Real-time identification system and method for stick-slip working condition of horizontal well drilling bit

Abstract

The invention belongs to the technical field of horizontal well drilling optimization, and discloses a real-time recognition system and method for the stick-slip working condition of a horizontal well drilling bit. Multi-source data such as bit pressure, rotating speed, torque and mechanical drilling speed are obtained based on the while-drilling logging technology; constructing a method for acquiring effective torque of a drill bit in combination with a test while drilling to obtain effective rock breaking torque at the drill bit under the conditions of different moments and well depths; and constructing a logging data normalization model, performing normalization processing on the multi-source logging data, and obtaining a change rule of the real rock breaking energy utilization rate of the drill bit along with time and well depth. According to the real-time automatic identification method for the stick-slip working condition of the horizontal well drilling bit, manual intervention is applied in time, the stick-slip working condition is prevented from developing into early-stage damage of the bit, and the method is of great significance in saving non-production time, prolonging the service life of the bit and a trip drilling ruler and guaranteeing drilling safety and production benefits.

Description

Technical Field

[0001] This invention belongs to the field of horizontal well drilling optimization technology, and particularly relates to a real-time identification system and method for stick-slip conditions of horizontal well drilling bits. Background Technology

[0002] Horizontal well drilling significantly increases the wellbore-reservoir contact area to improve recovery rates, making it an essential technology for non-oil and gas resource development. Taking horizontal well drilling in deep coal-rock gas reservoirs as an example, after the PDC drill bit enters the coal-rock reservoir, it can quickly shear the rock to break it, and the actual drill bit rotation speed and rock-breaking torque are relatively stable. However, due to the development of interbedded rock in the coal-rock reservoir, and the fact that the strength of the interbedded rock is significantly higher than that of the coal-rock and it is prone to rockfall, after the drill bit enters the interbedded rock under the action of drilling pressure, the drill bit torque needs to accumulate for a certain period of time (the torque growth rate is slow) before it is sufficient to shear the interbedded rock, and then the drill bit torque is released rapidly. During this process, the drill bit rotation speed and rock-breaking speed fluctuate, and the fluctuation frequency is significantly reduced, which means that the unfavorable working condition of drill bit stick-slip occurs. Stick-slip condition is the main cause of early damage to PDC drill bits in horizontal well drilling, which can easily lead to unexpected tripping and bit replacement, seriously affecting the drilling time and construction cost of horizontal wells.

[0003] If drill bit stick-slip occurs, early damage to the drill bit can be completely avoided if the drilling pressure and rotational speed are reduced in time. Currently, drilling operators mainly analyze the fluctuations in mechanical drilling speed and surface torque during drilling, observe the appearance of the drill bit after tripping, and conduct post-drilling analysis based on their work experience to assess the downhole conditions of the previous drilling run in order to guide subsequent operations. In recent years, some researchers have also tried to identify drill bit stick-slip by installing various vibration sensors on the drill bit. In response to actual technical needs, the current solution has three main problems: (1) poor timeliness of post-drilling analysis, making it impossible to identify downhole stick-slip conditions in real time and adjust accordingly; (2) Drill bit stick-slip will lead to a decrease in mechanical drilling speed and torque fluctuations, but it is not the only cause. For example, the collapse of the rock block and the jamming of the centralizer will also lead to a decrease in drilling speed and torque fluctuations. Therefore, the method of identifying stick-slip by analyzing mechanical drilling speed and torque fluctuations during drilling relies too much on the subjective experience of engineers, has low reliability, and cannot meet actual needs; (3) Stick-slip conditions can be identified more accurately through drill bit vibration sensors, but this requires increasing the cost of drill bits. Moreover, it is difficult for the surface and downhole to communicate in real time during drilling. In most cases, the vibration data measured must be analyzed and applied after the drill is pulled out. Unless additional real-time coding and downhole-surface communication systems and equipment are added, the cost is high and it is far from being suitable for widespread application. Therefore, horizontal well drilling urgently needs a quantitative prediction method that can identify drill bit stick-slip in real time and accurately under the conditions of technological technology.

[0004] Logging while drilling is a standard feature in almost all drilling operations. Through logging while drilling, the operator can obtain real-time data on drilling parameters (pressure on drill, torque, displacement, rotation speed, etc.) and mechanical drilling speed. Furthermore, logging data from adjacent wells are archived and a large database is built. The logging information includes bottom hole drilling condition information, but the current logging database needs further exploration in supporting drilling optimization. Summary of the Invention

[0005] To overcome the problems existing in related technologies, the present invention discloses an embodiment of a real-time identification system and method for stick-slip conditions of horizontal well drilling bits, specifically involving a real-time identification model and method for stick-slip conditions of horizontal well drilling bits based on logging data.

