Drilling construction method for controlling hole deviation

Abstract

The invention discloses a drilling construction method for controlling hole inclination, which comprises the following steps of: when a real-time stratigraphic dip angle value meets a triggering condition, extracting a preset number of stratigraphic dip angle value time sequence data including the real-time stratigraphic dip angle value; and a data processing and analysis algorithm based on artificial intelligence and deep learning is introduced at the rear end to analyze the time sequence data of the stratigraphic dip values, so that time sequence forward and backward interactive fusion feature representation information of the stratigraphic dip values is captured, and dynamic adjustment of drilling working parameters is carried out. And generating recommended drilling speed value, rotating speed value and drilling pressure value to meet the actual dynamic change of the current geological condition. Therefore, the drilling working parameters can be dynamically adjusted based on the change of the stratigraphic dip angle, so that more intelligent drilling construction control is realized, and the purposes of optimizing the drilling straightness and reducing the occurrence of hole deviation are achieved.

Description

Technical Field

[0001] The present application relates to the field of intelligent adjustment, and more specifically, to a drilling construction method for controlling hole inclination. Background Art

[0002] Bedrock coring under the Antarctic ice sheet is of great scientific value for studying the formation process of Antarctic glaciers and global climate change. By analyzing bedrock samples, researchers can obtain key information about past climate conditions, changes in the Earth's environment, and ice sheet dynamics. In addition, bedrock coring can also help predict future climate change trends and provide a scientific basis for formulating response strategies.

[0003] Bedrock coring under the Antarctic ice is an extremely challenging task. The extreme environmental conditions in the Antarctic, such as low temperatures, high wind speeds, and severe weather, make drilling operations face many difficulties. When the drilling equipment encounters an inclined interface between the ice and the rock, it is very easy for the hole to deflect, which will not only affect the quality of bedrock coring, but may even cause the entire coring operation to fail. In addition, improper selection of drilling parameters will also aggravate the hole deflection phenomenon. For example, too high a mechanical drilling speed will increase the bending moment of the drill bit, resulting in a deflected hole; while too low a drill bit speed may prevent the drill bit from effectively breaking the rock, which will also affect the straightness of the borehole.

[0004] At present, coring of bedrock under the Antarctic ice mainly relies on experienced operators to manually adjust drilling parameters according to geological conditions. Although this method can reduce hole deviation to a certain extent, it still has great uncertainty and limitations due to the lack of real-time and accurate geological data support. In addition, the process of manually adjusting parameters is time-consuming and inefficient, and it is difficult to meet the needs of large-scale, high-precision coring.

[0005] Therefore, an optimized drilling construction plan for controlling hole inclination is desired. Summary of the invention

[0006] In order to solve the above technical problems, the present application is proposed. The embodiment of the present application provides a drilling construction method for controlling hole inclination. When the real-time formation inclination value meets the trigger condition, a preset number of formation inclination value time series data including the real-time formation inclination value is extracted, and a data processing and analysis algorithm based on artificial intelligence and deep learning is introduced at the back end to analyze these formation inclination value time series data, so as to capture the forward and backward interactive fusion feature representation information of the formation inclination value, so as to dynamically adjust the drilling working parameters, generate recommended drilling speed values, rotation speed values ​​and drilling pressure values ​​to meet the actual dynamic changes of current geological conditions. In this way, the drilling working parameters can be dynamically adjusted based on the changes in the formation inclination, so as to achieve more intelligent drilling construction control, and achieve the purpose of optimizing the straightness of the borehole and reducing the occurrence of hole inclination.

[0007] According to one aspect of the present application, a drilling construction method for controlling hole deflection is provided, which comprises:

[0008] Acquire real-time formation dip values ​​through geological detection equipment, and store the real-time formation dip values ​​in a formation dip database;

[0009] In response to the real-time formation inclination value being less than a preset threshold, controlling the drilling equipment to perform drilling with previously set working parameters, wherein the working parameters include a drilling speed value, a rotation speed value, and a drilling pressure value;

[0010] In response to the real-time formation dip value being greater than or equal to a preset threshold, extracting a time queue of a preset number of formation dip values ​​including the real-time formation dip value from the formation dip database;

[0011] Based on the time queue of the formation inclination values, the working parameters of the drilling equipment are adjusted.

[0012] Compared with the prior art, the present application provides a drilling construction method for controlling hole inclination. When the real-time formation inclination value meets the trigger condition, a preset number of formation inclination value time series data including the real-time formation inclination value is extracted, and a data processing and analysis algorithm based on artificial intelligence and deep learning is introduced at the back end to analyze these formation inclination value time series data, so as to capture the forward and backward interactive fusion feature representation information of the formation inclination value, so as to dynamically adjust the drilling working parameters and generate the recommended drilling speed value, rotation speed value and drilling pressure value to meet the actual dynamic changes of the current geological conditions. In this way, the drilling working parameters can be dynamically adjusted based on the change of the formation inclination to achieve more intelligent drilling construction control, so as to achieve the purpose of optimizing the straightness of the borehole and reducing the occurrence of hole inclination. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0014] Figure 1 It is a flow chart of a drilling construction method for controlling hole inclination according to an embodiment of the present application;

[0015] Figure 2 A data flow diagram of a drilling construction method for controlling hole inclination according to an embodiment of the present application;

[0016] Figure 3It is a flowchart of sub-step S4 of the drilling construction method for controlling hole inclination according to an embodiment of the present application. DETAILED DESCRIPTION

[0017] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0018] As shown in this application and claims, unless the context clearly indicates an exception, the words "a", "an", "an" and / or "the" do not refer to the singular and may also include the plural. Generally speaking, the terms "include" and "comprise" only indicate the inclusion of the steps and elements that have been clearly identified, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0019] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules can be used and run on the user terminal and / or server. The modules are only illustrative, and different aspects of the system and method can use different modules.

