Drilling parameter recommendation method and device based on AI model

Abstract

The invention provides a drilling parameter recommendation method and device based on an AI model, and relates to the technical field of petroleum engineering, and the method comprises the steps: cleaning original data; inputting the cleaned data into a parameter recommendation subsystem; calculating a drilling performance measurement index according to the physical model, and constructing a function relationship between the drilling parameters and the drilling performance measurement index through dynamic learning by taking optimization of the drilling performance measurement index as a target; and recommending the optimized drilling parameters to a driller according to the function relationship. According to the method, the AI model is applied to drilling parameter recommendation, different working conditions can be identified, original data can be cleaned, and drilling parameters capable of optimizing target indexes in different modes are recommended by adopting a scheme of combining linear exploration and nonlinear application; according to the method, the defects that a traditional drilling parameter recommendation method is tedious in process, long in consumed time, poor in precision and the like are overcome, the optimal drilling parameters can be dynamically and efficiently recommended, the drilling speed is optimized, and therefore the purposes of reducing cost and improving efficiency in the drilling process are achieved.

Description

Technical Field

[0001] The present invention relates to the field of petroleum engineering technology, and in particular to a drilling parameter recommendation method and device based on an AI model. Background Art

[0002] This section is intended to provide a background or context to the embodiments of the invention recited in the claims. No admission is made that the description herein is prior art by inclusion in this section.

[0003] Drilling parameter optimization and recommendation is an important part of drilling engineering, which directly affects safe drilling, high-quality production and efficient utilization. How to quickly and accurately recommend drilling parameters that are conducive to achieving target indicators has become one of the key research directions of modern drilling control optimization technology. However, the complexity and diversity of drilling parameters lead to large-scale, high-risk, and high-investment problems in drilling projects. The method of recommending drilling parameters based on empirical models is difficult to meet the drilling construction goals, which seriously affects drilling efficiency.

[0004] In summary, there is an urgent need for a technical solution that can overcome the above-mentioned defects and realize the analysis, optimization and recommendation of drilling parameters. Summary of the invention

[0005] In order to solve the problems existing in the prior art, the present invention proposes a drilling parameter recommendation method and device based on an AI model.

[0006] In a first aspect of an embodiment of the present invention, a drilling parameter recommendation method based on an AI model is proposed, comprising:

[0007] Acquire raw data, and clean the raw data;

[0008] Input the cleaned data into the parameter recommendation subsystem; wherein the drilling performance metric index is calculated according to the physical model, and the functional relationship between the drilling parameters and the drilling performance metric index is constructed through dynamic learning with the goal of optimizing the drilling performance metric index;

[0009] The optimized drilling parameters are recommended to the driller according to the functional relationship.

[0010] In a second aspect of an embodiment of the present invention, a drilling parameter recommendation device based on an AI model is proposed, comprising:

[0011] A data preprocessing subsystem, used to obtain raw data and clean the raw data;

[0012] The parameter recommendation subsystem is used to calculate the drilling performance metrics according to the physical model, with the goal of optimizing the drilling performance metrics, and to construct the functional relationship between the drilling parameters and the drilling performance metrics through dynamic learning; and to recommend the optimized drilling parameters to the driller according to the functional relationship.

[0013] In a third aspect of an embodiment of the present invention, a computer device is proposed, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements a drilling parameter recommendation method based on an AI model when executing the computer program.

[0014] In a fourth aspect of an embodiment of the present invention, a computer-readable storage medium is proposed, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, a drilling parameter recommendation method based on an AI model is implemented.

[0015] In a fifth aspect of an embodiment of the present invention, a computer program product is proposed, wherein the computer program product comprises a computer program, and when the computer program is executed by a processor, a drilling parameter recommendation method based on an AI model is implemented.

[0016] The present invention discloses a drilling parameter recommendation method and device based on an AI model, which collects raw data from a well site and performs corresponding preprocessing, and provides the data to a parameter recommendation system for use. The parameter recommendation subsystem receives data from a data processing subsystem, calculates various metrics on the data, builds a functional relationship between drilling parameters and target indicators based on AI model learning, and recommends drilling operation parameters that can improve the target indicators based on the functional relationship. The present invention applies an AI model to drilling parameter recommendation, which can identify different working conditions and clean the raw data. It adopts a solution that combines linear exploration with nonlinear application to recommend drilling parameters that can optimize the target indicators under different modes. The present invention makes up for the shortcomings of traditional drilling parameter recommendation methods, such as cumbersome processes, long time consumption, and poor accuracy. It can dynamically and efficiently recommend optimal drilling parameters and optimize drilling speed, thereby achieving the purpose of reducing costs and increasing efficiency in the drilling process. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0018] Figure 1 It is a flowchart of a drilling parameter recommendation method based on an AI model according to an embodiment of the present invention.

[0019] Figure 2 It is a schematic diagram of a process of cleaning raw data according to an embodiment of the present invention.

[0020] Figure 3 It is a flowchart of performing data preprocessing according to working conditions according to an embodiment of the present invention.

[0021] Figure 4 It is a schematic diagram of the process of cleaning data using the expert threshold method according to an embodiment of the present invention.

[0022] Figure 5 It is a schematic diagram of the process of cleaning data using the speed constraint method according to an embodiment of the present invention.

[0023] Figure 6 It is a schematic diagram of a flow chart of a parameter recommendation subsystem of an embodiment of the present invention.

[0024] Figure 7 It is a schematic diagram of a process for optimizing drilling performance according to an embodiment of the present invention.

[0025] Figure 8 It is a schematic diagram of the principles of the three stages of exploration, learning and application of the parameter recommendation subsystem of an embodiment of the present invention.

[0026] Fig. 9 4 is a schematic diagram of a mode switching process of a parameter recommendation subsystem according to an embodiment of the present invention.

[0027] Fig.10 Schematic diagram of speed value distribution according to an embodiment of the present invention.

[0028] Fig.11 It is a schematic diagram of a support vector regression model according to an embodiment of the present invention.

[0029] Fig.12 It is a schematic diagram of adjusting (reducing) exploration mode parameters according to an embodiment of the present invention.

[0030] Fig.13 It is a schematic diagram of adjusting (increasing) exploration mode parameters according to an embodiment of the present invention.

[0031] Fig.14 Schematic diagram of a random forest algorithm according to an embodiment of the present invention.

[0032] Fig.15 It is a schematic diagram of the architecture of a drilling parameter recommendation device based on an AI model according to an embodiment of the present invention.

[0033] Fig.16 It is a schematic diagram of the structure of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0034] The principles and spirit of the present invention will be described below with reference to several exemplary embodiments. It should be understood that these embodiments are provided only to enable those skilled in the art to better understand and implement the present invention, and are not intended to limit the scope of the present invention in any way. On the contrary, these embodiments are provided to make the present disclosure more thorough and complete, and to fully convey the scope of the present disclosure to those skilled in the art.

[0035] Those skilled in the art will appreciate that the embodiments of the present invention may be implemented as a system, device, apparatus, method or computer program product. Therefore, the present disclosure may be implemented in the following forms, namely: complete hardware, complete software (including firmware, resident software, microcode, etc.), or a combination of hardware and software.

[0036] According to an embodiment of the present invention, a drilling parameter recommendation method and device based on an AI model are proposed, which relates to the field of petroleum engineering technology. The present invention uses an expert threshold method and a speed constraint method to clean the original data through a data preprocessing subsystem according to the working conditions, and then transmits the processed data to a parameter recommendation subsystem (ROP-OPT-AI), calculates drilling performance metrics based on a physical model, dynamically learns to construct a functional relationship between drilling parameters and drilling performance improvement goals, and recommends better drilling parameters to the driller based on the functional relationship.

[0037] The principle and spirit of the present invention are explained in detail below with reference to several representative embodiments of the present invention.

[0038] Figure 1 FIG. 1 is a flow chart of a drilling parameter recommendation method based on an AI model according to an embodiment of the present invention. Figure 1 As shown, the method includes:

[0039] Step 101, obtaining original data and cleaning the original data;

[0040] Step 102, inputting the cleaned data into a parameter recommendation subsystem; wherein the drilling performance metric index is calculated according to the physical model, with the goal of optimizing the drilling performance metric index, and constructing a functional relationship between the drilling parameters and the drilling performance metric index through dynamic learning;

[0041] Step 103: recommending optimized drilling parameters to the driller according to the functional relationship.

