A machine learning-based well bottom drilling parameter monitoring and optimization method

Abstract

The application provides a machine learning based well bottom drilling parameter monitoring and optimization method, and belongs to the technical field of oil geological exploration and drilling.The application accurately fits the well bottom bit pressure and bit torque through a soft rod model of drilling engineering, then judges whether the well bottom mechanical specific energy appears a real abnormal situation through a machine learning algorithm, and accurately identifies and processes the downhole abnormal state according to the engineering business logic.The application comprehensively evaluates whether the abnormal energy dissipation appears in the real-time drilling process through kmeans, LDA and SOS multiple judgment algorithms, and uses the MSE decision tree process to optimize the drilling parameters, so that the abnormal situation judgment success rate is improved to the reliable degree in actual engineering.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of oil geological exploration and drilling, and particularly relates to a well bottom drilling parameter monitoring and optimization method based on machine learning. BACKGROUND

[0002] In the drilling process, due to factors such as formation changes, different drilling tool combinations, different drill bit types, and different drilling parameters, the drilling mechanical drilling speed varies greatly, and problems such as excessive downhole vibration, accelerated drill bit wear, and excessive vibration of the drilling tool combination often occur, which leads to early tripping, low tripping footage, and low mechanical drilling speed.

[0003] The prior art uses the mechanical specific energy MSE parameter calculated based on ground comprehensive logging parameters as an index of drilling energy dissipation. In Chinese patent application document CN109145322A, an algorithm for calculating the bottom hole mechanical specific energy MSE is proposed, which aims to use the bottom hole mechanical specific energy MSE as a judgment index to analyze and judge, and optimize the drilling parameters. DH Since the parameters obtained during drilling are collected by ground sensors deployed by comprehensive logging services, problems such as abnormal sensor collection and data interruption often occur.

[0004] If the abnormal threshold method is used to judge the abnormality of MSEDH, false positives and false negatives often occur. In recent years, machine learning algorithms have been widely used in big data processing, especially in the abnormal monitoring of time series data.

[0005] The prior art at least has the following disadvantages:

[0006] 1. The abnormal threshold method is used to judge the abnormality of MSEDH, which often leads to false positives and false negatives;

[0007] 2. The drill bit pressure and drill bit torque obtained by conventional ground data approximation fitting are inaccurate;

[0008] 3. There is no method to correct the drill bit pressure and drill bit torque by soft rod model back calculation, which cannot effectively calculate the bottom hole mechanical specific energy, thus leading to the inability to judge the instability of the drill bit. SUMMARY

[0009] To address the problems existing in current technologies, this invention provides a method for monitoring and optimizing downhole drilling parameters based on machine learning. It adaptively uses machine learning algorithms to determine whether there are genuine anomalies in the downhole mechanical energy specific energy, and then combines this with engineering business logic to accurately identify and handle downhole anomalies. This invention comprehensively evaluates whether abnormal energy dissipation occurs during real-time drilling using multiple judgment algorithms, including k-means, LDA, and SOS, and fully utilizes the MSE decision tree process for drilling parameter optimization, thereby improving the success rate of anomaly detection to a level reliable for actual engineering projects. This invention can accurately calculate the drill bit pressure and drill bit torque at the bottom of the well, thus more accurately using machine learning algorithms to judge the actual downhole conditions.

[0010] This invention provides a machine learning-based method for monitoring bottom hole drilling parameters, comprising the following steps:

[0011] For each set of newly received integrated logging data, the current drill bit pressure, drill bit torque, drill bit speed, and mechanical drilling speed are calculated using a drilling rig status judgment algorithm. The integrated logging data includes suspended weight, mechanical speed, inlet flow rate, mud density, and standpipe pressure.

[0012] Based on the current drill bit pressure, drill bit torque, drill bit speed, and mechanical drilling speed, obtain the bottom hole mechanical energy (MSE) at the current moment. DH ;

[0013] The currently obtained MSE DH The values ​​are stored in the analysis data pool, and machine learning algorithms are used to determine the MSE in the current analysis data pool. DH Has the ratio of the triaxial compressive strength of the formation rocks deviated from the original average?

