conflicts, significantly improving the level of automation control and safety of the ultra-deep well drilling process.

[0091] Furthermore, a snapshot of the current drilling parameter status with a unique session identifier is generated. This step is fundamental to ensuring the traceability of system decisions and the recoverability of anomalies. Its execution is handled by the system's status snapshot manager module and is collaboratively completed under the triggering of the dynamic priority arbitrator. This step specifically includes:

[0092] When a command conflict is detected, immediately record the current drilling pressure, rotation speed, displacement, and well depth data;

[0093] When a command conflict is identified, it means that the dynamic priority arbitrator completes the conflict detection based on the predefined command conflict degree matrix and determines that there is a high-risk command conflict with a conflict degree value exceeding the preset conflict threshold. At this time, the dynamic priority arbitrator will immediately send a request to the status snapshot manager to generate a status snapshot. The status snapshot manager responds to this request and obtains and records the instantaneous values ​​of key drilling parameters at the moment the conflict occurs from the system's real-time data buffer or from real-time data events subscribed to through the event bus. These parameters mainly include drilling pressure, rotation speed, displacement, and well depth data. These data together constitute the instantaneous field state of the drilling system before the conflict decision point.

[0094] The recorded data is associated with and stored with a unique session identifier to form a traceable state snapshot;

[0095] The State Snapshot Manager generates a unique session identifier for this snapshot. This identifier is unique within the current system cycle or the current complex conflict resolution session. For example, it can be a string generated based on a combination of timestamp, sequence number, and system instance ID.

[0096] Subsequently, the manager binds the recorded drill pressure, rotation speed, displacement, well depth data, and other optional auxiliary status data (such as system time and risk level) to the unique session identifier and persists them in a structured data format (such as database records or files).

[0097] In this way, a traceable state snapshot is formed. This state snapshot not only contains the state data itself, but also associates it with the conflict arbitration event that triggered the generation of this snapshot, the subsequent execution sequence, and possible rollback operations through a unique session identifier, ensuring the integrity and traceability of the entire decision chain.

[0098] Similarly, taking the drilling encounter with a high-pressure fracture zone as an example, the dynamic priority arbitrator may simultaneously receive a leak prevention command A (reduce discharge rate) and a speed-up command C (increase drilling pressure). After querying the command conflict degree matrix, the conflict degree value is calculated to be 0.85, which is higher than the preset conflict threshold of 0.7. The system identifies the command conflict.

[0099] At this point, the dynamic priority arbitrator triggers the snapshot generation process, and the status snapshot manager is called. It quickly captures the real-time parameters at this moment, such as: drilling pressure = 180KN, rotation speed = 65rpm, displacement = 32L / s, and well depth = 4527.3m.

[0100] At the same time, the system generates a unique session identifier, such as "Arbitration_Session_20231027_142356_001". The state snapshot manager associates the captured parameter data with the identifier "Arbitration_Session_20231027_142356_001" and stores it in the snapshot database. This complete record is a traceable state snapshot.

[0101] Subsequently, whether other instructions are followed after the successful execution of the leak prevention instruction A, or if a rollback is required due to an execution anomaly and a reverse adjustment instruction is generated based on the state snapshot, this unique session identifier can be used to quickly and accurately locate and retrieve the precise state before the conflict occurred, providing a solid foundation for decision-making consistency and system recovery.

[0102] Through the above mechanism, the method of the present invention can automatically save a precise checkpoint each time it faces a critical multi-instruction conflict decision, which greatly enhances the robustness and controllability of the intelligent drilling system in complex and dynamic environments.

[0103] Furthermore, after identifying instruction conflicts and generating a state snapshot, the critical instruction scheduling and serialization step begins. This step sorts the conflicting instructions according to dynamic priorities and generates an ordered execution sequence. This step specifically includes two core operations executed by the system's dynamic priority arbitrator. Its purpose is to transform the concurrent stream of conflicting instructions into a sequentially executable and logically clear instruction sequence. Specifically, this step includes:

[0104] The instruction with the highest dynamic priority is selected as the currently executed instruction.

[0105] After completing the dynamic priority calculation and identifying the conflict, the dynamic priority arbitrator sorts all conflicting instructions according to their dynamic priority scores.

[0106] The instruction with the highest score is identified as the most urgent and priority goal under the current working conditions. The dynamic priority arbitrator selects this instruction as the current execution instruction and prepares to transform it into a control instruction that can be executed by the intelligent driller system. Selecting the current execution instruction is the core output of the arbitration decision. It determines the action that the system should perform immediately after the conflict is resolved. For example, when the risk of leakage is high, the leak prevention instruction of reducing the discharge rate is executed first.