[0042] In actual application scenarios, the drilling parameter recommendation method based on the AI ​​model proposed in the present invention works as follows: the data preprocessing subsystem uses an independent thread to obtain raw data through the HTTP protocol, and performs corresponding data preprocessing according to the working conditions. The expert threshold method and speed constraint method are used to clean the data, and the mechanical drilling rate (ROP) is calculated according to the well depth interval. The processed micro-element well section data is saved in the raw data buffer, and when the data accumulates to a certain extent, it is sent to the parameter recommendation subsystem.

[0043] The parameter recommendation subsystem calculates drilling performance metrics such as mechanical specific energy and optimization targets based on the data. The mechanical specific energy calculation process is divided into three conditions according to the working conditions: composite drilling with screw, composite drilling without screw, and sliding drilling. The construction of the optimization target considers maximizing the mechanical drilling speed and minimizing the mechanical specific energy at the same time.

[0044] With the goal of improving drilling performance metrics, a combination of linear exploration and nonlinear application is used to recommend drilling parameters, which can not only show the trend of data changes but also accurately fit the distribution of historical data. After the parameter recommendation subsystem is started, it enters the exploration mode. Based on the trend of linear model fitting, it explores and collects data in the direction of ROP increase; after sufficient exploration, it enters the learning mode, and learns the functional relationship between ROP and drilling parameters in the current exploration data based on the model, and gives the optimal drilling parameter combination; then it maintains the optimal drilling parameters and enters the application mode, which continuously monitors the current drilling status while breaking rocks efficiently and steadily, and automatically enters the learning mode or exploration mode according to the degree of ROP performance degradation, so as to adapt to the current drilling status.

[0045] In combination with the above working principle, the drilling parameter recommendation method based on the AI ​​model proposed in the present invention is described in detail below.

[0046] Step 101, data cleaning:

[0047] refer to Figure 2 In step 101, the original data is obtained, and the specific process of cleaning the original data is as follows:

[0048] Step 1011, using the data preprocessing subsystem, using an independent thread to obtain original data through the HTTP protocol;

[0049] Step 1012, preprocessing the data according to the working conditions, using the expert threshold method and the speed constraint method to clean the data, and calculating the mechanical drilling speed according to the well depth interval, and saving the processed micro-element well section data to the original data buffer;

[0050] Step 1013: When the data accumulates to a certain extent, it is sent to the parameter recommendation subsystem.

[0051] The following mainly describes in detail each processing link of step 1012.

[0052] In one embodiment, the detailed process of data preprocessing according to the working conditions is as follows: Figure 3 shown.

[0053] Working condition pretreatment:

[0054] According to the working condition of the currently received data and the switching status of the previous and next working conditions, the original data is preprocessed accordingly;

[0055] Among them, during preprocessing, the drilling status is first determined based on the well site data; if it is the drilling status, the sliding drilling status is further determined; if it is not the drilling status, the status parameters are updated, and the data buffer area and the mechanical drilling speed calculation queue are cleared;

[0056] When judging the sliding drilling state, if it is the sliding drilling state, the screw speed and screw torque are recalculated and the state parameters are updated; if it is not the sliding drilling state, no processing is performed.

[0057] In one embodiment, the detailed process of using the expert threshold method and the speed constraint method to clean data is as follows: Figure 4 and Figure 5 shown.

[0058] Expert Threshold Method:

[0059] Abnormal values ​​in the monitoring data are detected by setting expert thresholds, wherein the expert thresholds include at least: drilling pressure range, composite drilling torque range, sliding drilling torque range, composite drilling ground speed range, sliding drilling screw speed range, inlet flow range, riser pressure range, and drilling speed range.

[0060] Speed ​​Constraint Method:

[0061] The expert threshold method is used to clean outliers, and the retained data is further judged using the speed constraint method to clean up data that does not meet the upper and lower bounds of the speed value.

[0062] In one embodiment, the specific process of calculating the mechanical drilling speed according to the well depth interval is:

[0063] The mechanical drilling speed is determined according to the well depth interval, and a queue is maintained according to the well depth interval. The window length of the queue changes dynamically. Each time the mechanical drilling speed of a new data is calculated, the calculated data is added to the end of the queue. The calculation method of the mechanical drilling speed is:

[0064]

[0065] Among them, ROP now It is the tail data of the newly added team; Depth1, Depth nowThey are the first and last data in the well depth data queue respectively; Time1, Time now They are respectively the first data and the last data in the time data queue.

[0066] In one embodiment, the specific process of saving the processed micro-element well section data to the original data buffer is as follows:

[0067] The processed data is saved in the micro-element well section dynamic buffer; when the interval between the well depth of the last data in the micro-element well section dynamic buffer and the well depth of the first data in the micro-element well section dynamic buffer is greater than or equal to the buffer size, all the data in the micro-element well section dynamic buffer are sent to the parameter recommendation subsystem, and after the sending is completed, the micro-element well section dynamic buffer is cleared and the data is accumulated again.

[0068] refer to Figure 6 , is a flow chart of the recommended parameters of the parameter recommendation subsystem of one embodiment of the present invention. The parameter recommendation subsystem (ROP-OPT-AI) obtains micro-element well section data from the data preprocessing subsystem, filters and merges it into one piece of data, calculates the metrics (mechanical specific energy MSE, optimization target Objective) used to characterize the drilling performance based on the predefined physical model, and adds the calculation results to the predefined data structure. When the data is collected to a certain extent, the algorithm will dynamically learn the functional relationship between the drilling operation parameters (such as WOB, RPM, Qpdm) and the optimization target Objective and ROP, and recommend better drilling operation parameters to the driller based on the functional relationship, so as to gradually improve the drilling efficiency.

[0069] Step 102, optimizing drilling performance:

[0070] refer to Figure 7 In step 102, the cleaned data is input into the parameter recommendation subsystem; wherein, the drilling performance metric index is calculated according to the physical model, and the functional relationship between the drilling parameters and the drilling performance metric index is constructed through dynamic learning with the goal of optimizing the drilling performance metric index. The specific process is as follows:

[0071] Step 1021, calculating drilling performance metrics according to the cleaned data, wherein the drilling performance metrics at least include: mechanical specific energy and optimization target;

[0072] Step 1022, setting the goal of improving the drilling performance metric index, and adopting a solution combining linear exploration and nonlinear application to recommend drilling parameters; wherein, after the parameter recommendation subsystem is started, it enters the exploration mode, and based on the trend of linear model fitting, explores and collects data in the direction of increasing the mechanical drilling rate; after sufficient exploration, it enters the learning mode, and learns the functional relationship between the mechanical drilling rate and the drilling parameters in the current exploration data based on the model, and determines the optimal drilling parameter combination; maintains the optimal drilling parameters and enters the application mode, monitors the current drilling status while breaking the rock, and automatically enters the learning mode or the exploration mode according to the degree of performance degradation of the mechanical drilling rate.

[0073] In one embodiment, the method further comprises:

[0074] The time and well depth of the last data in the micro-element well section data are taken out as the time and well depth of the merged data, and the other data are averaged by Kalman filtering to obtain the merged data, which is then input into the parameter recommendation subsystem.

[0075] In one embodiment, the drilling performance metric is calculated based on the cleaned data, including:

[0076] The calculation formula of mechanical specific energy is:

[0077]

[0078] Wherein, MSE is mechanical specific energy; EFF is s is the efficiency coefficient; WOB is the bit pressure; T is the torque; RPM is the speed; ROP is the mechanical drilling speed; d B is the drill bit diameter; mechanical specific energy is the work done by the drill bit to break a unit volume of rock, which is used to evaluate drilling efficiency. The larger the mechanical specific energy value, the lower the rock breaking efficiency of the drill bit, the poor adaptability of the drill bit to the bottom layer, and the drilling parameters need to be optimized;

[0079] The calculation formula of the optimization target is as follows:

[0080]

[0081] Among them, Objective is the optimization target, and the optimization target is to maximize the mechanical drilling speed while minimizing the mechanical specific energy; ROP0 and MSE0 are normalized quantities used to remove the influence of dimension.

[0082] In one embodiment, the modes of the parameter recommendation subsystem include: exploration mode, learning mode and application mode. Figure 8 , which is a schematic diagram of the principles of the three stages of exploration, learning, and application of a parameter recommendation subsystem of an embodiment of the present invention.