[0006] The technical solution is as follows: A method for real-time identification of stick-slip conditions in horizontal well drilling bits, comprising the following steps: S1, based on logging-while-drilling technology, acquires multi-source data on drill pressure, rotational speed, torque, and mechanical drilling speed, and constructs a method for obtaining the effective rock-breaking torque of the drill bit combined with logging-while-drilling testing, reading the torque value recorded during normal rotary drilling. The torque recorded when the drill string is lifted off the bottom of the well and kept running at the original rotation speed. The effective rock-breaking torque at the drill bit under different drilling times and well depths was obtained. S2. Construct a normalized model for logging data. Input the effective rock-breaking torque of the drill bit under different time and well depth conditions and logging data into the normalized model to calculate the true rock-breaking energy utilization rate. The variation pattern with time and well depth; S3, Analysis of the actual rock-breaking energy utilization rate at the corresponding time and well depth. In addition to multi-source logging data, the dataset was divided into training and test sets in a 7:3 ratio. The data was normalized using prior physical knowledge, and a backpropagation (BP) neural network was constructed for parameter optimization, training, and validation. This led to the development of a real-time identification model for horizontal well bit stick-slip conditions based on logging data. The threshold values ​​of characteristic parameters such as amplitude, frequency, and time-domain rate of change are used to determine whether a stick-slip condition is present, enabling real-time identification and control of stick-slip conditions in horizontal well drill bits.

[0007] In step S1, acquiring multi-source data on drilling pressure, rotational speed, torque, and mechanical drilling speed includes: S101, Read the torque during normal rotary drilling and , Equal to drill string rotation torque and effective rock-breaking torque sum; S102. Raise the drill string, the drill bit leaves the bottom of the well, the drilling pressure drops to 0 but the rotational speed remains unchanged, and read the torque recorded by the logging system at this time. ; With step S101 The values ​​are the same; S103. Lower the drill bit back to the bottom of the well, increase the drilling pressure to the normal value, and set the effective rock-breaking torque at the drill bit. It is obtained by calculation using the following formula; ; In the formula, To achieve effective rock-breaking torque, For torque and, The torque recorded by the logging system; Torque recorded by the logging system The progress varies with well depth, but the daily footage in the horizontal section is limited, and short-run cleaning is required to obtain data at multiple well depths. And then obtain through linear interpolation The functional relationship between the well depth and the depth is expressed in the above formula. It is a function of well depth; the obtained By incorporating a normalized model of logging data that incorporates prior physical knowledge, the rock-breaking efficiency of the drill bit at the corresponding time and well depth can be obtained. .

[0008] In step S2, the normalization model for logging data is constructed as follows: ; In the formula, This represents the actual rock-breaking energy utilization rate of the drill bit. Drilling pressure, unit: ; The cross-sectional area of ​​the wellbore is expressed in units of... ; Rotary / top drive speed, unit: ; For the output speed of downhole power drilling tools, the manufacturer will provide the relationship between its output speed and displacement, in units of... ; The real-time torque recorded by the logging system during normal drilling, in units of... ; The average torque recorded by the logging system during normal circulation and rotation of the drill bit as it is lifted from the bottom of the well, in units of... ; This refers to the mechanical drilling speed, measured in units of... ; This represents the actual rotational speed of the drill bit. This represents the actual rock-breaking torque of the drill bit.

[0009] In step S3, constructing a real-time identification model for stick-slip conditions of horizontal well drill bits based on logging data includes: S301. The actual rock-breaking energy utilization rate of the drill bit at the corresponding time and well depth is transmitted to the first input layer of the neural network, with each input variable corresponding to an input neuron. S302. The actual rock-breaking energy utilization rate of the drill bit at the corresponding time and well depth is weighted and nonlinearly transformed in the input layer and then transmitted to the next hidden layer or output layer of the network. Repeat the above steps until the data is transmitted to the last output layer of the network and the output result of the network is calculated. S303. Compare the output results with the actual labels, and use MAE and R2 to calculate the average relative error and goodness of fit between the output value and the true value; divide the dataset into training set and test set in a 7:3 ratio, and combine physical prior knowledge to construct a real-time identification model for horizontal well bit stick-slip conditions based on logging data according to the neural network modeling steps.