[0020] Flowcharts are used in the present application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed accurately in order. On the contrary, various steps may be processed in reverse order or simultaneously as required. Meanwhile, other operations may also be added to these processes, or a certain step or several steps of operations may be removed from these processes.

[0021] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0022] During drilling, the formation dip value refers to the inclination angle of the formation layer relative to the horizontal plane. Changes in the formation dip value have a direct impact on the straightness of the borehole (i.e., hole inclination). That is, when the drill bit encounters an inclined formation, the inclination angle of the formation will generate lateral force on the drill bit. This lateral force will cause the drill bit to deviate from the predetermined vertical path, resulting in a hole inclination. In addition, the inclined formation will cause the drill bit to experience different resistance at different locations, further exacerbating the hole inclination phenomenon.

[0023] Not only that, during the drilling process, the change in the formation inclination value has an important influence on the selection and adjustment of drilling working parameters (such as mechanical drilling speed, drill bit speed and drilling pressure). Reasonable selection and adjustment of these parameters can effectively control the hole inclination and improve the straightness and coring quality of the borehole. Specifically, when the mechanical drilling speed is high, the forward speed of the drill bit is accelerated, and the bending moment of the drill bit will also increase. If the formation inclination is large, the lateral force will be more obvious, which is easy to cause the hole to be inclined. A higher drill bit speed can improve the cutting efficiency of the drill bit and reduce the residence time of the drill bit in the formation, thereby reducing the influence of the lateral force, which helps to maintain the straightness of the borehole. A higher drilling pressure will increase the pressure of the drill bit on the formation, making it easier for the drill bit to sink into the formation, especially when the formation inclination is large, the lateral force will be more obvious, which is easy to cause the hole to be inclined.

[0024] Based on this, in the technical solution of the present application, a drilling construction method for controlling hole inclination is proposed. Figure 1 It is a flow chart of a drilling construction method for controlling hole inclination according to an embodiment of the present application. Figure 2 Schematic diagram of data flow of a drilling construction method for controlling hole inclination according to an embodiment of the present application. Figure 1 and Figure 2 As shown, according to the embodiment of the present application, the wind-solar power generation energy storage management method includes the following steps: S1, obtaining a real-time formation inclination value through geological detection equipment, and storing the real-time formation inclination value in a formation inclination database; S2, in response to the real-time formation inclination value being less than a preset threshold, controlling the drilling equipment to drill with the originally set working parameters, the working parameters including the drilling speed value, the rotation speed value and the drilling pressure value; S3, in response to the real-time formation inclination value being greater than or equal to the preset threshold, extracting a time queue of a preset number of formation inclination values ​​including the real-time formation inclination value from the formation inclination database; S4, adjusting the working parameters of the drilling equipment based on the time queue of the formation inclination value.

[0025] In particular, the S1, obtains the real-time formation dip value through the geological detection equipment, and stores the real-time formation dip value in the formation dip database. Specifically, the geological detection equipment is deployed on the drilling bit; the formation dip value is collected through the geological detection equipment to obtain the measured formation dip value; the measured formation dip value is preprocessed and format converted to obtain the real-time formation dip value; the real-time formation dip value is stored in the formation dip database.

[0026] Geological exploration equipment usually includes high-precision sensors and data transmission modules, which can collect stratum inclination values ​​in real time. These devices need to have high reliability and low temperature resistance to adapt to the extreme environmental conditions in Antarctica. After the geological exploration equipment collects the stratum inclination values, it is preprocessed and format converted to ensure the accuracy and consistency of the data. The preprocessing steps may include operations such as filtering, correction, and standardization to eliminate noise and errors. Specifically, digital filters are used to remove high-frequency noise to ensure the smoothness of the data; the collected data is corrected according to known standard values ​​to eliminate sensor errors; and the data is converted into a unified format and unit for subsequent processing and storage.

[0027] The processed real-time formation dip values ​​are stored in the formation dip database for subsequent analysis and use. The formation dip database can be stored locally or in the cloud to ensure data security and accessibility. The database should be designed with efficient read and write performance to support real-time data processing. Common database systems include MySQL, PostgreSQL, and NoSQL databases (such as MongoDB).

[0028] In particular, in S2, in response to the real-time formation inclination value being less than a preset threshold, the drilling equipment is controlled to perform drilling with previously set working parameters, wherein the working parameters include a drilling speed value, a rotation speed value and a drilling pressure value.

[0029] The system continuously monitors the latest formation dip values ​​in the formation dip database and compares them with the preset threshold. The threshold setting needs to be determined according to the specific geological conditions and drilling requirements. For example, it can be set to 5°. The selection of the threshold needs to take into account the complexity of the geological structure and the performance of the drilling equipment to ensure that it can respond to geological changes in a timely manner without adjusting the parameters too frequently. The setting of the threshold is usually based on the formation dip range in the geological report, the response time and adjustment ability of the drilling equipment, and reference to previous successful cases under similar geological conditions.