[0083] In one embodiment, reference Fig. 9, the mode switching process of the parameter recommendation subsystem is:

[0084] Determine whether the parameter recommendation subsystem is in exploration mode; if in exploration mode, further determine whether the exploration is sufficient; if not in exploration mode, evaluate drilling performance;

[0085] When judging whether the exploration is sufficient, if the exploration is sufficient, switch to the learning mode and use the nonlinear model to learn the optimal parameters; if the exploration is insufficient, switch to the exploration mode and use the linear model to explore the space;

[0086] When evaluating drilling performance metrics, determine whether the drilling performance has declined; if not, switch to application mode and maintain the current parameters for rock breaking; if decline occurs, determine whether the decline indicator reaches the decline threshold;

[0087] When judging whether the recession indicator has reached the recession threshold, if it has not reached it, it switches to the learning mode and uses the nonlinear model to learn the optimal parameters; if it has reached it, it switches to the exploration mode and performs spatial exploration through the linear model.

[0088] In one embodiment, the mode switching method in the exploration mode includes:

[0089] The exploration mode of the parameter recommendation subsystem is used to guide the driller to explore in the direction of increasing the mechanical drilling speed, obtain the corresponding relationship between the operating parameters and the mechanical drilling speed by exploring the parameter space, and find the area with the maximum mechanical drilling speed;

[0090] When the exploration is sufficient, the mode switches from exploration mode to learning mode. The switching condition is to determine whether the predetermined number of explorations has been completed. Among them, a minimum number of explorations and a maximum number of explorations are set. When the minimum number of explorations is reached, the current drilling speed is determined before each exploration. If the drilling speed exceeds the drilling speed threshold, the exploration continues, otherwise it switches to learning mode. When the maximum number of explorations is exceeded, the exploration ends.

[0091] In one embodiment, a mode switching method in a learning mode or an application mode includes:

[0092] After the parameter recommendation subsystem switches to the learning mode or the application mode, it is in the stage of dynamically monitoring the drilling performance; if the drilling performance reaches the performance threshold, the current operating parameters are maintained to enter the application mode; if the drilling performance declines and does not reach the decline threshold, the drilling parameters are adjusted to enter the learning mode; if the drilling performance declines and reaches the decline threshold, it switches to the exploration stage;

[0093] Among them, the mechanical drilling speed score is used to judge the current drilling performance, and the current drilling effect is judged by setting two mechanical drilling speed score thresholds. If the mechanical drilling speed score falls into an interval below the first threshold, it is determined that the drilling performance has seriously declined. If the mechanical drilling speed score falls into an interval above the second threshold, it is determined that the current drilling performance is stable. If the mechanical drilling speed score falls into an interval between the first threshold and the second threshold, it is determined that the drilling performance has slightly declined. The calculation method of the mechanical drilling speed score is:

[0094]

[0095] ROP score Score for rate of penetration; ROP now Current mechanical drilling speed; ROP history is the historical mechanical drilling speed; the parameter recommendation subsystem calculates the mechanical drilling speed scores of the current anchor point and the historical anchor point every time it receives 10 micro-element well section data.

[0096] In order to improve the robustness of the determination result, the method further includes:

[0097] Calculate multiple consecutive ROP scores;

[0098] According to the mechanical drilling speed score threshold, multiple mechanical drilling speed scores are divided into different intervals, and an interval with the largest number of divided mechanical drilling speed scores is selected as the current determination result.

[0099] Step 103, recommend drilling parameters:

[0100] In step 103, optimized drilling parameters are recommended to the driller according to the functional relationship, including parameter recommendations under different working conditions and parameter recommendations under different modes.

[0101] 1) Parameter recommendations under different working conditions:

[0102] When making parameter recommendations under different working conditions, for composite drilling, the functional relationship between drilling operation parameters and drilling performance improvement targets is:

[0103] ROP=f(WOB,RPM surface ,Qpdm)

[0104] OBJ=f(WOB,RPM surface ,Qpdm)

[0105] ROP is the mechanical drilling speed; OBJ is the optimization target; WOB is the bit pressure; RPM surface is the ground speed; Qpdm is the displacement; the recommended parameters are given according to the functional relationship;

[0106] For sliding drilling, the functional relationship between drilling operation parameters and drilling performance improvement targets is:

[0107] ROP=f(WOB,Qpdm)

[0108] OBJ=f(WOB,Qpdm)

[0109] Recommended parameters are given according to the functional relationship.

[0110] 2) Recommended exploration mode parameters:

[0111] When making parameter recommendations in the exploration mode, the support vector regression model is used to fit the relationship between drilling operation parameters, mechanical drilling speed and optimization objectives, and the recommended parameters are determined based on the fitted relationship;

[0112] Among them, for compound drilling or sliding drilling, the support vector regression model is used to fit the linear relationship of the optimization target calculation method. According to the slope obtained by parameter fitting and based on the boundary value of the explored parameter interval, the recommended parameters are adjusted in the direction of increasing the mechanical drilling speed according to the change of the slope. If the adjusted recommended parameters exceed the values ​​specified in the drilling design, the drilling design will be used as the recommended parameters.

[0113] 3) Learning mode parameter recommendation:

[0114] When making parameter recommendations in the learning mode, a random forest regression model is used for nonlinear fitting to fit the relationship between drilling operation parameters, mechanical drilling speed and optimization objectives, and the recommended parameters are determined based on the fitted relationship;

[0115] Among them, for compound drilling or sliding drilling, the random forest regression model is used to fit the linear relationship of the optimization target calculation method. After fitting, a grid search is performed in the parameter space based on the random forest regression model to obtain the parameters corresponding to the maximum point of the mechanical drilling speed or optimization target, and to determine whether the parameters meet the conditions. If they meet the conditions, recommendations are made.

[0116] 4) Application mode parameter recommendation:

[0117] When the parameter recommendation is performed in the application mode, since the drilling performance of this mode is relatively stable, the previously recommended parameters are adopted to keep the system in a steady state.

[0118] In order to explain the above-mentioned drilling parameter recommendation method based on the AI ​​model more clearly, a detailed description is given below in conjunction with a specific embodiment.

[0119] The drilling parameter recommendation method based on the AI ​​model proposed in the present invention is mainly implemented through a data preprocessing subsystem and a parameter recommendation subsystem.

[0120] 1. Data preprocessing subsystem

[0121] The data preprocessing subsystem uses an independent thread to continuously obtain raw data from the well site equipment through the HTTP protocol, preprocesses it accordingly according to the working conditions, cleans the data, removes outliers, calculates ROP, and puts the processed data into the raw data buffer. When the data in the buffer accumulates to a certain level, it is sent to the parameter recommendation system.

[0122] 1.1. Working condition pretreatment

[0123] Working condition preprocessing is to preprocess the raw data according to the working condition of the currently received data (drilling, non-drilling, sliding drilling and compound drilling) and the switching of the previous and next working conditions (drilling switching to non-drilling, non-drilling switching to drilling). Figure 3 The schematic diagram of the working condition preprocessing process shown in the figure includes: drilling state judgment, sliding drilling state judgment and calculation of screw speed and screw torque under sliding drilling.

[0124] When a piece of data is received, first determine whether it is drilling data. If it is not drilling data, indicate to the system that the current state is non-drilling, skip the processing of the current data, and clear the data buffer and ROP calculation queue. If it is a drilling state, first determine whether it is a sliding drilling state based on the observed values ​​of the speed and torque. If it is a sliding drilling state, it is necessary to recalculate the screw speed and screw torque as the actual speed and torque of the current data to replace the original values. If it is a compound drilling state, no processing is performed.

[0125] 1.1.1. Drilling status judgment

[0126] When the received data simultaneously meets the two conditions that the absolute values ​​of the well depth and the drill bit position differ by 0.05 meters and the drilling pressure value is greater than or equal to 10kN, it indicates that the current state is drilling, otherwise it is non-drilling.

[0127] 1.1.2. Determine whether it is in sliding drilling state

[0128] If the rotary table speed of the current data is less than 5 revolutions or the torque is less than 0.8 kN·m, it is a sliding drilling state.

[0129] 1.1.3. Calculate the screw speed (RPM) and screw torque (Torque) under sliding drilling

[0130] In the sliding drilling state, the screw speed and screw torque need to be recalculated. The screw speed RPM motor The calculation formula is as follows:

[0131]

[0132] Wherein, Qpdm represents the displacement and Kn represents the speed-to-flow ratio (Qpdm / RPM).