[0014] If you determine the current MSE DH If a deviation occurs, an engineering analysis and judgment will be performed. Based on the engineering judgment parameters, a machine learning algorithm will be called to traverse each engineering judgment parameter to determine whether any abnormality has occurred in each engineering judgment parameter and whether corresponding risks may have occurred.

[0015] Preferably, the drill bit pressure and drill bit torque are obtained by the following method:

[0016] Drill bit pressure = suspended weight during drilling - suspended weight during stable idling circulation after lifting off the bottom of the well;

[0017] Drill bit torque = Torque during drilling - Torque during stable idle circulation after lifting off the bottom of the well.

[0018] Preferably, the drill bit pressure and drill bit torque are obtained by the following method:

[0019] The soft rod model is used for iterative back calculation until the surface drilling pressure and surface torque calculated by the soft rod model are consistent with the measured logging data. Then, the drill bit drilling pressure and drill bit torque corrected by the soft rod model are obtained.

[0020] Preferably, the drill bit pressure and drill bit torque are obtained by iterative back calculation using a soft rod model to correct for the soft rod model. Specifically, this includes the following steps:

[0021] Assume the initial estimated value of drill bit pressure and drill bit torque is 0;

[0022] Repeat the following process until the calculated ground drilling pressure and ground torque converge with the measured values. At this point, the corresponding drill bit drilling pressure and drill bit torque are accurate values:

[0023] The ground drilling pressure and ground torque were calculated by reverse engineering based on the soft rod model.

[0024] The difference between the actual measured ground drilling pressure and ground torque is obtained and then added to the drill bit drilling pressure and drill bit torque.

[0025] Preferably, the drill bit rotation speed is obtained by the following method:

[0026] When there is no screw drill bit, the drill bit speed equals the ground speed;

[0027] When a screw drill bit is used, the drill bit speed = ground speed + screw speed.

[0028] Preferably, the bottom hole mechanical energy (MSE) is obtained based on the current drill bit pressure, drill bit torque, drill bit rotation speed, and mechanical drilling speed. DH .

[0029] Preferably, the engineering analysis and judgment adopts the bottom-hole mechanical energy ratio decision tree method obtained by calculation.

[0030] Preferably, the machine learning algorithm includes the following algorithms: kmeans algorithm, LDA algorithm, and SOS algorithm. The kmeans algorithm is used for preliminary determination of MSE. DH Does a multi-state distribution exist? Then, use the LDA or SOS algorithm to determine the characteristics of abnormal trends.

[0031] Preferably, the engineering judgment parameters include the drill bit mud bag risk index, the stick-slip index, and the severity of eddy currents.

[0032] Preferably, the drill bit mud bag risk index corresponds to the drill bit mud bag risk, the stick-slip index corresponds to the stick-slip risk, and the vortex severity corresponds to the vortex risk.

[0033] This invention provides a method for optimizing bottom hole drilling parameters. Using the aforementioned machine learning-based bottom hole drilling parameter monitoring method, monitoring results are obtained.

[0034] If the monitoring results show an anomaly, an alarm will be triggered indicating that there may be a risk downhole corresponding to the abnormal engineering judgment parameters;

[0035] Based on the potential risks, a final bottom hole drilling parameter optimization strategy is proposed.

[0036] Preferably, the step of providing the final bottom hole drilling parameter optimization strategy based on possible risks includes:

[0037] To mitigate the risk of drill bit mud buildup, increase displacement and reduce drilling pressure and rotational speed.

[0038] To mitigate stick-slip risk, reduce drilling pressure and increase drilling speed;

[0039] To mitigate the risk of vortexing, lift the drill bit off the bottom of the well. Once the severity of vortexing reaches the preset normal value, lower the drill bit back to the bottom of the well, reduce the rotation speed, and increase the drilling pressure.