[0030] The specific steps of real-time monitoring are as follows: The system regularly reads the latest formation inclination value from the formation inclination database. The reading frequency can be adjusted according to actual needs, usually once per second. The real-time formation inclination value read is compared with the preset threshold to determine whether the drilling parameters need to be adjusted. If the real-time formation inclination value is less than the preset threshold, the system will continue to maintain the original parameters; if it is greater than or equal to the preset threshold, the system will enter the parameter adjustment mode.

[0031] When the real-time formation inclination value is less than the preset threshold, the system considers that the current geological conditions are relatively stable and there is no need to adjust the drilling parameters. At this time, the system will enter the working mode of maintaining the original parameters. The system will control the drilling equipment to continue drilling with the previously set working parameters. These working parameters include the forward speed of the drill bit (drilling speed value), the rotation speed of the drill bit (speed value) and the pressure of the drill bit on the formation (bit pressure value). For example, the initial setting of the drilling speed value is 10m / min, the speed value is 100rpm, and the bit pressure value is 50kN.

[0032] The drilling equipment is equipped with an automatic control system that can receive data from the formation inclination database and perform parameter control according to the preset logic. The automatic control system usually includes sensors, controllers and actuators, which can adjust the drilling parameters in real time. The sensors monitor various parameters in the drilling process in real time, such as drilling speed, rotation speed and drilling pressure. The sensors need to have high precision and high reliability to ensure the accuracy of the data. The controller controls the working parameters of the drilling equipment according to the data in the formation inclination database and the preset logic. The controller is usually implemented using an embedded system or PLC (programmable logic controller). The actuator adjusts the actual working parameters of the drilling equipment according to the instructions of the controller. The actuator includes motors, hydraulic systems, pneumatic systems, etc.

[0033] During the system initialization phase, the operator sets the initial drilling speed, rotation speed, and drilling pressure values ​​based on geological reports and experience. These initial parameters need to be optimized based on specific geological conditions and the performance of the drilling equipment. The steps for setting the initial parameters include: understanding the geological structure and characteristics of the drilling area based on the geological report; determining reasonable initial parameters based on successful cases under similar geological conditions in the past; and selecting appropriate initial parameters based on the performance indicators of the drilling equipment. Equipment performance indicators include maximum drilling speed, maximum rotation speed, and maximum drilling pressure.

[0034] When the real-time formation inclination value is less than the preset threshold, the control system will keep these parameters unchanged to ensure the stability and continuity of the drilling process. Specifically, the controller continuously reads the formation inclination value and drilling parameters to ensure the real-time and accuracy of the data. The controller locks the current working parameters to prevent unnecessary adjustments. The controller detects abnormal conditions during the drilling process in real time, such as sensor failure or actuator failure. Once an abnormal condition is detected, the system will sound an alarm and stop drilling to prevent equipment damage.

[0035] The system continuously monitors the formation inclination value and drilling parameters to ensure the smooth progress of the drilling process. The monitoring data can be displayed in real time through a visual interface, which is convenient for operators to observe and adjust. The graphical interface usually includes charts, curves and dashboards, so that operators can intuitively understand the drilling status. When an abnormal situation is detected, the system will issue an alarm to remind the operator to take measures. Abnormal situations include sudden changes in the formation inclination value, abnormal increase or decrease in drilling speed, abnormal fluctuations in speed and abnormal increase in drilling pressure. Alarms can be issued in the form of sound, light or screen prompts.

[0036] All monitoring data and control actions will be recorded for subsequent analysis and optimization. The recorded data include formation inclination values, drilling speed values, rotation speed values, and drilling pressure values, which can be used to evaluate the effect of drilling operations and improve system performance. The recorded data is stored in log files and stored in databases or cloud storage. Log files usually include information such as timestamps, parameter values, and operating instructions. Back up data regularly to ensure data security and integrity. Backup can be achieved through local storage or cloud storage to prevent data loss. By analyzing the recorded data, the effect of drilling operations can be evaluated, potential problems can be discovered, and system performance can be optimized. Data analysis methods include statistical analysis, trend analysis, and machine learning. Through data analysis, the laws of geological structures can be discovered, drilling parameters can be optimized, and drilling efficiency and success rate can be improved.

[0037] In this way, the system can maintain stable drilling parameters when the formation inclination angle is small, ensure the straightness of the borehole and the quality of coring, thereby improving the success rate and efficiency of the drilling operation.

[0038] In particular, in S3, in response to the real-time formation dip value being greater than or equal to a preset threshold, a time queue of a preset number of formation dip values ​​including the real-time formation dip value is extracted from the formation dip database.

[0039] When the real-time formation dip value is greater than or equal to the preset threshold, the system considers that the current geological conditions are unstable and the drilling parameters need to be adjusted. At this time, the system will extract a time queue of a preset number of formation dip values, including the real-time formation dip value, from the formation dip database. A time queue refers to a series of formation dip values ​​arranged in chronological order. These values ​​reflect the trend of formation dip changes over time, which helps the system to understand the changes in geological conditions more comprehensively. The length of the time queue can be adjusted according to actual conditions, usually data from the most recent period of time. For example, the time queue can be set to contain the most recent 10 formation dip values.

[0040] The specific steps of data extraction are as follows: the system first reads the current real-time formation dip value; determines the data range to be extracted according to the preset time window length (e.g. 10 data points); extracts a preset number of formation dip values ​​including the real-time formation dip value from the formation dip database; and arranges the extracted data in chronological order to form a time queue. For example, if the current real-time formation dip value is the 10th data point, the system will extract the formation dip values ​​from the 1st to the 10th data points.