[0133] Screw Torque motor The calculation formula is as follows:

[0134]

[0135] Where SPP is the riser pressure, CSPP is the circulating riser pressure, T max is the maximum rated torque of the screw, P max is the maximum rated pressure difference.

[0136] 1.2 Data Cleaning

[0137] In order to reduce the impact of data anomalies, the received data is cleaned and the obvious outliers are removed. The expert threshold algorithm is combined with the speed constraint to monitor the outliers. The data first enters the expert threshold cleaning algorithm. If it is an outlier, the speed constraint cleaning algorithm is skipped and directly output. Otherwise, the speed constraint cleaning algorithm is entered for further judgment. Figure 4 , which is a flowchart of data cleaning.

[0138] 1.2.1 Expert Threshold Cleaning Algorithm

[0139] By setting expert thresholds, outliers in the data can be monitored. Each expert threshold can be set as follows:

[0140] Weight on bit (WOB) range: 0~300kN;

[0141] Composite drilling torque (Torque) range: 0 ~ 50kN m;

[0142] Sliding drilling torque range: 0~30kN·m;

[0143] Compound drilling ground speed (RPM) range: 0~150r;

[0144] Sliding drilling screw speed range: 0~300r;

[0145] Inlet flow rate (Flowin) range: 0~100L / s;

[0146] Standpipe pressure (SPP) range: 0~100MPa;

[0147] ROP range: 0-60m / hr.

[0148] 1.2.2 Speed ​​Constrained Cleaning Algorithm

[0149] After the data is cleaned by the expert threshold algorithm, it will enter the speed constraint algorithm for further judgment. Figure 5 , which is a flow chart of the speed constraint algorithm. Define the data point received at the current time t as x, then the corresponding speed value Δx is: Δx=(x t -x t-k ) / k, where k is the interval between two data points. Data analysis shows that the speed value follows a Gaussian distribution. Fig.10 The left dashed line is the upper limit of the selected speed, the right dashed line is the lower limit of the selected speed, and the black solid line is the Gaussian distribution. In order to select the upper and lower limits of the speed value according to the distribution, the upper and lower limits of the speed value are given by calculating the mean and standard deviation of the data in the window and combining the n-sigma criterion. The mean of the window data (μ t ) and standard deviation (σ t ) is calculated as follows:

[0150]

[0151]

[0152] According to experience and data testing, take n = 5 and calculate the upper bound of the speed constraint Δx ub and the lower bound Δx lb :

[0153] Δx ub =μ t +n×σ t (5)

[0154] Δx lb =μ t -n×σ t (6)

[0155] The calculation window size W affects the cleaning effect. When the data volume is too small, the Gaussian distribution fitting effect is poor and the cleaning effect is not good. Therefore, it is necessary to set the window size based on the well site data situation. Preferably, W>9000 (considering that when the time step is 1s, at least 2.5H of velocity data is used for fitting).

[0156] 1.3 Calculating ROP

[0157] ROP calculation is performed according to the well depth interval, and a queue is maintained according to the well depth interval. The window length of the queue changes dynamically (the micro-element well segment dynamic buffer gives the queue window length). Each time the ROP of a new data is calculated, the data is added to the end of the queue.

[0158] Assume that in the ROP calculation queue, the well depth data can be expressed as [Depth1, ..., Depth now], time data can be expressed as [Time1, …, Time now ], then the newly added tail data ROP is:

[0159]

[0160] ΔDepth and ΔTime represent the well depth difference and time difference respectively.

[0161] 1.4. Design of dynamic buffer zone of micro-element well section

[0162] The micro-element well section dynamic buffer stores the processed data and defines the frequency at which the parameter recommendation system obtains data for calculation.

[0163] The size of the buffer is accumulated according to the depth interval and changes dynamically with the ROP. When the interval between the depth of the last data in the buffer and the depth of the first data is greater than or equal to the buffer size, all the data in the buffer is provided to the recommendation algorithm, and then the data buffer is cleared and data accumulation begins again.

[0164] If the anchor point ROP is less than or equal to 6 m / s, the buffer size is set to 0.05 m; if the anchor point ROP is greater than 6 m / s and less than or equal to 10 m / s, the buffer size is 0.08 m; if the anchor point ROP is greater than 10 m / s, the buffer size is 0.1 m. The dynamic buffer size of the micro-element well section is expressed as follows:

[0165]

[0166] The buffer stores logically continuous drilling data. The data in the data buffer needs to be cleared when the working condition switches, including switching between drilling and non-drilling, sliding drilling and compound drilling.

[0167] 2. Parameter recommendation subsystem (ROP-OPT-AI)

[0168] refer to Figure 6 , which is a flow chart of parameter recommendation by the parameter recommendation subsystem. The parameter recommendation subsystem obtains micro-element well section data from the data preprocessing subsystem, filters and merges it into one piece of data, calculates the metrics (mechanical specific energy MSE, optimization target Objective) used to characterize drilling performance based on the predefined physical model, and adds the calculation results to the predefined data structure. When the data is collected to a certain extent, the algorithm will dynamically learn the functional relationship between the drilling operation parameters (such as WOB, RPM, Qpdm) and the optimization targets Objective and ROP, and recommend better drilling operation parameters to the driller based on the functional relationship, so as to gradually improve the drilling efficiency.

[0169] 2.1 Micro-element well section data processing

[0170] After receiving the data from the data preprocessing subsystem, the time, well depth and drilling speed of the last data in the micro-element well section are taken as the time and well depth of the merged data, and the other data are averaged using Kalman filtering. The merged data segment is the input of the recommendation algorithm.

[0171] 2.2. Calculation of metrics

[0172] The mechanical specific energy (MSE) and optimization target (Objective) are calculated based on the physical model as drilling performance metrics.

[0173] Mechanical specific energy is the work done by the drill bit to break a unit volume of rock (i.e. the mechanical energy required). Mechanical specific energy provides a method to evaluate drilling efficiency. The larger the specific energy value, the lower the rock breaking efficiency of the drill bit, the poor adaptability of the drill bit to the bottom layer, and the drilling parameters need to be optimized. The model calculation formula is:

[0174]

[0175] Among them, EFF s Efficiency coefficient is a constant, usually EFF s =35%; MSE is mechanical specific energy, MPa; WOB is bit pressure, N; T is torque, Nm; RPM is speed, rev / min; ROP is mechanical drilling speed, m / hr; d B is the drill bit diameter, mm.

[0176] The calculation of mechanical specific energy (MSE) requires total speed and total torque. The calculation of total speed is divided into the following three cases:

[0177] Compound drilling screw: RPM = RPM surface +RPM motor

[0178] Compound drilling without screw: RPM = RPM surface

[0179] Sliding drilling: RPM = RPM motor =(Qpdm / Kn)×60

[0180] The total torque can be divided into the following two cases according to the working conditions:

[0181] Compound drilling: Torque = Torque surface

[0182] Sliding drilling: Torque = Torque motor

[0183] The optimization objective Objective is to maximize the mechanical drilling speed while minimizing the mechanical specific energy. The normalized quantities ROP0 and MSE0 are defined to remove the influence of the dimension. Currently, ROP0 = 60, MSE0 = 20000. The optimization objective calculation formula is as follows:

[0184]

[0185] 2.3 Mode Switching

[0186] The current system is divided into three modes: exploration, learning, and application. Mode switching includes mode switching in exploration mode and mode switching in learning and application modes.

[0187] 2.3.1. Mode switching in exploration mode

[0188] The task of the exploration mode is to guide the driller to continuously explore in the direction of increasing ROP, fully explore the parameter space to obtain as many corresponding relationships between operating parameters and ROP as possible, and find the area with the largest ROP. When the exploration is sufficient, the system switches from the exploration mode to the learning mode, and the switching condition is to determine whether the predetermined number of explorations has been completed. Considering the timeliness requirements of actual drilling operations, the system adopts a dynamic exploration number based on the drilling speed. When the drilling speed is high, more explorations are performed, and when the drilling speed is low, the number of explorations is reduced to reduce the impact on the drilling operation. Specifically, the system sets the minimum number of explorations (5 times) and the maximum number of explorations (9 times). When the minimum number of explorations is reached, the current drilling speed is judged before each exploration. If the drilling speed is fast (the judgment threshold is 6m / h), the exploration continues, otherwise it switches to the learning mode (a switching judgment is performed every time 5 micro-element well section data are received).