[0010] Furthermore, after constructing a real-time identification model for stick-slip conditions of horizontal well drill bits based on logging data, if the output value of the model is 2, it indicates a stick-slip condition. When the deviation between the model's predicted value and the actual value exceeds a set threshold, it is necessary to compare the characteristic parameters of the amplitude and frequency of the actual rock-breaking energy utilization rate of the drill bit with the changes in time and well depth with the threshold to determine whether the condition is stick-slip, and further verify the reliability of the real-time identification model for stick-slip conditions of horizontal well drill bits based on logging data.

[0011] In step S3, constructing a real-time identification model for stick-slip conditions of horizontal well drill bits based on logging data specifically includes the following steps: (1) Record the true value of the well section under normal working conditions as 1, the true value of the well section where stick-slip occurs as 2, set the number of hidden layer nodes to 4, select the trainscg algorithm for training, select tansing and purelin for excitation functions, and import multiple parameters such as mechanical drilling speed, torque and drilling pressure as input vectors into the BP neural network for training and testing. (2) Record the true value of the well section under normal working conditions as 1, and the true value of the well section where stick-slip condition occurs as 2. Set the number of hidden layer nodes to 4, select the trainscg algorithm for training, select tansing and purelin for excitation functions, and select multiple parameters such as mechanical drilling speed, torque and drilling pressure as input vectors for training and testing. (3) Compare steps (1) and (2), using the MSEb single-item parameter as the input vector; (4) Select the optimal number of nodes by changing the number of hidden layer nodes; use the Trainlm / LMs algorithm, select the maximum number of iteration steps as 1000, and input MSE to train the neural network with a hidden layer node count of 4-20; (5) Fix the maximum number of iterations to 1000, select Trianscg as the optimal algorithm, and optimize the number of hidden layer nodes; (6) Under the optimal algorithm and the optimal number of nodes, only the maximum number of iterations is changed for testing. The maximum number of iterations is selected as 100-1500 for training the BP artificial neural network.

[0012] In step S3, the actual rock-breaking energy utilization rate at the corresponding time and well depth of the drill bit and the real-time identification model of stick-slip condition of horizontal well drill bit based on logging data are embedded into the logging system to identify stick-slip condition in real time during drilling and provide prompts.

[0013] Another objective of this invention is to provide a real-time identification system for stick-slip conditions of horizontal well drilling bits. This system implements the aforementioned real-time identification method for stick-slip conditions of horizontal well drilling bits. The system includes: (1) Calculation module for effective rock-breaking torque of drill bit. A method for obtaining effective drill bit torque combined with drilling test is constructed to obtain the effective rock-breaking torque of drill bit at different times and well depths. Then, the rock-breaking torque of drill bit is fitted as a function of well depth to predict the rock-breaking torque under subsequent drilling conditions in real time. The fitting function of drill bit rock-breaking torque and well depth is continuously iterated and corrected in combination with subsequent drilling test results. The effective rock-breaking torque of drill bit at different times and well depths under normal drilling pressure is brought into the logging data normalization model combined with physical prior knowledge to obtain multiple logging data of drill bit stick-slip during drilling test, including the actual rock-breaking energy utilization rate of drill bit at the corresponding time and well depth. (2) Module for calculating the actual rock-breaking energy utilization rate at the drill bit. A normalization model for logging data is constructed, and the multi-source logging data of effective rock-breaking torque, top drive speed, displacement, and drilling pressure of the drill bit are normalized to obtain the actual rock-breaking energy utilization rate of the drill bit; (3) A real-time identification model and method for stick-slip condition of horizontal well drill bit based on logging data. Combined with prior data of well history, the evolution data of drill bit rock breaking energy utilization rate with time and well depth under normal drilling and stick-slip conditions are obtained respectively. Then, according to the training and testing neural network modeling steps, a real-time identification model for stick-slip condition of horizontal well drill bit based on logging data is constructed, and then the real-time identification and control of stick-slip condition of horizontal well drill bit is carried out.

[0014] Furthermore, the system is mounted on a computer device, which includes at least one processor, a memory, and a computer program stored in the memory and running on the at least one processor, wherein the processor executes the computer program to perform the above-mentioned functions.

[0015] Furthermore, the system is mounted on a computer-readable storage medium that stores a computer program, which, when executed by a processor, can perform the aforementioned functions.

[0016] Combining all the above technical solutions, the beneficial effects of this invention are as follows: First, this invention constructs a real-time automatic identification method for stick-slip conditions of horizontal well drilling bits based on well logging databases and real-time logging data, thereby enabling timely manual intervention to prevent stick-slip conditions from developing into premature drill bit damage. This is of great significance for saving non-productive time and ensuring drilling safety and production efficiency.