[0040] Preferably, the step of providing the final bottom hole drilling parameter optimization strategy based on possible risks includes:

[0041] To address the risk of drill bit mud buildup, increase displacement by 10%, reduce drilling pressure by 10%, and reduce rotational speed by 10%.

[0042] To mitigate the risk of stick-slip, reduce drilling pressure by 10% and increase rotational speed by 10%.

[0043] To mitigate the risk of vortexing, lift the drill bit off the bottom of the well. Once the severity of vortexing reaches the preset normal value, lower the drill bit back to the bottom of the well, reduce the rotation speed by 10%, and increase the drilling pressure by 10%.

[0044] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0045] (1) The present invention uses the drill bit pressure and drill bit torque after correction by the soft rod model to calculate the bottom hole mechanical energy (MSE). DH This allows us to determine whether there are any abnormal parameters.

[0046] (2) Based on the drill bit pressure and drill bit torque after correction using a soft rod model, this invention calculates the bottom hole mechanical energy (MSE). DH The ratio of bottom-hole mechanical specific energy to triaxial compressive strength was calculated, and a combination of multiple machine learning algorithms, including the k-means algorithm, LDA algorithm, and SOS algorithm, was used to preliminarily determine the MSE. DH Whether there is a multi-state distribution, and then judge the abnormal trend characteristics based on LDA or SOS algorithm, and judge whether the engineering judgment parameters are abnormal, so as to make the judgment more accurate;

[0047] (2) When the engineering judgment parameters are abnormal, the present invention adopts the bottom hole mechanical energy MSE decision tree method. Using the bottom hole mechanical energy ratio MSE calculated based on the soft rod model, different optimization suggestions for adjusting parameters such as displacement, drilling pressure and rotation speed are given for different risks, so that the engineering judgment parameters can be restored to normal as soon as possible.

[0048] (3) This invention proposes to make risk prediction based on engineering judgment parameters such as drill bit mud bag risk index, stick-slip index and eddy severity, which can comprehensively avoid the occurrence of various risks. Attached Figure Description

[0049] Figure 1 This is a flowchart of a machine learning-based bottom hole drilling parameter monitoring method according to an embodiment of the present invention;

[0050] Figure 2 This is a flowchart of a bottom hole drilling parameter optimization method according to an embodiment of the present invention. Detailed Implementation

[0051] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0052] This invention provides a machine learning-based method for monitoring bottom hole drilling parameters, comprising the following steps:

[0053] For each set of newly received integrated logging data, the current drill bit pressure, drill bit torque, drill bit speed, and mechanical drilling speed are calculated using a drilling rig status judgment algorithm. The integrated logging data includes suspended weight, mechanical speed, inlet flow rate, mud density, and standpipe pressure.

[0054] Based on the current drill bit pressure, drill bit torque, drill bit speed, and mechanical drilling speed, obtain the bottom hole mechanical energy (MSE) at the current moment. DH ;

[0055] The currently obtained MSE DH The values ​​are stored in the analysis data pool, and machine learning algorithms are used to determine the MSE in the current analysis data pool. DH Has the ratio of the formation rock's triaxial compressive strength to the formation rock deviated from the original average? According to drilling vibration theory, if this ratio shifts, it indicates an energy dissipation event has occurred downhole, which is considered an abnormal operating condition. The machine learning algorithm used here is a modified k-means unsupervised learning algorithm, which does not require a learning process based on historical data.

[0056] If you determine the current MSE DH If a deviation occurs, an engineering analysis and judgment will be performed. Based on the engineering judgment parameters, a machine learning algorithm will be called to traverse each engineering judgment parameter to determine whether any abnormality has occurred in each engineering judgment parameter and whether corresponding risks may have occurred.

[0057] According to a specific embodiment of the present invention, the drill bit pressure and drill bit torque are obtained by the following method:

[0058] Drill bit pressure = suspended weight during drilling - suspended weight during stable idling circulation after lifting off the bottom of the well;

[0059] Drill bit torque = Torque during drilling - Torque during stable idle circulation after lifting off the bottom of the well.