[0041] In particular, the S4 adjusts the working parameters of the drilling equipment based on the time queue of the formation inclination value. Accordingly, in the above-mentioned drilling construction method for controlling hole inclination, if the real-time formation inclination value is greater than or equal to a preset threshold, the time queue of the formation inclination value is extracted from the formation inclination database, and the working parameters of the drilling equipment are adaptively adjusted accordingly. This process ensures that the drilling operation can respond to changes in formation inclination in a timely manner, thereby effectively reducing hole inclination and improving the straightness of the borehole and the quality of coring. Specifically, in a specific example of the present application, Figure 3 As shown, the S4 includes: S41, performing formation inclination forward time series encoding on the time queue of the formation inclination values ​​to obtain forward implicit correlation characteristics of the formation inclination value time series; S42, performing formation inclination backward time series encoding on the time queue of the formation inclination values ​​to obtain backward implicit correlation characteristics of the formation inclination value time series; S43, performing feature complementary fusion processing on the forward implicit correlation characteristics of the formation inclination value time series and the backward implicit correlation characteristics of the formation inclination value time series to obtain a forward-backward interactive compensation fusion representation of the formation inclination value time series; S44, dynamically adjusting working parameters based on the forward-backward interactive compensation fusion representation of the formation inclination value time series to determine recommended drilling speed, rotation speed and drilling pressure values.

[0042] Specifically, the S41 performs forward temporal encoding of the formation dip angle on the time queue of the formation dip angle value to obtain the forward implicit correlation feature of the formation dip angle value time series. That is, in the technical solution which is itself light, the time queue of the formation dip angle value is input into the sequence encoder based on the forward LSTM model to obtain the forward implicit correlation feature vector of the formation dip angle value time series as the forward implicit correlation feature of the formation dip angle value time series. The sequence encoder based on the forward LSTM model can start processing from the earliest time point, can capture the forward trend of the formation dip angle value changing over time and the forward temporal correlation feature, which can reflect the influence of the historical formation dip angle value on the current formation state, and is helpful to understand the cumulative effect of formation changes.

[0043] Specifically, the S42 performs backward temporal encoding of the formation inclination values ​​on the time queue of the formation inclination values ​​to obtain the backward implicit association characteristics of the formation inclination value time series. Here, similarly, the time queue of the formation inclination values ​​is input into the sequence encoder based on the backward LSTM model to obtain the backward implicit association feature vector of the formation inclination value time series as the backward implicit association characteristics of the formation inclination value time series. In particular, the sequence encoder based on the backward LSTM model can capture the backward trend of the formation inclination value changing over time and the backward temporal association characteristics, which can quickly capture the recent changes in the formation inclination value, help to adjust the drilling parameters in time, and reduce the risk of hole deviation.

[0044] Specifically, the S43 performs feature complementary fusion processing on the forward implicit association feature of the formation inclination value time series and the backward implicit association feature of the formation inclination value time series to obtain the forward-backward interactive compensation fusion representation of the formation inclination value time series. It should be understood that the forward implicit association feature vector of the formation inclination value time series and the backward implicit association feature vector of the formation inclination value time series respectively contain the forward time series association feature and the backward time series association feature related to the formation inclination value in the time dimension, which can respectively reflect the historical evolution and instantaneous change of the formation inclination value. Therefore, by fusing the formation inclination value time series implicit association feature vector extracted by the forward and backward LSTM models, more comprehensive and richer formation inclination value time series information can be obtained, which is helpful to more accurately perform dynamic adjustment of drilling working parameters, so as to control the hole inclination. In the process of fusing the forward implicit association features of the formation dip value time series and the backward implicit association features of the formation dip value time series, these two formation dip value time series features will have their own unique semantic information, and there will also be certain redundant or overlapping information. Traditional feature fusion methods may ignore the unique information of these forward and backward time series features of the formation dip value, resulting in an incomplete final fusion feature representation. In addition, there may be misalignment between different time series features, which will further affect the effect of feature fusion. Based on this, in the technical solution of the present application, the forward implicit association features of the formation dip value time series and the backward implicit association features of the formation dip value time series are further subjected to feature complementary fusion processing to obtain a forward-backward interactive compensation fusion representation of the formation dip value time series.

[0045] Specifically, in an embodiment of the present application, the specific process of performing feature complementary fusion processing on the forward implicit association feature of the formation dip value time series and the backward implicit association feature of the formation dip value time series includes: first, performing common feature extraction on the forward implicit association feature vector of the formation dip value time series and the backward implicit association feature vector of the formation dip value time series to obtain a common feature representation vector between the forward and backward features of the formation dip value time series. Specifically, performing common feature extraction on the forward implicit association feature vector of the formation dip value time series and the backward implicit association feature vector of the formation dip value time series using the following common feature extraction formula to obtain a common feature representation vector between the forward and backward features of the formation dip value time series; wherein, the common feature extraction formula is:

[0046]

[0047] Wherein, V1 is the forward implicit correlation feature vector of the formation dip value time series, V2 is the backward implicit correlation feature vector of the formation dip value time series, and W VT and b VT are the modulation weight matrix and the modulation bias vector, respectively, tanh(·) is the tanh function, V J It is a common feature representation vector between the forward and backward features of the formation dip value time series.