[0107] Secondly, for other conflicting instructions that were not selected for immediate execution, the system performs the following: placing the remaining conflicting instructions into the delay queue according to their dynamic priority.

[0108] The delay queue is a temporary storage data structure used to store in order the instructions identified in this arbitration that conflict with the currently executing instruction and have been deferred. The dynamic priority arbitrator will arrange these remaining conflicting instructions (i.e., all conflicting instructions other than the currently executing instruction) in descending order of their respective dynamic priority scores and place them in the delay queue. The order in which the instructions are arranged in the delay queue determines the order in which they will be considered for execution in the future.

[0109] This mechanism ensures that even if instructions are delayed, their execution order still follows the urgency logic determined by real-time operating conditions, rather than a simple arrival order.

[0110] As mentioned above, the dynamic priority of the leak prevention instruction A is 1.2, the anti-blocking instruction B is 0.285, and the speed-up instruction C is 0.1. Instructions A have high conflicts with both B and C. In this scenario, the dynamic priority arbitrator selects the instruction with the highest dynamic priority as the currently executed instruction according to the above rules, that is, selects instruction A as the currently executed instruction.

[0111] Meanwhile, the remaining conflicting instructions are placed into the delay queue according to their dynamic priority. Since the dynamic priority of instruction B is higher than that of instruction C, the dynamic priority arbitrator places instruction B in the delay queue before instruction C. Thus, an ordered execution sequence is generated: instruction A is executed immediately, and then instructions B and C are considered for execution in order depending on the conditions.

[0112] Furthermore, after generating an ordered execution sequence, the process proceeds to the specific execution and effect verification stage: instructions in the execution sequence are executed sequentially, and post-execution performance indicators are monitored within a preset monitoring period, specifically including:

[0113] First, the intelligent driller system executes the current execution command. The intelligent driller system receives the current execution command issued by the dynamic priority arbitrator and converts it into specific equipment control signals (such as adjusting the pump speed valve position to change the displacement, adjusting the top drive setting to change the speed, etc.), driving the drilling equipment to perform the corresponding parameter adjustments.

[0114] Within the preset monitoring time window, assess whether the key operating condition indicators have achieved the expected goals of the instruction;

[0115] Once the instruction begins execution, the system starts a preset monitoring time window (e.g., 30 seconds). Within this window, the system continuously collects data through sensors and assesses changes in key operating parameters related to the instruction's objective.

[0116] The expected goal is to establish pre-defined quantitative success criteria for different types of instructions, specifically including:

[0117] For the leak prevention target, the leakage rate is reduced to below the third threshold and remains stable. Preferably, the third threshold is set to 5 m³ / h. The system monitors the wellhead leakage rate and evaluates whether it decreases and stabilizes below 5 m³ / h within the monitoring period after the execution of the leak prevention command (such as reducing the discharge rate).

[0118] For the anti-jamming target, the torque fluctuation amplitude is reduced to below the fourth threshold. Preferably, the fourth threshold is set to 30%. The system calculates the torque fluctuation amplitude and evaluates whether it is reduced to below 30% after executing the anti-jamming command (such as increasing the speed).

[0119] For the speed-up target, the mechanical drilling speed increase rate reaches the fifth threshold. Preferably, the fifth threshold is set to 5%. The system calculates the percentage increase in mechanical drilling speed after executing the speed-up command (such as increasing drilling pressure) and evaluates whether it reaches or exceeds 5%.

[0120] If the expected goal is achieved and no new high-risk events are added, the next priority instruction is taken out from the delay queue, and after verifying the compatibility of its parameter adjustment with the current state, it is used as the new current execution instruction.

[0121] When the system confirms that the expected goal of the current execution instruction has been achieved, such as successful leakage prevention, leakage amount stably below 5m³ / h, and no new high-risk events appear within the preset monitoring time window, it is considered that the current sub-stage goal has been safely completed, and the next goal can be attempted.

[0122] At this point, the dynamic priority arbiter will access the delay queue and retrieve the instruction with the next lower priority. However, the system will not directly issue it as the new current execution instruction. Considering the continuity of the drilling status, the system will first verify the compatibility of its parameter adjustment with the current status.