[0060] According to a specific embodiment of the present invention, the drill bit pressure and drill bit torque are obtained by the following method:

[0061] The soft rod model is used for iterative back calculation until the surface drilling pressure and surface torque calculated by the soft rod model are consistent with the measured logging data. Then, the drill bit drilling pressure and drill bit torque corrected by the soft rod model are obtained.

[0062] The drill bit pressure and torque values ​​do not rely on field-transmitted data. Instead, they are determined by a drilling rig status algorithm based on the suspended weight and idle torque during stable idling after lifting off the bottom of the well. These values ​​are used as the corresponding zero-values ​​for the calibrated drill bit pressure and torque. A drilling engineering soft rod model is used to correct the drill string friction and torque model, thereby accurately calculating the drill bit pressure and torque. Since the drill bit pressure and torque are obtained using the above method, without using the drill bit pressure and torque values ​​from the integrated logging data, the calculated MSE is... DH It is more accurate than conventional calculation methods.

[0063] According to a specific embodiment of the present invention, the drill bit pressure and drill bit torque after correction by the soft rod model are obtained through iterative back calculation. The specific steps include the following:

[0064] Assume the initial estimated value of drill bit pressure and drill bit torque is 0;

[0065] Repeat the following process until the calculated ground drilling pressure and ground torque converge with the measured values. At this point, the corresponding drill bit drilling pressure and drill bit torque are accurate values:

[0066] The ground drilling pressure and ground torque were calculated by reverse engineering based on the soft rod model.

[0067] The difference between the actual measured ground drilling pressure and ground torque is obtained and then added to the drill bit drilling pressure and drill bit torque.

[0068] According to a specific embodiment of the present invention, the drill bit rotation speed is obtained by the following method:

[0069] When there is no screw drill bit, the drill bit speed equals the ground speed;

[0070] When a screw drill bit is used, the drill bit speed = ground speed + screw speed.

[0071] According to a specific embodiment of the present invention, the bottom hole mechanical energy (MSE) at the current moment is obtained based on the current drill bit pressure, drill bit torque, drill bit rotation speed, and mechanical drilling speed, using the standard mechanical energy ratio formula. DH .

[0072] According to a specific embodiment of the present invention, the engineering analysis and judgment adopts the bottom hole mechanical energy ratio decision tree method obtained by calculation.

[0073] According to a specific embodiment of the present invention, the machine learning algorithm includes the following algorithms: kmeans algorithm, LDA algorithm, and SOS algorithm. The kmeans algorithm is used for preliminary determination of MSE. DH Does a multi-state distribution exist? Then, use the LDA or SOS algorithm to determine the characteristics of abnormal trends.

[0074] According to a specific embodiment of the present invention, the engineering judgment parameters include the drill bit mud bag risk index, the stick-slip index, and the eddy severity.

[0075] According to a specific embodiment of the present invention, the drill bit mud bag risk index corresponds to the drill bit mud bag risk, the stick-slip index corresponds to the stick-slip risk, and the eddy severity corresponds to the eddy risk.

[0076] This invention provides a method for optimizing bottom hole drilling parameters. Using the aforementioned machine learning-based bottom hole drilling parameter monitoring method, monitoring results are obtained.

[0077] If the monitoring results show an anomaly, an alarm will be triggered indicating that there may be a risk downhole corresponding to the abnormal engineering judgment parameters;

[0078] Based on the potential risks, a final bottom hole drilling parameter optimization strategy is proposed.

[0079] According to a specific embodiment of the present invention, the step of providing a final bottom hole drilling parameter optimization strategy based on possible risks includes:

[0080] To mitigate the risk of drill bit mud buildup, increase displacement and reduce drilling pressure and rotational speed.

[0081] To mitigate stick-slip risk, reduce drilling pressure and increase drilling speed;

[0082] To mitigate the risk of vortexing, lift the drill bit off the bottom of the well. Once the severity of vortexing reaches the preset normal value, lower the drill bit back to the bottom of the well, reduce the rotation speed, and increase the drilling pressure.