[0048] That is, a common feature extraction network is used to extract the common patterns between the forward implicit association feature vector of the formation dip value time series and the backward implicit association feature vector of the formation dip value time series. In some specific embodiments, the common feature extraction network may include multiple levels, each level is responsible for detecting abstract features of a specific level, thereby capturing the intrinsic connection between different inputs and providing a basis for subsequent compensation.

[0049] Next, based on the common feature representation vector between the forward and backward features of the formation dip value time series, the forward implicit association feature vector of the formation dip value time series and the backward implicit association feature vector of the formation dip value time series are subjected to fine-grained compensation optimization processing to obtain the forward fine-grained feature compensation vector of the formation dip value time series and the backward fine-grained feature compensation vector of the formation dip value time series. In a specific example of the present application, based on the common feature representation vector between the forward and backward features of the formation dip value time series, the forward implicit association feature vector of the formation dip value time series and the backward implicit association feature vector of the formation dip value time series are subjected to fine-grained compensation optimization processing to obtain the forward fine-grained feature compensation vector of the formation dip value time series and the backward fine-grained feature compensation vector of the formation dip value time series, including: respectively performing probabilistic processing on the forward implicit association feature vector of the formation dip value time series, the backward implicit association feature vector of the formation dip value time series and the common feature representation vector between the forward and backward features of the formation dip value time series to obtain the probabilistic forward implicit association feature vector of the formation dip value time series, the probabilistic backward implicit association feature vector of the formation dip value time series and the probabilistic common feature representation vector between the forward and backward features of the formation dip value time series; calculating the A forward information fine-grained compensation vector of the formation dip value time series forward implicit association feature vector relative to the common feature representation vector between the forward and backward features of the probabilistic formation dip value time series; a backward information fine-grained compensation vector of the formation dip value time series backward implicit association feature vector relative to the common feature representation vector between the forward and backward features of the probabilistic formation dip value time series; based on the fine-grained compensation vector of the formation dip value time series forward information, the forward implicit association feature vector of the formation dip value time series is compensated to obtain the forward fine-grained feature compensation vector of the formation dip value time series; based on the fine-grained compensation vector of the backward information of the formation dip value time series, the backward implicit association feature vector of the formation dip value time series is compensated to obtain the backward fine-grained feature compensation vector of the formation dip value time series. Specifically, the forward implicit correlation feature vector of the formation dip value time series and the backward implicit correlation feature vector of the formation dip value time series are subjected to fine-grained compensation optimization processing using the following fine-grained compensation optimization formula to obtain the forward fine-grained feature compensation vector of the formation dip value time series and the backward fine-grained feature compensation vector of the formation dip value time series; wherein the fine-grained compensation optimization formula is:

[0050] V a =sigmoid(V J )

[0051] V b =sigmoid(V1)

[0052] V c =sigmoid(V2)

[0053]

[0054]

[0055] V 1b =V1⊙v ba

[0056] V 2c =V2⊙v ca

[0057] Among them, V a 、V b and V c are respectively the common feature representation vector between the forward and backward features of the probabilistic formation dip value time series, the forward implicit association feature vector of the probabilistic formation dip value time series, and the backward implicit association feature vector of the probabilistic formation dip value time series, sigmoid(·) is the sigmoid function, V ai 、V bi and V ci are the common feature representation vector between the forward and backward features of the probabilistic formation dip value time series, the forward implicit association feature vector of the probabilistic formation dip value time series, and the feature values ​​of each position in the backward implicit association feature vector of the probabilistic formation dip value time series, log is the logarithmic function value with base 2, exp(·) is the natural exponential function value with base e as the natural constant, softmax(·) is the softmax function, and v ba is the fine-grained compensation vector of the forward information of the formation dip value time series, v ca is the fine-grained compensation vector of the backward information of the formation dip value time series, V 1b and V 2c They are respectively a forward fine-grained feature compensation vector of the formation dip angle value time series and a backward fine-grained feature compensation vector of the formation dip angle value time series.

[0058] That is, the forward implicit association feature vector of the formation dip value time series, the backward implicit association feature vector of the formation dip value time series, and the common feature representation vector between the forward and backward features of the formation dip value time series are respectively input into the vector probabilization unit based on the Sigmoid function. The Sigmoid function is an activation function that can map real values ​​to the (0,1) interval. In this step, the role of the Sigmoid function is to normalize the feature vector and convert the feature value into a probability form, so as to better measure the importance of the feature, that is, the probabilistic feature vector can more intuitively reflect which features are significant and which are secondary.

[0059] Then, the forward information fine-grained compensation vector of the formation dip value time series of the forward implicit association feature vector of the probabilistic formation dip value time series is calculated relative to the common feature representation vector between the forward and backward features of the probabilistic formation dip value time series. This process involves calculating the difference between the two probabilistic vectors, that is, finding out the information unique to the forward implicit association feature vector of the probabilistic formation dip value time series outside the common features. In this way, those details that are not reflected in the common features and are very critical for model prediction can be determined. At the same time, the backward information fine-grained compensation vector of the formation dip value time series of the probabilistic formation dip value time series is calculated relative to the common feature representation vector between the forward and backward features of the probabilistic formation dip value time series. Similar to the former, the goal is to identify the information in the backward implicit association feature vector of the probabilistic formation dip value time series that is not fully covered in the common features. In this way, it can be ensured that as much unique information as possible is retained in the final feature fusion.