[0123] For example, suppose the instruction just retrieved from the delay queue is a speed-up command (requiring an increase in drilling pressure), and the current state is in low-flow mode because a leak prevention command has just been executed. The dynamic priority arbitrator needs to verify whether it is safe and feasible to increase drilling pressure under the constraints of current pump pressure capacity, formation fracture pressure, etc. If it is compatible, the arbitrator sets the instruction as the new current execution instruction and starts a new round of "execution-monitoring" cycle; if it is incompatible, the dynamic priority arbitrator may return the instruction to the queue and reorder it, or request the drilling platform to generate a new optimized instruction based on the latest state.

[0124] Similarly, continuing the previous high-pressure fracture zone scenario, the system first executed the leak prevention command A, which was selected as the current execution command. In the subsequent 30-second preset monitoring time window, the system evaluated the key operating condition indicator of wellhead leakage and confirmed that it decreased from 10 m³ / h to 3 m³ / h and remained stable, thus achieving the expected goal of reducing the leakage to below the third threshold, and no new alarms were added during the period.

[0125] Once the conditions are met, the dynamic priority arbitrator retrieves the next priority instruction from the delay queue, namely the anti-card instruction B.

[0126] The compatibility of the arbitrator verification command B with the current state (low displacement, reduced bottom hole pressure) was determined by calculation. Under the current parameters, moderately increasing the rotation speed helps to clean the wellbore and will not cause new pressure risks, so it is judged to be compatible.

[0127] Therefore, instruction B is set as the new current execution instruction, handed over to the intelligent driller system for execution, and enters a new round of monitoring and evaluation cycle.

[0128] Furthermore, upon detecting a deterioration in operating conditions and triggering a rollback operation, the following core recovery steps are executed: A progressive reverse adjustment instruction is generated based on the aforementioned state snapshot to perform the rollback, specifically including:

[0129] Get the most recently generated valid state snapshot;

[0130] First, when the abnormal rollback conditions are met, the dynamic priority arbitrator sends a rollback request to the state snapshot manager. Based on the request, the state snapshot manager retrieves and obtains the most relevant and most recently generated valid state snapshot from its stored snapshot records. Typically, this valid state snapshot refers to the state snapshot saved before the instruction that caused the abnormal state was executed. It records the last known and stable operating point state of the system before it enters an unstable or risky state. Obtaining this snapshot is the benchmark for performing an accurate rollback.

[0131] Next, calculate the difference between the current drilling parameter state and the target state recorded in the state snapshot;

[0132] The status snapshot manager or dynamic priority arbitrator compares the current drilling parameter status (such as the current drill pressure, rotation speed, and displacement values) obtained from the sensors in real time with the target status (i.e., the drill pressure, rotation speed, and displacement values ​​recorded in the snapshot) read from the valid status snapshot. Through subtraction, it calculates the difference value of each key parameter. These difference values ​​quantify the degree and direction of the system's current state deviating from the safety benchmark.

[0133] Finally, a parameter recovery command is generated based on the difference value. This command controls the drilling parameters to be adjusted step by step to the target state within a preset safe time window.

[0134] Based on the calculated difference value, the state snapshot manager or dynamic priority arbitrator generates a structured parameter recovery instruction. The core feature of this instruction is that it does not immediately and all at once adjust the parameters to the target state, but rather plans a smooth recovery path.

[0135] For example, if the current drilling pressure is 220 kN and the drilling pressure recorded in the snapshot is 180 kN, then the difference in drilling pressure is +40 kN, and an instruction to reduce drilling pressure is generated. If the current displacement is 28 L / s and the displacement recorded in the snapshot is 32 L / s, then the difference in displacement is -4 L / s, and an instruction to increase displacement is generated.

[0136] To prevent secondary risks caused by sudden parameter changes, the instruction will decompose the total adjustment amount and specify that it will be completed step by step within a preset safe time window (e.g., 20 seconds). For example, for a 40 kN reduction in drilling pressure differential, the instruction may be designed to reduce the pressure by 10 kN in four steps within 20 seconds, with each step reducing the pressure by 10 kN every 5 seconds.

[0137] The endpoint of the step-by-step adjustment, that is, the final state to be restored, is the target state recorded in the effective state snapshot.

[0138] Furthermore, after performing the rollback operation based on the state snapshot, a key adaptive learning step is included: readjusting the priority strategy for subsequent instructions, specifically including:

[0139] If the same type of instruction triggers a rollback operation a set number of times within a preset time period, the basic weight of that type of instruction will be automatically reduced.