[0189] 2.3.2 Mode switching between learning and application modes

[0190] After the system enters the learning and application mode, it is in the stage of dynamically monitoring drilling performance. If the drilling performance is good, the current operating parameters are maintained and the system enters the application mode. If the drilling performance deteriorates slightly, the drilling parameters are fine-tuned and the system enters the learning mode to adapt to the current drilling status changes. If the drilling performance deteriorates severely, it means that formation switching may be in progress or complex downhole conditions have occurred. The current model is no longer adapted to the new formation environment and the system needs to be switched back to the exploration stage.

[0191] Using ROP score (ROP score) to judge the current drilling performance. By setting two ROP score thresholds to judge the current drilling effect, if it is lower than the minimum threshold (0.7), it means that the drilling performance has seriously declined. If it is higher than the maximum threshold (0.9), it means that the current drilling performance is stable. If it is between the two thresholds, it means that the drilling performance has declined to a small extent. The system calculates the ROP score of the anchor point and the historical anchor point every time it receives 10 micro-element well section data (ROP score ), the calculation formula is as follows:

[0192]

[0193] In order to make the judgment more robust, the “three consecutive ROP scores” judgment method is adopted. The ROP score is calculated by comparing the current ROP anchor point with the three most recent ROP anchor points in history. A voting method is used to enter the mode that falls into the interval with the highest ROP score.

[0194] The voting details are shown in Table 1. If the ROP scores of the three intervals are 1:1:1, considering that the drilling status fluctuates greatly at this time, the system maintains the application mode and waits for one round.

[0195] In actual application scenarios, the specific number of calculations can be selected and adjusted according to actual conditions.

[0196] Table 1 Three consecutive ROP scores

[0197]

[0198]

[0199] 2.4 Parameter Recommendations

[0200] 2.4.1 Parameter recommendations under different working conditions

[0201] Composite drilling: The functional relationship between drilling operation parameters and drilling performance improvement goals, and the recommended parameters are given based on this functional relationship:

[0202] ROP=f(WOB,RPM surface ,Qpdm)

[0203] OBJ=f(WOB,RPM surface ,Qpdm) (12)

[0204] Among them, ROP is the mechanical drilling speed; OBJ is the optimization target; WOB is the drilling pressure; RPM surface is the ground speed; Qpdm is the displacement; the recommended parameters are given according to the functional relationship;

[0205] Sliding drilling: The functional relationship between drilling operation parameters and drilling performance improvement goals, and the recommended parameters are given based on this functional relationship:

[0206] ROP=f(WOB,Qpdm)

[0207] OBJ=f(WOB,Qpdm) (13)

[0208] 2.4.2. Exploration Mode Parameter Recommendations

[0209] The support vector regression (SVR) model was used to fit the relationship between drilling operation parameters and ROP and OBJ. Fig.11 This is a schematic diagram of the support vector regression model.

[0210] The fitting objective of the SVR model is to maximize the interval band distance d (i.e., ) while minimizing the loss loss (i.e., ), the mathematical expression of the model is as follows:

[0211]

[0212]

[0213] Taking compound drilling (the rotation speed is not considered during sliding drilling) as an example, the SVR model is used to fit the linear relationship of formula (10). According to the slopes obtained by fitting each parameter, based on the boundary values ​​of the explored parameter interval, the recommended parameters are adjusted in the direction of increasing ROP according to the change of the slope. The recommended parameter size should not exceed the value specified in the drilling design, otherwise the drilling design will be used as the recommended parameter.

[0214] The explored parameter interval boundary values ​​refer to the maximum and minimum anchor points of WOB, RPM, and Qpdm received during the exploration phase. Taking drilling pressure as an example, if the model requires a reduction in drilling pressure, the minimum historical drilling pressure anchor point during the exploration phase will be used as the benchmark value for drilling pressure adjustment; if the model requires an increase in drilling pressure, the maximum historical drilling pressure anchor point during the exploration phase will be used as the benchmark value for current drilling pressure adjustment. Fig.12 This is a schematic diagram of adjusting (reducing) the parameters of the exploration mode. Fig.13 Schematic diagram of parameter adjustment (increase) in exploration mode.

[0215] 2.4.3 Learning mode parameter recommendation

[0216] The random forest regression model was used for nonlinear fitting to fit the relationship between drilling operation parameters and ROP and OBJ. Fig.14 This is a schematic diagram of the random forest algorithm. Fig.14The letters in the figure represent Random Forest, Instance, Tree 1, 2, 3, Value A, B, C, Average, and Final Value. The model uses the data set to train multiple decision trees that can complete the fitting task independently, resamples the data independently multiple times, and integrates the results of multiple decision trees to obtain the model output. The output results are not easily affected by noise data and outlier data, and have stronger robustness and generalization.

[0217] Taking compound drilling (sliding drilling without considering the rotation speed) as an example, the random forest regression model is used to fit the nonlinear relationship of formula (10). After fitting, a grid search is performed in the parameter space based on the model to obtain the parameters corresponding to the maximum point of OBJ (ROP) and determine whether the parameters meet the conditions. If they meet the conditions, recommendations are made.

[0218] The method for determining the interval and step size of each parameter in the grid search is as follows: calculate the maximum and minimum values ​​of WOB, RPM, and Qpdm in the historical data, the interval of WOB (unit: kN) is [minimum value -3, maximum value +3], and the exploration step size is 1; the interval of RPM (unit: r) is [minimum value -2, maximum value +2], and the exploration step size is 0.5; the interval of Qpdm (unit: L / s) is [minimum value -1, maximum value +1], and the exploration step size is 0.2. When recommending parameters, select the WOB, RPM, and Qpdm recommendations corresponding to the maximum value of OBJ (ROP) during the grid search process.

[0219] 2.4.4 Application mode parameter recommendations

[0220] When the system is in application mode, the drilling performance is relatively stable. At this time, keep the last recommended parameters unchanged to keep the system in a steady state.

[0221] The drilling parameter recommendation method based on the AI ​​model proposed in the present invention is implemented through a data preprocessing subsystem and a parameter recommendation subsystem; wherein the data preprocessing subsystem is mainly connected to the drilling equipment, collects the original data of the well site and performs corresponding preprocessing, and provides it to the parameter recommendation system for use. The parameter recommendation subsystem receives the data of the data processing subsystem, calculates various metrics for the data, builds a functional relationship between the drilling parameters and the target indicators based on AI model learning, and recommends drilling operation parameters that can improve the target indicators based on the functional relationship. The present invention applies the AI ​​model to drilling parameter recommendation, can identify different working conditions and clean the original data, adopts a solution combining linear exploration with nonlinear application, and recommends drilling parameters that can optimize the target indicators under different modes. The present invention makes up for the shortcomings of the traditional drilling parameter recommendation method, such as cumbersome process, long time consumption, and poor accuracy, and can dynamically and efficiently recommend the best drilling parameters and optimize the drilling speed, thereby achieving the purpose of reducing costs and increasing efficiency in the drilling process.

[0222] It should be noted that, although the operations of the method of the present invention are described in a specific order in the above embodiments and the accompanying drawings, this does not require or imply that the operations must be performed in the specific order, or that all the operations shown must be performed to achieve the desired results. Additionally or alternatively, some steps may be omitted, multiple steps may be combined into one step, and / or one step may be decomposed into multiple steps.

[0223] After introducing the method of the exemplary embodiment of the present invention, next, refer to Fig.15 A drilling parameter recommendation device based on an AI model according to an exemplary embodiment of the present invention is introduced.

[0224] The implementation of the drilling parameter recommendation device based on the AI ​​model can refer to the implementation of the above method, and the repeated parts will not be repeated. The term "module" or "unit" used below can be a combination of software and / or hardware that implements the predetermined function. Although the device described in the following embodiments is preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0225] Based on the same inventive concept, the present invention also proposes a drilling parameter recommendation device based on an AI model, such as Fig.15 As shown, the device comprises:

[0226] A data preprocessing subsystem, used to obtain raw data and clean the raw data;

[0227] The parameter recommendation subsystem is used to calculate the drilling performance metrics according to the physical model, with the goal of optimizing the drilling performance metrics, and to construct the functional relationship between the drilling parameters and the drilling performance metrics through dynamic learning; and to recommend the optimized drilling parameters to the driller according to the functional relationship.