[0060] Specifically, the forward implicit association feature vector of the formation dip value time series and the fine-grained compensation vector of the forward information of the formation dip value time series are input into the fine-grained compensation module to obtain the forward fine-grained feature compensation vector of the formation dip value time series. A fine-grained compensation module is usually a mechanism designed to strengthen or supplement specific features, where the module receives the original feature vector and its corresponding compensation vector, and adds the compensation vector to the original feature vector in some way, such as multiplication, to generate a new feature vector containing more detailed information. At the same time, the backward implicit association feature vector of the formation dip value time series and the fine-grained compensation vector of the backward information of the formation dip value time series are also input into the fine-grained compensation module to obtain the backward fine-grained feature compensation vector of the formation dip value time series. In this way, it is ensured that the forward implicit association feature vector of the formation dip value time series and the backward implicit association feature vector of the formation dip value time series have undergone the same compensation processing, thereby ensuring the consistency and integrity of the final feature fusion.

[0061] Furthermore, the forward fine-grained feature compensation vector of the formation dip value time series, the backward fine-grained feature compensation vector of the formation dip value time series, and the common feature representation vector between the forward and backward features of the formation dip value time series are cascaded to obtain a forward-backward interactive compensation fusion representation vector of the formation dip value time series as the forward-backward interactive compensation fusion representation of the formation dip value time series. Specifically, the forward fine-grained feature compensation vector of the formation dip value time series, the backward fine-grained feature compensation vector of the formation dip value time series, and the common feature representation vector between the forward and backward features of the formation dip value time series are cascaded according to the following cascade formula to obtain a forward-backward interactive compensation fusion representation vector of the formation dip value time series; wherein the cascade formula is:

[0062] v f=concat[V J ; V 1b ; V 2c ]

[0063] Among them, ⊙ is the positional point multiplication, concat[·;·;·] is the cascade operation, and v f It is a representation vector of the forward-backward interactive compensation fusion of the formation dip value time series.

[0064] That is, the forward fine-grained feature compensation vector of the formation dip value time series, the backward fine-grained feature compensation vector of the formation dip value time series and the common feature representation vector between the forward and backward features of the formation dip value time series are cascaded to obtain the forward-backward interactive compensation fusion representation vector of the formation dip value time series as the forward-backward interactive compensation fusion representation feature of the formation dip value time series. Specifically, the process of feature complementary fusion processing is based on capturing the common features between the forward implicit correlation features of the formation dip value time series and the backward implicit correlation features of the formation dip value time series, so as to compensate for the unique information existing in the forward implicit correlation feature vector of the formation dip value time series and the backward implicit correlation feature vector of the formation dip value time series, respectively, and thus achieve a richer and more comprehensive representation of the formation dip value time series features. In this way, not only the common information between the two time series can be retained, but also the unique information in the forward and backward features of the formation dip value time series can be effectively supplemented, effectively solving the problem of information loss in the fusion of forward and backward time series features of the formation dip value, thereby improving the effect of feature fusion.

[0065] Specifically, the S44 dynamically adjusts the working parameters based on the forward-backward interactive compensation fusion representation of the formation inclination value time series to determine the recommended drilling speed value, rotation speed value and drilling pressure value. In a specific example of the present application, the forward-backward interactive compensation fusion representation vector of the formation inclination value time series is input into the working parameter dynamic adjustment module based on the decoder to obtain the recommended drilling speed value, rotation speed value and drilling pressure value. Decoding regression is performed by interactive compensation fusion feature information between the forward time series association feature and the backward time series association feature of the formation inclination value, so as to dynamically adjust the drilling working parameters and generate recommended drilling speed value, rotation speed value and drilling pressure value to meet the actual dynamic changes of the current geological conditions. In this way, the drilling working parameters can be dynamically adjusted based on the change of the formation inclination to achieve more intelligent drilling construction control, so as to achieve the purpose of optimizing the straightness of the borehole and reducing the occurrence of hole deviation.

[0066] Preferably, the forward-backward interactive compensation fusion representation vector of the formation inclination value time series is passed through a decoder-based working parameter dynamic adjustment module to obtain a recommended drilling speed value, rotation speed value and drilling pressure value, including:

[0067] Determine the characteristic mean and characteristic variance of the formation dip value time series forward-backward interactive compensation fusion representation vector, and divide the characteristic mean by the characteristic variance to obtain the formation dip value time series forward-backward interactive compensation fusion probability statistics value:

[0068]

[0069] Wherein, μ represents the characteristic mean of the forward-backward interactive compensation fusion representation vector of the formation dip value time series, σ represents the characteristic variance of the forward-backward interactive compensation fusion representation vector of the formation dip value time series, and ρ represents the probability statistic value of the forward-backward interactive compensation fusion of the formation dip value time series;

[0070] The forward-backward interactive compensation fusion representation vector of the formation dip value time series is point-multiplied with the inverse of the maximum eigenvalue in the forward-backward interactive compensation fusion representation vector of the formation dip value time series to obtain the forward-backward interactive compensation fusion probability constraint vector of the formation dip value time series:

[0071] V1=V⊙v max -1

[0072] Wherein, V represents the forward-backward interactive compensation fusion representation vector of the formation dip value time series, v max -1 represents the inverse of the maximum eigenvalue in the forward-backward interactive compensation fusion representation vector of the formation dip value time series, ⊙ represents the dot product, and V1 represents the forward-backward interactive compensation fusion probability constraint vector of the formation dip value time series;