[0140] This rule is triggered based on monitoring the failure mode of instruction execution. The same type of instruction refers to instructions with the same target type, such as those that all belong to the speed-up or leak prevention type. The preset time is a configurable time window used to define a continuous time range, and the set number is a configurable threshold.

[0141] When the system detects that within the preset time period, the number of times the same type of instruction triggers a rollback operation due to the deterioration of the working conditions after execution reaches or exceeds the preset number, it is determined that the instruction of this type has repeatedly failed under the current working conditions and needs to be intervened in its decision priority.

[0142] When the above conditions are met, the dynamic priority arbitrator will perform weight adjustment, that is, automatically reduce the base weight of this type of instruction. Automatic reduction means that it is executed autonomously by the system algorithm without manual intervention. The reduction magnitude and duration can be preset. For example, the base weight of this type of instruction can be temporarily reduced by 50%, and the reduced state can be maintained for a fixed duration, such as 10 minutes.

[0143] This adjustment directly affects the calculation of subsequent dynamic priorities, causing the importance of such instructions to be temporarily reduced in subsequent arbitration decisions. This guides the system to focus more on other more likely successful objectives for a period of time, avoiding repeated operation in the same error mode.

[0144] For example, when drilling through a formation with well-developed high-pressure fractures, assume that the initial basic weight of the system is set to prevent leakage by 0.6 and increase speed by 0.3;

[0145] The system executes an acceleration command, but detects a sharp increase in leakage within the preset monitoring period, triggering a rollback.

[0146] Within the next preset time, such as 1 minute, another acceleration command generated based on the new data is executed again, but this also causes the leakage to worsen and triggers a second rollback.

[0147] At this point, the system determines that the acceleration command has triggered the rollback operation twice within the preset time.

[0148] Therefore, the dynamic priority arbiter automatically reduces the base weight of acceleration instructions, for example, from 0.3 to 0.15, for 10 minutes;

[0149] In the next 10 minutes, when new speed-up commands and leakage prevention commands are executed concurrently, their calculated dynamic priority will decrease significantly because their base weight has been reduced. The system will be more inclined to execute leakage prevention commands first, thereby adaptively adapting to the current high-risk formation and avoiding blindly pursuing drilling speed and repeatedly causing well leakage.

[0150] Furthermore, the ultra-deep well drilling method also includes an auxiliary data maintenance and update step to ensure the accuracy and timeliness of decision-making. This step specifically includes:

[0151] Periodically check the timeliness of the operating condition data on which each instruction in the delay queue is based;

[0152] This operation is performed by a dynamic priority arbitrator. Periodic checks refer to triggering a check process at preset time intervals or before each attempt to retrieve an instruction from the delay queue for execution. The goal of the check is to evaluate whether the sensor data on which each instruction stored in the delay queue awaiting execution was generated is still valid. The timeliness of the data is determined by comparing the timestamp of the data recorded in the instruction event with the current system time. Typically, a timeliness threshold is set (for example, if the data generation time exceeds the current time by 5 seconds, it is considered outdated).

[0153] Secondly, regarding the inspection results, if the operating condition data on which a certain instruction is based is outdated, the instruction will be removed from the queue and an update instruction will be requested.

[0154] When an inspection finds that the timestamp of the working condition data associated with a certain instruction in the delay queue exceeds the preset timeliness threshold, it is determined that the timeliness of the data has been lost.

[0155] The dynamic priority arbitrator will immediately remove the instruction from the delay queue to ensure that subsequent execution decisions are not based on outdated information that may no longer reflect the true downhole conditions, thereby avoiding control risks caused by data lag.

[0156] While removing the instruction from the queue, the system sends a data update request or event to the multi-source decision model that generated the instruction via the event bus, notifying it that the instruction has been discarded due to outdated data, and prompting it to recalculate and evaluate based on the latest real-time sensor data to generate a new optimized instruction that reflects the current operating conditions. This mechanism ensures that even if the instruction is delayed, its decision basis can remain synchronized with the real-time operating conditions, maintaining the accuracy and agility of the overall system response.

[0157] Furthermore, the multi-source decision model is the core of the intelligent drilling system's decision-making process. It is not a single model, but a collection of collaborative models. Specifically, the multi-source decision model includes a geological model, a mechanical model, and a hydraulic model.

[0158] The geological model is mainly based on geological engineering data such as logging while drilling and well logging. Its optimization goal focuses on speed improvement, that is, by analyzing formation lithology, drillability, etc., it generates parameter optimization instructions aimed at improving mechanical drilling speed, such as suggesting to increase drilling pressure.