[0083] According to a specific embodiment of the present invention, the step of providing a final bottom hole drilling parameter optimization strategy based on possible risks includes:

[0084] To address the risk of drill bit mud buildup, increase displacement by 10%, reduce drilling pressure by 10%, and reduce rotational speed by 10%.

[0085] To mitigate the risk of stick-slip, reduce drilling pressure by 10% and increase rotational speed by 10%.

[0086] To mitigate the risk of vortexing, lift the drill bit off the bottom of the well. Once the severity of vortexing reaches the preset normal value, lower the drill bit back to the bottom of the well, reduce the rotation speed by 10%, and increase the drilling pressure by 10%.

[0087] Example 1

[0088] According to a specific embodiment of the present invention, and in conjunction with the accompanying drawings, the machine learning-based well bottom drilling parameter monitoring method of the present invention will be described in detail.

[0089] This invention provides a machine learning-based method for monitoring bottom hole drilling parameters, comprising the following steps:

[0090] For each set of newly received integrated logging data, the current drill bit pressure, drill bit torque, drill bit speed, and mechanical drilling speed are calculated using a drilling rig status judgment algorithm. The integrated logging data includes suspended weight, mechanical speed, inlet flow rate, mud density, and standpipe pressure.

[0091] Based on the current drill bit pressure, drill bit torque, drill bit speed, and mechanical drilling speed, obtain the bottom hole mechanical energy (MSE) at the current moment. DH ;

[0092] The currently obtained MSE DH The values ​​are stored in the analysis data pool, and machine learning algorithms are used to determine the MSE in the current analysis data pool. DH Has the ratio of the formation rock's triaxial compressive strength to the formation rock deviated from the original average? According to drilling vibration theory, if this ratio shifts, it indicates an energy dissipation event has occurred downhole, which is considered an abnormal operating condition. The machine learning algorithm used here is a modified k-means unsupervised learning algorithm, which does not require a learning process based on historical data.

[0093] If you determine the current MSE DH If a deviation occurs, an engineering analysis and judgment will be performed. Based on the engineering judgment parameters, a machine learning algorithm will be called to traverse each engineering judgment parameter to determine whether any abnormality has occurred in each engineering judgment parameter and whether corresponding risks may have occurred.

[0094] Example 2

[0095] According to a specific embodiment of the present invention, and in conjunction with the accompanying drawings, the machine learning-based well bottom drilling parameter monitoring method of the present invention will be described in detail.

[0096] This invention provides a machine learning-based method for monitoring bottom hole drilling parameters, comprising the following steps:

[0097] For each set of newly received integrated logging data, the current drill bit pressure, drill bit torque, drill bit speed, and mechanical drilling speed are calculated using a drilling rig status judgment algorithm. The integrated logging data includes suspended weight, mechanical speed, inlet flow rate, mud density, and standpipe pressure.

[0098] Drill bit pressure and drill bit torque are obtained using the following methods:

[0099] Drill bit pressure = suspended weight during drilling - suspended weight during stable idling circulation after lifting off the bottom of the well;

[0100] Drill bit torque = Torque during drilling - Torque during stable idle circulation after lifting off the bottom of the well;

[0101] or,

[0102] Drill bit pressure and drill bit torque are obtained using the following methods:

[0103] The process involves iterative back-calculation using a soft rod model until the surface drilling pressure and surface torque calculated by the soft rod model are consistent with the measured logging data. This yields the corrected drill bit drilling pressure and drill bit torque based on the soft rod model. The specific steps include the following:

[0104] Assume the initial estimated value of drill bit pressure and drill bit torque is 0;

[0105] Repeat the following process until the calculated ground drilling pressure and ground torque converge with the measured values. At this point, the corresponding drill bit drilling pressure and drill bit torque are accurate values:

[0106] The ground drilling pressure and ground torque were calculated by reverse engineering based on the soft rod model.

[0107] The difference between the actual measured ground drilling pressure and ground torque is obtained and then added to the drill bit drilling pressure and drill bit torque.