[0073] After dot-wise addition of the forward-backward interactive compensation fusion probability constraint vector of the formation dip value time series and the forward-backward interactive compensation fusion probability statistical value of the formation dip value time series, the logarithmic value with base 2 is calculated to obtain the forward-backward interactive compensation fusion information interaction vector of the formation dip value time series:

[0074] V2=log(V1⊕ρ)

[0075] Wherein, log represents a logarithmic operation with base 2, ⊕ represents point addition, and V2 represents the forward-backward interactive compensation fusion information interactive vector of the formation dip value time series;

[0076] After performing point subtraction on the forward-backward interactive compensation fusion probability constraint vector of the formation dip value time series with the forward-backward interactive compensation fusion probability statistic value of the formation dip value time series, the inverse of each eigenvalue is calculated to obtain the forward-backward interactive compensation fusion sequence constraint vector of the formation dip value time series:

[0077]

[0078] in, represents point subtraction, V3 represents the forward-backward interactive compensation fusion sequence constraint vector of the formation dip value time series;

[0079] The formation dip value time series forward-backward interactive compensation fusion information interaction vector is interpolated with the formation dip value time series forward-backward interactive compensation fusion sequence constraint vector to obtain an optimized formation dip value time series forward-backward interactive compensation fusion representation vector;

[0080] The optimized formation inclination value time series forward-backward interactive compensation fusion representation vector is passed through a decoder-based working parameter dynamic adjustment module to obtain recommended drilling speed value, rotation speed value and drilling pressure value.

[0081] In the case where the forward implicit association feature vector of the formation dip value time series and the backward implicit association feature vector of the formation dip value time series respectively represent the time series implicit association features of the formation dip values ​​in the forward time dimension direction and the backward time dimension direction, when performing inter-feature compensatory interaction with common features as a mask, the formation dip value time series forward-backward interactive compensation fusion representation vector will also have inter-feature temporal compensation interactive compensation diversity under the time series dimension difference, thereby having feature interaction time series distribution diversity based on feature mapping deviation compensation, thereby, when the formation dip value time series forward-backward interactive compensation fusion representation vector is decoded and regressed through the decoder, it will affect the accuracy of the decoding result.

[0082] That is, considering that the weight matrix of the decoder acts on the forward-backward interactive compensation fusion representation vector of the formation inclination value timing, the distribution diversity of the forward-backward interactive compensation fusion representation vector of the formation inclination value timing causes the uncertainty of the weight stimulus parameters of the weight matrix of the decoder, resulting in the lack of posterior probability density inference of the forward-backward interactive compensation fusion representation vector of the formation inclination value timing through the weight matrix, which will affect the accuracy of the decoding result.

[0083] In this case, the probability statistics characteristics of the forward-backward interactive compensation fusion representation vector of the formation inclination value time series are used to simulate the mesoscale interaction structure under probability constraints between the eigenvalue scale and the eigenvector scale of the forward-backward interactive compensation fusion representation vector of the formation inclination value time series, so as to construct a bidirectional latent variable motif based on the mesoscale short sequence relative to the probability statistics value to perform mesoscale bidirectional migration, and perform posterior recovery based on the sequence constraints on the interactive information, so as to improve the convergence effect in the probability density domain, and improve the accuracy of the recommended drilling speed value, rotation speed value and drilling pressure value obtained by the forward-backward interactive compensation fusion representation vector of the formation inclination value time series through the decoder-based working parameter dynamic adjustment module. In this way, the drilling working parameters can be dynamically adjusted based on the change of the formation inclination to achieve more intelligent drilling construction control.

[0084] In summary, according to the drilling construction method for controlling hole inclination according to the embodiment of the present application, when the real-time formation inclination value meets the trigger condition, a preset number of formation inclination value time series data including the real-time formation inclination value is extracted, and a data processing and analysis algorithm based on artificial intelligence and deep learning is introduced at the back end to analyze these formation inclination value time series data, so as to capture the forward and backward interactive fusion feature representation information of the formation inclination value, so as to dynamically adjust the drilling working parameters, generate recommended drilling speed values, rotation speed values ​​and drilling pressure values ​​to meet the actual dynamic changes of current geological conditions. In this way, the drilling working parameters can be dynamically adjusted based on the changes in the formation inclination, so as to realize more intelligent drilling construction control, and achieve the purpose of optimizing the straightness of the borehole and reducing the occurrence of hole inclination.

[0085] The embodiments of the present disclosure have been described above, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and changes will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The selection of terms used herein is intended to best explain the principles of the embodiments, practical applications, or improvements to the technology in the market, or to enable other persons of ordinary skill in the art to understand the embodiments disclosed herein.

Claims

1. A drilling construction method for controlling hole inclination, characterized in that: include: Acquire real-time formation dip values ​​through geological detection equipment, and store the real-time formation dip values ​​in a formation dip database; In response to the real-time formation inclination value being less than a preset threshold, controlling the drilling equipment to perform drilling with previously set working parameters, wherein the working parameters include a drilling speed value, a rotation speed value, and a drilling pressure value; In response to the real-time formation dip value being greater than or equal to a preset threshold, extracting a time queue of a preset number of formation dip values ​​including the real-time formation dip value from the formation dip database; Based on the time queue of the formation inclination values, the working parameters of the drilling equipment are adjusted.