[0159] The mechanical model is mainly based on data such as wellbore mechanics and formation pressure monitoring. Its optimization objectives focus on leakage prevention and wellbore stability. That is, by calculating the bottom hole pressure window, formation stress, etc., it generates parameter optimization instructions aimed at preventing well leakage and collapse, such as suggesting to reduce the discharge rate to control the equivalent circulation density.

[0160] The hydraulic model is mainly based on data such as annular hydraulic parameters, drilling fluid properties, and cuttings transport. Its optimization objectives focus on preventing stuck pipe and keeping the wellbore clean. That is, by evaluating cuttings carrying efficiency, friction torque, etc., it generates parameter optimization instructions aimed at preventing stuck pipe accidents caused by filter cake adhesion to the wellbore when the drill string is stationary and ensuring wellbore cleanliness, such as suggesting to increase rotation speed and displacement.

[0161] The key point is that the geological model, mechanical model and hydraulic model generate optimization instructions independently based on different data sources. This means that the three models are parallel and independent computing processes when running. They subscribe to and process data streams in their respective professional fields, and independently and asynchronously calculate what they consider to be the optimal drilling parameter adjustment suggestions based on built-in algorithms and rules.

[0162] However, it is precisely in complex strata that when different models make judgments based on local targets and data almost simultaneously, the instruction space conflict problem mentioned above will occur. In order to solve the problem of coupling between models and direct conflict of instructions, and to achieve efficient and decoupled communication, the instructions generated by the geological model, mechanical model and hydraulic model are asynchronously published through the event bus.

[0163] As the central nervous system-like communication infrastructure, the event bus allows decision-making models to no longer send instructions directly to the execution system or to each other. Instead, they encapsulate the generated instructions into structured instruction events in a unified format and then publish the events to the event bus. The event bus is responsible for the storage, routing, and distribution of events.

[0164] Asynchronous means that the model's decision-making process is non-blocking. After generating an instruction event and publishing it to the bus, the model can immediately continue its next round of data processing and decision calculation without waiting for the instruction to be received, processed, or executed. This ensures the independence and efficiency of each model's own calculation cycle and avoids computational delays or blocking caused by waiting for instruction feedback.

[0165] For example, when drilling encounters a complex stratigraphic intersection zone where both easily lost fractures and easily narrowed gypsum-salt rocks coexist, the three models operate independently based on their respective real-time data:

[0166] The geological model detected a decrease in drilling speed and determined that the formation was hardening based on its data source. Therefore, it generated a speed-up command event (target: speed up, action: increase drilling pressure) and published it asynchronously through the event bus.

[0167] The mechanical model detected a shrinking pressure window margin and, based on its data source, determined that the risk of leakage was increasing. It then generated a leak prevention command event (target: leak prevention, action: reduce displacement) and published it asynchronously via the event bus almost simultaneously.

[0168] The hydraulic model analyzes torque and cuttings concentration data to determine that the wellbore cleanliness has deteriorated and there is a risk of stuck drill bit. Therefore, an anti-sticking command event is generated (target: anti-sticking, action: increase rotation speed), which is also published asynchronously through the event bus.

[0169] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for drilling ultra-deep wells based on a spiral toothed PDC drill bit, characterized in that, Includes the following steps: Receive parameter optimization instructions generated concurrently by a multi-source decision model for the goals of speed improvement, leakage prevention, and jamming prevention; Based on real-time downhole operating data, a priority calculation model including basic weight, real-time operating condition adjustment coefficient and time decay coefficient is adopted to calculate dynamic priority for multiple concurrently generated instructions. At the same time, conflicting instructions are identified based on a predefined instruction conflict degree matrix. When a conflict is identified, a snapshot of the current drilling parameter status with a unique session identifier is generated, and the conflicting instructions are sorted according to dynamic priority and an ordered execution sequence is generated. The instructions in the execution sequence are executed sequentially, and the working condition indicators after execution are monitored within a preset monitoring period; If the detected operating condition indicators deteriorate beyond the safety threshold, a reverse adjustment instruction is generated based on the state snapshot to roll back the operation and the priority strategy of subsequent instructions is readjusted.