[0108] The drill bit rotation speed is obtained using the following method:

[0109] When there is no screw drill bit, the drill bit speed equals the ground speed;

[0110] When a screw drill bit is used, the drill bit speed = ground speed + screw speed.

[0111] Based on the current drill bit pressure, drill bit torque, drill bit speed, and mechanical drilling speed, the bottom hole mechanical energy (MSE) at the current moment is obtained using the standard mechanical energy ratio formula. DH .

[0112] The currently obtained MSE DHThe values ​​are stored in the analysis data pool, and machine learning algorithms are used to determine the MSE in the current analysis data pool. DH Has the ratio of the triaxial compressive strength of the formation rocks deviated from the original average?

[0113] If you determine the current MSE DH If a deviation occurs, an engineering analysis and judgment are performed. Combining the engineering judgment parameters, a machine learning algorithm is invoked to traverse each engineering judgment parameter, determining whether any parameter is abnormal and potentially carries corresponding risks. This engineering analysis and judgment employs a calculated bottom-hole mechanical energy ratio decision tree method. The machine learning algorithm includes the following algorithms: kmeans algorithm, LDA algorithm, and SOS algorithm. The kmeans algorithm is used for preliminary judgment of MSE (Mechanical Energy Sequence). DH Does a multi-state distribution exist? Then, use the LDA or SOS algorithm to determine the characteristics of abnormal trends.

[0114] The engineering assessment parameters include the drill bit mud bag risk index, the stick-slip index, and the eddy severity; the drill bit mud bag risk index corresponds to the drill bit mud bag risk, the stick-slip index corresponds to the stick-slip risk, and the eddy severity corresponds to the eddy risk.

[0115] Example 3

[0116] According to a specific embodiment of the present invention, the bottom hole drilling parameter optimization method of the present invention will be described in detail with reference to the accompanying drawings.

[0117] This invention provides a method for optimizing bottom hole drilling parameters, using any of the aforementioned machine learning-based bottom hole drilling parameter monitoring methods to obtain monitoring results.

[0118] If the monitoring results show an anomaly, an alarm will be triggered indicating that there may be a risk downhole corresponding to the abnormal engineering judgment parameters;

[0119] Based on the potential risks, a final bottom hole drilling parameter optimization strategy is proposed.

[0120] The final bottom hole drilling parameter optimization strategy based on potential risks includes:

[0121] To address the risk of drill bit mud buildup, increase displacement by 10%, reduce drilling pressure by 10%, and reduce rotational speed by 10%.

[0122] To mitigate the risk of stick-slip, reduce drilling pressure by 10% and increase rotational speed by 10%.

[0123] To mitigate the risk of vortexing, lift the drill bit off the bottom of the well. Once the severity of vortexing reaches the preset normal value, lower the drill bit back to the bottom of the well, reduce the rotation speed by 10%, and increase the drilling pressure by 10%.

[0124] Through calculation and comparison in this embodiment, traditional methods have a false alarm or missed alarm rate of over 40%, while the false alarm or missed alarm rate obtained according to the present invention can be reduced to less than 5%. Compared with the traditional method of calculating mechanical specific energy using ground parameters and determining abnormal mechanical specific energy conditions through threshold values, the present invention significantly improves the accuracy and reliability of alarms.

[0125] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention are included within the scope of protection of the present invention.