[0228] In one embodiment, the data preprocessing subsystem obtains raw data and cleans the raw data, including:

[0229] Use independent threads to obtain raw data through HTTP protocol;

[0230] Data preprocessing is performed according to the working conditions, and the expert threshold method and speed constraint method are used to clean the data. The mechanical drilling speed is calculated according to the well depth interval, and the processed micro-element well section data is saved in the original data buffer.

[0231] When the data accumulates to a certain level, it is sent to the parameter recommendation subsystem.

[0232] In one embodiment, the data preprocessing subsystem performs data preprocessing according to the working conditions, including:

[0233] According to the working condition of the currently received data and the switching status of the previous and next working conditions, the original data is preprocessed accordingly;

[0234] Among them, during preprocessing, the drilling status is first determined based on the well site data; if it is the drilling status, the sliding drilling status is further determined; if it is not the drilling status, the status parameters are updated, and the data buffer area and the mechanical drilling speed calculation queue are cleared;

[0235] When judging the sliding drilling state, if it is the sliding drilling state, the screw speed and screw torque are recalculated and the state parameters are updated; if it is not the sliding drilling state, no processing is performed.

[0236] In one embodiment, the data preprocessing subsystem uses an expert threshold method and a speed constraint method to clean data, including:

[0237] The abnormal values ​​in the monitoring data are monitored by setting expert thresholds, wherein the expert thresholds include at least: drilling pressure range, composite drilling torque range, sliding drilling torque range, composite drilling ground speed range, sliding drilling screw speed range, inlet flow range, riser pressure range, and drilling speed range;

[0238] The expert threshold method is used to clean outliers, and the retained data is further judged using the speed constraint method to clean up data that does not meet the upper and lower bounds of the speed value.

[0239] In one embodiment, the data preprocessing subsystem calculates the mechanical drilling rate according to the well depth interval, including:

[0240] The mechanical drilling speed is determined according to the well depth interval, and a queue is maintained according to the well depth interval. The window length of the queue changes dynamically. Each time the mechanical drilling speed of a new data is calculated, the calculated data is added to the end of the queue. The calculation method of the mechanical drilling speed is:

[0241]

[0242] Among them, ROP now It is the tail data of the newly added team; Depth1, Depth now They are the first and last data in the well depth data queue respectively; Time1, Time now They are respectively the first data and the last data in the time data queue.

[0243] In one embodiment, the data preprocessing subsystem saves the processed micro-element well section data to the original data buffer, including:

[0244] The processed data is saved in the micro-element well section dynamic buffer; when the interval between the well depth of the last data in the micro-element well section dynamic buffer and the well depth of the first data in the micro-element well section dynamic buffer is greater than or equal to the buffer size, all the data in the micro-element well section dynamic buffer are sent to the parameter recommendation subsystem, and after the sending is completed, the micro-element well section dynamic buffer is cleared and the data is accumulated again.

[0245] In one embodiment, the parameter recommendation subsystem is used to calculate the drilling performance metric index according to the physical model, with the goal of optimizing the drilling performance metric index, construct the functional relationship between the drilling parameters and the drilling performance metric index through dynamic learning; and recommend the optimized drilling parameters to the driller according to the functional relationship, including:

[0246] Calculating drilling performance metrics based on the cleaned data, wherein the drilling performance metrics at least include: mechanical specific energy and optimization target;

[0247] The goal is to improve the drilling performance metrics, and a solution combining linear exploration and nonlinear application is used to recommend drilling parameters. After the parameter recommendation subsystem is started, it enters the exploration mode, and based on the trend of linear model fitting, it explores and collects data in the direction of increasing the mechanical drilling rate. After sufficient exploration, it enters the learning mode, and learns the functional relationship between the mechanical drilling rate and drilling parameters in the current exploration data based on the model to determine the optimal drilling parameter combination. The optimal drilling parameters are maintained and the application mode is entered. The current drilling status is monitored while breaking the rock, and the learning mode or exploration mode is automatically switched to according to the degree of mechanical drilling rate performance degradation.

[0248] In one embodiment, the parameter recommendation subsystem is further used to:

[0249] The time and well depth of the last data in the micro-element well section data are taken out as the time and well depth of the merged data, and the other data are averaged by Kalman filtering to obtain the merged data, which is then input into the parameter recommendation subsystem.

[0250] In one embodiment, the parameter recommendation subsystem calculates the drilling performance metric based on the cleaned data, including:

[0251] The calculation formula of mechanical specific energy is:

[0252]

[0253] Wherein, MSE is mechanical specific energy; EFF is s is the efficiency coefficient; WOB is the bit pressure; T is the torque; RPM is the speed; ROP is the mechanical drilling speed; d B is the drill bit diameter; mechanical specific energy is the work done by the drill bit to break a unit volume of rock, which is used to evaluate drilling efficiency. The larger the mechanical specific energy value, the lower the rock breaking efficiency of the drill bit, the poor adaptability of the drill bit to the bottom layer, and the drilling parameters need to be optimized;

[0254] The calculation formula of the optimization target is as follows:

[0255]

[0256] Among them, Objective is the optimization target, and the optimization target is to maximize the mechanical drilling speed while minimizing the mechanical specific energy; ROP0 and MSE0 are normalized quantities used to remove the dimension.

[0257] In one embodiment, the parameter recommendation subsystem, the modes of the parameter recommendation subsystem include: exploration mode, learning mode and application mode.

[0258] In one embodiment, the parameter recommendation subsystem, the mode switching method of the parameter recommendation subsystem is:

[0259] Determine whether the parameter recommendation subsystem is in exploration mode; if in exploration mode, further determine whether the exploration is sufficient; if not in exploration mode, evaluate drilling performance;

[0260] When judging whether the exploration is sufficient, if the exploration is sufficient, switch to the learning mode and use the nonlinear model to learn the optimal parameters; if the exploration is insufficient, switch to the exploration mode and use the linear model to explore the space;

[0261] When evaluating drilling performance metrics, determine whether the drilling performance has declined; if not, switch to application mode and maintain the current parameters for rock breaking; if decline occurs, determine whether the decline indicator reaches the decline threshold;

[0262] When judging whether the recession indicator has reached the recession threshold, if it has not reached it, it switches to the learning mode and uses the nonlinear model to learn the optimal parameters; if it has reached it, it switches to the exploration mode and performs spatial exploration through the linear model.

[0263] In one embodiment, the parameter recommendation subsystem, in the exploration mode, a mode switching method includes:

[0264] The exploration mode of the parameter recommendation subsystem is used to guide the driller to explore in the direction of increasing the mechanical drilling speed, obtain the corresponding relationship between the operating parameters and the mechanical drilling speed by exploring the parameter space, and find the area with the maximum mechanical drilling speed;

[0265] When the exploration is sufficient, the mode switches from exploration mode to learning mode. The switching condition is to determine whether the predetermined number of explorations has been completed. Among them, a minimum number of explorations and a maximum number of explorations are set. When the minimum number of explorations is reached, the current drilling speed is determined before each exploration. If the drilling speed exceeds the drilling speed threshold, the exploration continues, otherwise it switches to learning mode. When the maximum number of explorations is exceeded, the exploration ends.

[0266] In one embodiment, the parameter recommendation subsystem, a mode switching method in a learning mode or an application mode includes:

[0267] After the parameter recommendation subsystem switches to the learning mode or the application mode, it is in the stage of dynamically monitoring the drilling performance; if the drilling performance reaches the performance threshold, the current operating parameters are maintained to enter the application mode; if the drilling performance declines and does not reach the decline threshold, the drilling parameters are adjusted to enter the learning mode; if the drilling performance declines and reaches the decline threshold, it switches to the exploration stage;

[0268] Among them, the mechanical drilling speed score is used to judge the current drilling performance, and the current drilling effect is judged by setting two mechanical drilling speed score thresholds. If the mechanical drilling speed score falls into an interval below the first threshold, it is determined that the drilling performance has seriously declined. If the mechanical drilling speed score falls into an interval above the second threshold, it is determined that the current drilling performance is stable. If the mechanical drilling speed score falls into an interval between the first threshold and the second threshold, it is determined that the drilling performance has slightly declined. The calculation method of the mechanical drilling speed score is:

[0269]

[0270] ROP score Score for rate of penetration; ROP now Current mechanical drilling speed; ROP history is the historical mechanical drilling speed; the parameter recommendation subsystem calculates the mechanical drilling speed scores of the current anchor point and the historical anchor point every time it receives 10 micro-element well section data.