2. The drilling construction method for controlling hole inclination according to claim 1, characterized in that: Acquiring real-time formation dip values ​​through geological detection equipment and storing the real-time formation dip values ​​in a formation dip database includes: deploying the geological exploration equipment on a drilling bit; Collecting the formation dip value by means of the geological exploration equipment to obtain the measured formation dip value; Preprocessing and formatting the measured formation dip value to obtain the real-time formation dip value; The real-time formation dip value is stored in the formation dip database.

3. The drilling construction method for controlling hole inclination according to claim 2 is characterized in that: Adjusting the working parameters of the drilling equipment based on the time queue of the formation dip angle value includes: Performing formation dip forward time series coding on the time queue of the formation dip values ​​to obtain a formation dip value time series forward implicit correlation feature; Performing formation dip backward time series coding on the time queue of the formation dip values ​​to obtain a formation dip value time series backward implicit correlation feature; Performing feature complementary fusion processing on the forward implicit correlation feature of the formation dip value time series and the backward implicit correlation feature of the formation dip value time series to obtain a forward-backward interactive compensation fusion representation of the formation dip value time series; Based on the forward-backward interactive compensation fusion representation of the formation inclination value time series, the working parameters are dynamically adjusted to determine the recommended drilling speed value, rotation speed value and drilling pressure value.

4. The drilling construction method for controlling hole inclination according to claim 3 is characterized in that: Performing formation dip forward time series encoding on the time queue of the formation dip values ​​to obtain a formation dip value time series forward implicit association feature, including: inputting the time queue of the formation dip values ​​into a sequence encoder based on a forward LSTM model to obtain a formation dip value time series forward implicit association feature vector as the formation dip value time series forward implicit association feature.

5. The drilling construction method for controlling hole inclination according to claim 4 is characterized in that: The time queue of the formation dip angle values ​​is subjected to backward temporal encoding of the formation dip angle to obtain a backward implicit correlation feature of the formation dip angle value time series, including: inputting the time queue of the formation dip angle values ​​into a sequence encoder based on a backward LSTM model to obtain a backward implicit correlation feature vector of the formation dip angle value time series as the backward implicit correlation feature of the formation dip angle value time series.

6. The drilling construction method for controlling hole inclination according to claim 5, characterized in that: The forward implicit correlation feature of the formation dip value time series and the backward implicit correlation feature of the formation dip value time series are subjected to feature complementary fusion processing to obtain a forward-backward interactive compensation fusion representation of the formation dip value time series, including: Extracting common features from the forward implicit correlation feature vector of the formation dip value time series and the backward implicit correlation feature vector of the formation dip value time series to obtain a common feature representation vector between the forward and backward features of the formation dip value time series; Based on the common feature representation vector between the forward and backward features of the formation dip value time series, fine-grained compensation optimization processing is performed on the forward implicit association feature vector of the formation dip value time series and the backward implicit association feature vector of the formation dip value time series to obtain a forward fine-grained feature compensation vector of the formation dip value time series and a backward fine-grained feature compensation vector of the formation dip value time series; The forward fine-grained feature compensation vector of the formation dip value time series, the backward fine-grained feature compensation vector of the formation dip value time series and the common feature representation vector between the forward and backward features of the formation dip value time series are cascaded to obtain a forward-backward interactive compensation fusion representation vector of the formation dip value time series as the forward-backward interactive compensation fusion representation of the formation dip value time series.

7. The drilling construction method for controlling hole inclination according to claim 6, characterized in that: Based on the common feature representation vector between the forward and backward features of the formation dip value time series, fine-grained compensation optimization processing is performed on the forward implicit association feature vector of the formation dip value time series and the backward implicit association feature vector of the formation dip value time series to obtain a forward fine-grained feature compensation vector of the formation dip value time series and a backward fine-grained feature compensation vector of the formation dip value time series, including: Probabilistically processing the forward implicit association feature vector of the formation dip value time series, the backward implicit association feature vector of the formation dip value time series, and the common feature representation vector between the forward and backward features of the formation dip value time series to obtain a probabilistic forward implicit association feature vector of the formation dip value time series, a probabilistic backward implicit association feature vector of the formation dip value time series, and a probabilistic common feature representation vector between the forward and backward features of the formation dip value time series; Calculate the forward information fine-grained compensation vector of the formation dip value time series of the forward implicit correlation feature vector of the probabilistic formation dip value time series relative to the common feature representation vector between the forward and backward features of the probabilistic formation dip value time series; Calculate a fine-grained compensation vector of the backward information of the formation dip value time series of the probabilistic formation dip value time series of the backward implicit correlation feature vector relative to the common feature representation vector between the forward and backward features of the probabilistic formation dip value time series; Based on the formation dip value time series forward information fine-grained compensation vector, the formation dip value time series forward implicit associated feature vector is compensated to obtain the formation dip value time series forward fine-grained feature compensation vector; The backward implicit association feature vector of the formation dip value time series is compensated based on the backward information fine-grained compensation vector of the formation dip value time series to obtain the backward fine-grained feature compensation vector of the formation dip value time series.

8. The drilling construction method for controlling hole inclination according to claim 7, characterized in that: Based on the forward-backward interactive compensation fusion representation of the formation inclination value time series, the working parameters are dynamically adjusted to determine the recommended drilling speed value, rotation speed value and drilling pressure value, including: inputting the forward-backward interactive compensation fusion representation vector of the formation inclination value time series into a decoder-based working parameter dynamic adjustment module to obtain the recommended drilling speed value, rotation speed value and drilling pressure value.

Images