Claims

1. A method for monitoring bottom hole drilling parameters based on machine learning, characterized in that, Includes the following steps: For each set of newly received integrated logging data, the current drill bit pressure and drill bit torque are calculated using a drilling rig status algorithm. The integrated logging data includes suspended weight, machine speed, inlet flow rate, mud density, and standpipe pressure. Based on the current drill bit pressure, drill bit torque, drill bit speed, and mechanical drilling speed, obtain the bottom hole mechanical energy (MSE) at the current moment. DH ; The currently obtained MSE DH The values ​​are stored in the analysis data pool, and machine learning algorithms are used to determine the MSE in the current analysis data pool. DH Has the ratio of the triaxial compressive strength of the formation rocks deviated from the original average? If you determine the current MSE DH If a deviation occurs, an engineering analysis and judgment are performed. Combining the engineering judgment parameters, the calculated bottom hole mechanical energy ratio decision tree method is used, and a machine learning algorithm is called to traverse each engineering judgment parameter. The engineering judgment parameters include the drill bit mud bag risk index, stick-slip index, and eddy severity. The analysis determines whether each engineering judgment parameter is abnormal and whether the corresponding risk may occur. The machine learning algorithm specifically includes: performing a preliminary judgment of MSE using the k-means algorithm. DH Does a multi-state distribution exist? Then, determine the characteristics of abnormal trends using the LDA or SOS algorithm. The drill bit pressure and drill bit torque are obtained through the following methods: The soft rod model is used for iterative back calculation until the surface drilling pressure and surface torque calculated by the soft rod model are consistent with the measured logging data. Then, the drill bit drilling pressure and drill bit torque are obtained after the soft rod model is corrected. The drill bit pressure and drill bit torque are obtained by iterative back calculation using a soft rod model, which includes the following steps: Assume the initial estimated value of drill bit pressure and drill bit torque is 0; Repeat the following process until the calculated ground drilling pressure and ground torque converge with the measured values. At this point, the corresponding drill bit drilling pressure and drill bit torque are accurate values: The ground drilling pressure and ground torque were calculated by reverse engineering based on the soft rod model. The difference between the actual measured ground drilling pressure and ground torque is obtained and then added to the drill bit drilling pressure and drill bit torque.

2. The method for monitoring bottom hole drilling parameters based on machine learning according to claim 1, characterized in that, Drill bit pressure and drill bit torque are obtained using the following methods: Drill bit pressure = Suspended weight during drilling - Suspended weight during stable idling circulation after lifting off the bottom of the well; Drill bit torque = Torque during drilling - Torque during stable idle circulation after lifting off the bottom of the well.

3. The method for monitoring bottom hole drilling parameters based on machine learning according to claim 1, characterized in that, The drill bit rotation speed is obtained using the following method: When there is no screw drill bit, the drill bit speed equals the ground speed. When a screw drill bit is used, the drill bit speed = ground speed + screw speed.

4. The method for monitoring bottom hole drilling parameters based on machine learning according to claim 1, characterized in that, The drill bit mud bag risk index corresponds to the drill bit mud bag risk, the stick-slip index corresponds to the stick-slip risk, and the vortex severity corresponds to the vortex risk.

5. A method for optimizing bottom hole drilling parameters, characterized in that, Using the machine learning-based wellbore drilling parameter monitoring method according to any one of claims 1-4, the monitoring results are obtained. If the monitoring results show an anomaly, an alarm will be triggered indicating that there may be a risk downhole corresponding to the abnormal engineering judgment parameters; Based on the potential risks, a final bottom hole drilling parameter optimization strategy is proposed.

6. The wellbore drilling parameter optimization method according to claim 5, characterized in that, The final bottom hole drilling parameter optimization strategy based on potential risks includes: To mitigate the risk of drill bit mud buildup, increase displacement and reduce drilling pressure and rotational speed. To mitigate stick-slip risk, reduce drilling pressure and increase drilling speed; To mitigate the risk of vortexing, lift the drill bit off the bottom of the well. Once the severity of vortexing reaches the preset normal value, lower the drill bit back to the bottom of the well, reduce the rotation speed, and increase the drilling pressure.

7. The method for optimizing bottom hole drilling parameters according to claim 6, characterized in that, The final bottom hole drilling parameter optimization strategy based on potential risks includes: To address the risk of drill bit mud buildup, increase displacement by 10%, reduce drilling pressure by 10%, and reduce rotational speed by 10%. To mitigate the risk of stick-slip, reduce drilling pressure by 10% and increase drilling speed by 10%. To address the risk of vortexing, lift the drill bit off the bottom of the well. Once the severity of vortexing reaches the preset normal value, lower the drill bit back to the bottom of the well, reduce the rotation speed by 10%, and increase the drilling pressure by 10%.

Images