2. The ultra-deep well drilling method based on a spiral toothed PDC drill bit according to claim 1, characterized in that, A priority calculation model is adopted, which includes basic weights, real-time operating condition adjustment coefficients, and time decay coefficients. Specifically, it includes: Basic weights are assigned to the three types of instructions: speed-up, leak prevention, and jam prevention. Based on the pressure window margin and the rate of change of mechanical drilling speed in the real-time operating data, the real-time operating condition adjustment coefficient is dynamically calculated. When the pressure margin is lower than the first threshold, the real-time operating condition adjustment coefficient of the leak prevention command is increased to the first coefficient; when the mechanical drilling speed decrease rate is higher than the second threshold, the real-time operating condition adjustment coefficient of the speed-up command is increased to the second coefficient. The time decay coefficient is calculated based on the generation timestamp of each instruction, and the delay time is positively correlated with the degree of decay. The dynamic priority of the instruction is obtained by multiplying the basic weight, the real-time operating condition adjustment coefficient, and the time decay coefficient.

3. The ultra-deep well drilling method based on a spiral toothed PDC drill bit according to claim 1, characterized in that, Conflicting instructions are identified based on a predefined instruction conflict degree matrix, specifically including: The rows and columns of the conflict degree matrix represent the adjustment directions of the three parameters: drilling pressure, rotation speed, and displacement, respectively. The matrix element values ​​are the probabilities of the corresponding parameter adjustment combinations leading to the deterioration of the working conditions, obtained based on historical data statistics. Calculate the conflict degree value corresponding to the conflict degree matrix for the direction of parameter adjustment between instructions; If a conflict value exceeds the preset conflict threshold, the instructions are determined to be conflicting.

4. The ultra-deep well drilling method based on a spiral toothed PDC drill bit according to claim 3, characterized in that, Generate a snapshot of the current drilling parameter status with a unique session identifier, specifically including: When a command conflict is detected, immediately record the current drilling pressure, rotation speed, displacement, and well depth data; The recorded data is associated with a unique session identifier and stored to form a traceable state snapshot.

5. The ultra-deep well drilling method based on a spiral toothed PDC drill bit according to claim 1, characterized in that, Conflicting instructions are sorted according to dynamic priority and an ordered execution sequence is generated, specifically including: The instruction with the highest dynamic priority is selected as the currently executed instruction. The remaining conflicting instructions are placed into the delay queue in order of their dynamic priority.

6. The ultra-deep well drilling method based on a spiral toothed PDC drill bit according to claim 5, characterized in that, The instructions in the execution sequence are executed sequentially, and the post-execution operating conditions are monitored within a preset monitoring period, specifically including: Execute the currently executing instruction; Within the preset monitoring time window, assess whether the key operating condition indicators have achieved the expected goals of the instruction; For the leak prevention target, the leakage amount is reduced to below the third threshold and remains stable; for the jam prevention target, the torque fluctuation range is reduced to below the fourth threshold; and for the speed increase target, the mechanical drilling speed increase rate reaches the fifth threshold. If the expected goal is achieved and no new high-risk events are added, the instruction with the next priority is taken from the delay queue, and after verifying the compatibility of its parameter adjustment with the current state, it is used as the new current execution instruction.

7. The ultra-deep well drilling method based on a spiral toothed PDC drill bit according to claim 1, characterized in that, Based on the aforementioned state snapshot, a reverse adjustment instruction is generated for rollback, specifically including: Get the most recently generated valid state snapshot; Calculate the difference between the current drilling parameter status and the target status recorded in the status snapshot; A parameter recovery command is generated based on the difference value. This command controls the drilling parameters to be adjusted step by step to the target state within a preset safe time window.

8. The ultra-deep well drilling method based on a spiral toothed PDC drill bit according to claim 1, characterized in that, The priority strategy for subsequent instructions is readjusted, specifically including: If the same type of instruction triggers a rollback operation a set number of times within a preset time period, the basic weight of that type of instruction will be automatically reduced.

9. A method for drilling ultra-deep wells based on a spiral toothed PDC drill bit according to claim 6, characterized in that, The ultra-deep well drilling method also includes: Periodically check the timeliness of the operating condition data on which each instruction in the delay queue is based; If the operating condition data on which an instruction is based is outdated, the instruction is removed from the queue and an update request is sent.

10. The ultra-deep well drilling method based on a spiral toothed PDC drill bit according to claim 1, characterized in that, The multi-source decision model includes a geological model, a mechanical model, and a hydraulic model, each of which independently generates optimization instructions based on different data sources. The instructions generated by the geological model, mechanical model, and hydraulic model are asynchronously published via the event bus.

Images