[0271] In one embodiment, the parameter recommendation subsystem, the method further includes:

[0272] Calculate multiple consecutive ROP scores;

[0273] According to the mechanical drilling speed score threshold, multiple mechanical drilling speed scores are divided into different intervals, and an interval with the largest number of divided mechanical drilling speed scores is selected as the current determination result.

[0274] In one embodiment, the parameter recommendation subsystem recommends optimized drilling parameters to the driller according to the functional relationship, including:

[0275] When making parameter recommendations under different working conditions, for composite drilling, the functional relationship between drilling operation parameters and drilling performance improvement targets is:

[0276] ROP=f(WOB,RPM surface ,Qpdm)

[0277] OBJ=f(WOB,RPM surface ,Qpdm)

[0278] ROP is the mechanical drilling speed; OBJ is the optimization target; WOB is the bit pressure; RPM surface is the ground speed; Qpdm is the displacement; the recommended parameters are given according to the functional relationship;

[0279] For sliding drilling, the functional relationship between drilling operation parameters and drilling performance improvement targets is:

[0280] ROP=f(WOB,Qpdm)

[0281] OBJ=f(WOB,Qpdm)

[0282] Recommended parameters are given according to the functional relationship.

[0283] In one embodiment, the parameter recommendation subsystem recommends optimized drilling parameters to the driller according to the functional relationship, including:

[0284] When making parameter recommendations in the exploration mode, the support vector regression model is used to fit the relationship between drilling operation parameters, mechanical drilling speed and optimization objectives, and the recommended parameters are determined based on the fitted relationship;

[0285] Among them, for compound drilling or sliding drilling, the support vector regression model is used to fit the linear relationship of the optimization target calculation method. According to the slope obtained by parameter fitting and based on the boundary value of the explored parameter interval, the recommended parameters are adjusted in the direction of increasing the mechanical drilling speed according to the change of the slope. If the adjusted recommended parameters exceed the values ​​specified in the drilling design, the drilling design will be used as the recommended parameters.

[0286] In one embodiment, the parameter recommendation subsystem recommends optimized drilling parameters to the driller according to the functional relationship, including:

[0287] When making parameter recommendations in the learning mode, a random forest regression model is used for nonlinear fitting to fit the relationship between drilling operation parameters, mechanical drilling speed and optimization objectives, and the recommended parameters are determined based on the fitted relationship;

[0288] Among them, for compound drilling or sliding drilling, the random forest regression model is used to fit the linear relationship of the optimization target calculation method. After fitting, a grid search is performed in the parameter space based on the random forest regression model to obtain the parameters corresponding to the maximum point of the mechanical drilling speed or optimization target, and to determine whether the parameters meet the conditions. If they meet the conditions, recommendations are made.

[0289] In one embodiment, the parameter recommendation subsystem recommends optimized drilling parameters to the driller according to the functional relationship, including:

[0290] When making parameter recommendations in application mode, the last recommended parameters are used.

[0291] It should be noted that although several modules of the drilling parameter recommendation device based on the AI ​​model are mentioned in the above detailed description, this division is only exemplary and not mandatory. In fact, according to an embodiment of the present invention, the features and functions of two or more modules described above can be embodied in one module. Conversely, the features and functions of one module described above can be further divided into multiple modules for embodiment.

[0292] Based on the above invention concept, Fig.16 As shown, the present invention also proposes a computer device 1600, including a memory 1610, a processor 1620, and a computer program 1630 stored in the memory 1610 and executable on the processor 1620, wherein the processor 1620 implements the aforementioned AI model-based drilling parameter recommendation method when executing the computer program 1630.

[0293] Based on the aforementioned inventive concept, the present invention proposes a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the aforementioned AI model-based drilling parameter recommendation method is implemented.

[0294] Based on the aforementioned inventive concept, the present invention proposes a computer program product, which includes a computer program, and when the computer program is executed by a processor, a drilling parameter recommendation method based on an AI model is implemented.

[0295] The present invention discloses a drilling parameter recommendation method and device based on an AI model, which collects raw data from a well site and performs corresponding preprocessing, and provides the data to a parameter recommendation system for use. The parameter recommendation subsystem receives data from a data processing subsystem, calculates various metrics on the data, builds a functional relationship between drilling parameters and target indicators based on AI model learning, and recommends drilling operation parameters that can improve the target indicators based on the functional relationship. The present invention applies an AI model to drilling parameter recommendation, which can identify different working conditions and clean the raw data. It adopts a solution that combines linear exploration with nonlinear application to recommend drilling parameters that can optimize the target indicators under different modes. The present invention makes up for the shortcomings of traditional drilling parameter recommendation methods, such as cumbersome processes, long time consumption, and poor accuracy. It can dynamically and efficiently recommend optimal drilling parameters and optimize drilling speed, thereby achieving the purpose of reducing costs and increasing efficiency in the drilling process.

[0296] The acquisition, storage, use, and processing of data in the technical solution of this application comply with the relevant provisions of laws and regulations.

[0297] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, devices, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0298] The present invention is described with reference to flowcharts and / or block diagrams of methods and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0299] These computer program instructions may also be stored in a computer readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0300] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0301] Finally, it should be noted that the above-described embodiments are only specific implementations of the present invention, which are used to illustrate the technical solutions of the present invention, rather than to limit them. The protection scope of the present invention is not limited thereto. Although the present invention is described in detail with reference to the above-described embodiments, ordinary technicians in the field should understand that any technician familiar with the technical field can still modify the technical solutions recorded in the above-described embodiments within the technical scope disclosed by the present invention, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. A drilling parameter recommendation method based on an AI model, characterized in that: include: Acquiring raw data and cleaning the raw data; The cleaned data is input into the parameter recommendation subsystem, which calculates drilling performance metrics based on the physical model. With the goal of optimizing the drilling performance metrics, the functional relationship between drilling parameters and drilling performance metrics is constructed through dynamic learning. The optimized drilling parameters are recommended to the driller based on the functional relationship.

2. The method according to claim 1, characterized in that Acquiring raw data and cleaning the raw data includes: Through the data preprocessing subsystem, an independent thread is used to obtain raw data through the HTTP protocol; Data preprocessing is performed based on the working conditions. The expert threshold method and speed constraint method are used to clean the data. The mechanical drilling rate is calculated according to the well depth interval. The processed micro-element well section data is saved in the original data buffer. When the data accumulates to a certain level, it is sent to the parameter recommendation subsystem.

3. The method according to claim 2, characterized in that Data preprocessing is performed according to the working conditions, including: Pre-process the raw data accordingly based on the current working condition of the received data and the switching status of the previous and next working conditions; Among them, during preprocessing, the drilling status is first determined based on the well site data; if it is in the drilling state, the sliding drilling state is further determined; if it is not in the drilling state, the state parameters are updated, and the data buffer area and mechanical drilling speed calculation queue are cleared; When judging the sliding drilling state, if it is the sliding drilling state, the screw speed and screw torque are recalculated and the state parameters are updated; if it is not the sliding drilling state, no processing is performed.

4. The method according to claim 2, characterized in that Use expert thresholding and speed constraint methods to clean data, including: Abnormal values ​​in the monitoring data are detected by setting expert thresholds, wherein the expert thresholds include at least: a weight on bit range, a composite drilling torque range, a sliding drilling torque range, a composite drilling surface speed range, a sliding drilling screw speed range, an inlet flow range, a riser pressure range, and a drilling speed range; The expert threshold method is used to clean outliers, and the retained data is further judged using the speed constraint method to clean data that does not meet the upper and lower bounds of the speed value.

5. The method according to claim 2, characterized in that Calculate ROP by well depth interval, including: The ROP is determined based on the well depth interval, and a queue is maintained based on the well depth interval. The window length of the queue changes dynamically. Each time the ROP of a new data point is calculated, the calculated data is added to the end of the queue. The ROP is calculated as follows: Among them, ROP now The newly added tail data; Depth1, Depth now They are the first and last data in the well depth data queue respectively; Time1, Time now They are respectively the first data and the last data in the time data queue.

6. The method according to claim 2, characterized in that Save the processed micro-element well section data to the original data buffer, including: The processed data is saved in the micro-element well section dynamic buffer. When the interval between the well depth of the last data and the well depth of the first data in the micro-element well section dynamic buffer is greater than or equal to the buffer size, all the data in the micro-element well section dynamic buffer is sent to the parameter recommendation subsystem. After the sending is completed, the micro-element well section dynamic buffer is cleared and the data is accumulated again.

7. The method according to claim 1, characterized in that The cleaned data is input into the parameter recommendation subsystem, which calculates drilling performance metrics based on the physical model. With the goal of optimizing the drilling performance metrics, the functional relationship between drilling parameters and drilling performance metrics is constructed through dynamic learning, including: Calculating drilling performance metrics based on the cleaned data, wherein the drilling performance metrics include at least: mechanical specific energy and optimization target; With the goal of improving drilling performance metrics, a scheme combining linear exploration and nonlinear application is used to recommend drilling parameters. After the parameter recommendation subsystem is started, it enters exploration mode. Based on the trend of linear model fitting, it explores and collects data in the direction of increasing mechanical penetration rate. After sufficient exploration, it enters learning mode. Based on the model, it learns the functional relationship between mechanical penetration rate and drilling parameters in the current exploration data to determine the optimal drilling parameter combination. Maintaining the optimal drilling parameters, it enters application mode, monitors the current drilling status while breaking rock, and automatically switches to learning mode or exploration mode based on the degree of mechanical penetration rate performance degradation.

8. The method according to claim 7, characterized in that The method further includes: The time and well depth of the last data in the micro-element well section data are taken as the time and well depth of the merged data, and the other data are averaged using Kalman filtering to obtain the merged data, which is then input into the parameter recommendation subsystem.

9. The method according to claim 7, characterized in that Calculate drilling performance metrics based on cleaned data, including: The calculation formula of mechanical specific energy is: Wherein, MSE is mechanical specific energy; EFF is s is the efficiency coefficient; WOB is the bit weight; T is the torque; RPM is the speed; ROP is the mechanical drilling speed; d B is the drill bit diameter; mechanical specific energy is the work done by the drill bit to break a unit volume of rock and is used to evaluate drilling efficiency. A larger mechanical specific energy value indicates lower rock breaking efficiency, poor adaptability of the drill bit to the bottom layer, and the drilling parameters need to be optimized; The calculation formula of the optimization objective is as follows: Among them, Objective is the optimization objective, which is to maximize the mechanical drilling rate while minimizing the mechanical specific energy; ROP0 and MSE0 are normalized quantities used to remove dimensions.

10. The method according to claim 7, characterized in that The modes of the parameter recommendation subsystem include: exploration mode, learning mode and application mode.

11. The method according to claim 10, characterized in that The mode switching method of the parameter recommendation subsystem is: determining whether the parameter recommendation subsystem is in an exploration mode; if in the exploration mode, further determining whether the exploration is sufficient; if not in the exploration mode, evaluating the drilling performance; When judging whether the exploration is sufficient, if the exploration is sufficient, switch to the learning mode and use the nonlinear model to learn the optimal parameters; If the exploration is not sufficient, switch to exploration mode and perform spatial exploration through linear models; When evaluating drilling performance metrics, determine whether drilling performance has declined; If no recession occurs, switch to application mode and keep the current parameters for rock breaking; If a recession occurs, whether the indicators for determining the recession have reached the recession threshold; When judging whether the recession indicator reaches the recession threshold, if it does not reach it, it switches to the learning mode and uses the nonlinear model to learn the optimal parameters; if it reaches it, it switches to the exploration mode and performs spatial exploration through the linear model.

12. The method according to claim 11, characterized in that Mode switching methods in exploration mode include: The exploration mode of the parameter recommendation subsystem is used to guide the driller to explore in the direction of increasing the mechanical penetration rate, obtain the corresponding relationship between the operating parameters and the mechanical penetration rate by exploring the parameter space, and find the area with the maximum mechanical penetration rate; When the exploration is sufficient, the exploration mode switches to the learning mode. The switching condition is to determine whether the predetermined number of explorations has been completed. Among them, the minimum and maximum exploration times are set. When the minimum number of explorations is reached, the current drilling speed is judged before each exploration. If the drilling speed exceeds the drilling speed threshold, the exploration continues, otherwise it switches to the learning mode. When the maximum number of explorations is exceeded, the exploration ends.

13. The method according to claim 11, characterized in that Mode switching methods in learning mode or application mode include: After the parameter recommendation subsystem switches to the learning mode or the application mode, it enters the stage of dynamically monitoring drilling performance; if the drilling performance reaches the performance threshold, the current operating parameters are maintained and the application mode is entered; if the drilling performance deteriorates but does not reach the degradation threshold, the drilling parameters are adjusted and the learning mode is entered; if the drilling performance deteriorates and reaches the degradation threshold, the exploration mode is switched; The ROP score is used to judge the current drilling performance. Two ROP score thresholds are set to evaluate the current drilling effect. If the ROP score falls below the first threshold, it is determined that the drilling performance has severely declined. If the ROP score falls above the second threshold, it is determined that the current drilling performance is stable. If the ROP score falls between the first and second thresholds, it is determined that the drilling performance has slightly declined. The ROP score is calculated as follows: ROP score Score for rate of penetration; ROP now ROP is the current mechanical drilling speed; history is the historical mechanical drilling speed; the parameter recommendation subsystem calculates the mechanical drilling speed scores of the current anchor point and the historical anchor points every time it receives 10 micro-element well section data.

14. The method according to claim 13, characterized in that The method further includes: Calculate multiple consecutive ROP scores; According to the mechanical drilling speed score threshold, multiple mechanical drilling speed scores are divided into different intervals, and the interval with the largest number of divided mechanical drilling speed scores is selected as the current determination result.

15. The method according to claim 1, wherein Recommending optimized drilling parameters to the driller based on the functional relationship includes: When recommending parameters under different working conditions, for composite drilling, the functional relationship between drilling operation parameters and drilling performance improvement targets is as follows: ROP=f(WOB,RPM surface ,Qpdm) OBJ=f(WOB,RPM surface ,Qpdm) ROP is the mechanical drilling speed; OBJ is the optimization target; WOB is the bit weight; RPM surface is the ground speed; Qpdm is the displacement; the recommended parameters are given according to the functional relationship; For sliding drilling, the functional relationship between drilling operation parameters and drilling performance improvement targets is: ROP=f(WOB,Qpdm) OBJ=f(WOB,Qpdm) Recommended parameters are given according to the functional relationship.

16. The method according to claim 1, wherein Recommending optimized drilling parameters to the driller based on the functional relationship includes: When performing parameter recommendations in the exploration mode, a support vector regression model is used to fit the relationship between drilling operation parameters, mechanical penetration rate, and optimization objectives, and the recommended parameters are determined based on the fitted relationship. Among them, for composite drilling or sliding drilling, the support vector regression model is used to fit the linear relationship of the optimization target calculation method. According to the slope obtained by parameter fitting and the boundary values ​​of the explored parameter interval, the recommended parameters are adjusted in the direction of increasing the mechanical penetration rate according to the change of the slope. If the adjusted recommended parameters exceed the values ​​specified in the drilling design, the drilling design will be used as the recommended parameters.

17. The method according to claim 1, wherein Recommending optimized drilling parameters to the driller based on the functional relationship includes: When recommending parameters in the learning mode, a random forest regression model is used for nonlinear fitting to fit the relationship between drilling operation parameters, mechanical drilling speed, and optimization objectives, and the recommended parameters are determined based on the fitted relationship. Among them, for compound drilling or sliding drilling, the random forest regression model is used to fit the linear relationship of the optimization target calculation method. After fitting, a grid search is performed in the parameter space based on the random forest regression model to obtain the parameters corresponding to the maximum point of the mechanical drilling rate or optimization target, and to determine whether the parameters meet the conditions. If they meet the conditions, recommendations are made.

18. The method according to claim 1, wherein Recommending optimized drilling parameters to the driller based on the functional relationship includes: When making parameter recommendations in application mode, the previously recommended parameters are used.

19. A drilling parameter recommendation device based on an AI model, characterized in that: include: A data preprocessing subsystem is used to obtain raw data and clean the raw data; The parameter recommendation subsystem is used to calculate drilling performance metrics based on the physical model, with the goal of optimizing the drilling performance metrics. It constructs the functional relationship between drilling parameters and drilling performance metrics through dynamic learning. The optimized drilling parameters are recommended to the driller based on the functional relationship.

20. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the method according to any one of claims 1 to 18 is implemented.

21. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 18 is implemented.

22. A computer program product, characterized in that The computer program product comprises a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 18 is implemented.

Images