[0017] Secondly, the present invention provides a real-time identification model and method for stick-slip conditions of horizontal well drilling bits based on logging data. This technology proposes a method for obtaining drill bit torque, constructs a calculation model for drill bit rock-breaking energy utilization efficiency, realizes the normalization processing of multi-source logging data, and establishes a stick-slip condition identification method based on neural network methods. It can identify stick-slip conditions in real time during actual drilling, apply timely human intervention, avoid early damage to the drill bit, and improve drilling safety and efficiency. Attached Figure Description

[0018] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure; Figure 1 This is a flowchart of a method for real-time identification of stick-slip conditions in horizontal well drilling provided in an embodiment of the present invention; Figure 2 This is a schematic diagram showing the normal drilling torque and the rotational torque after the drill bit is lifted off the bottom of the well, recorded during the actual drilling process of this invention. Figure 3 This is an image showing the effect of the drill bit being brought out of the well due to "core-peeling" damage caused by stick-slip conditions according to the present invention. Figure 4 This is a schematic diagram of logging data and MSEb data during drill bit stick-slip according to the present invention; Figure 5 This is a schematic diagram of the prediction results of the neural network model when multi-source data is used as input vectors in this invention; Figure 6 This invention provides only MSE b The neural network model prediction results when used as input vectors; Figure 7 This is a schematic diagram illustrating the impact of the hidden layer and its number of nodes on the training results provided by the present invention; Figure 8 These are training result images under different algorithms provided by this invention; Figure 9 This is a diagram showing the impact of the maximum number of iterations on the training results provided by this invention; Figure 10 This is a schematic diagram of the identification and control of drill bit stickiness and slippage during actual drilling provided by the present invention; Figure 11This is a schematic diagram of the drill bit after it has been identified and controlled for stick-slip drilling, as provided by the present invention. Detailed Implementation

[0019] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Many specific details are set forth in the following description to provide a thorough understanding of the present invention. However, the present invention can be practiced in many other ways different from those described herein, and those skilled in the art can make similar modifications without departing from the spirit of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0020] The innovation of this invention lies in the following: This invention relies on the logging system without changing the existing drilling and logging processes, and without adding new measurement-while-drilling systems and processes; This invention constructs a calculation and analysis method for the actual rock-breaking torque and rock-breaking energy utilization efficiency of the drill bit, rather than relying on the total torque and total rock-breaking energy input from the ground, which can more accurately reflect the actual working conditions of the drill bit.

[0021] Example 1, such as Figure 1 As shown, the real-time identification method for stick-slip conditions of horizontal well drilling bits provided in this embodiment of the invention includes: S1, based on logging-while-drilling technology, acquires multi-source data on drill pressure, rotational speed, torque, and mechanical drilling speed, and constructs a method for obtaining the effective rock-breaking torque of the drill bit combined with logging-while-drilling testing, reading the torque value recorded during normal rotary drilling. The torque recorded when the drill string is lifted off the bottom of the well and kept running at the original rotation speed. The effective rock-breaking torque at the drill bit under different drilling times and well depths was obtained. S2. Construct a normalized model for logging data. Input the effective rock-breaking torque of the drill bit under different time and well depth conditions and logging data into the normalized model to calculate the true rock-breaking energy utilization rate. The variation pattern with time and well depth; S3, Analysis of the actual rock-breaking energy utilization rate at the corresponding time and well depth. In addition to multi-source logging data, the dataset is divided into training and test sets in a 7:3 ratio. Combining prior physical knowledge and following the neural network modeling steps, a real-time identification model for stick-slip conditions of horizontal well drill bits based on logging data is constructed. The threshold values ​​of characteristic parameters such as amplitude, frequency, and time-domain rate of change of MSEb are determined to determine whether it is a stick-slip condition, and real-time identification and control of stick-slip conditions of horizontal well drill bits are carried out.

[0022] For example, taking the BP neural network method as an example, a real-time, intelligent method for identifying drill bit stick-slip is constructed through training, testing, and prediction. The first step of the BP neural network method is to pass the input data to the first layer (input layer) of the network, with each input variable corresponding to an input neuron. The second step is to pass the input data through weighted and nonlinear transformations in the input layer to the next layer (hidden layer or output layer). This second step is repeated until the data reaches the last layer (output layer), and the network output is calculated. The third step is to compare the output with the actual labels and calculate the error. For logging-while-drilling data, the training process of the BP neural network requires a large amount of computational and storage resources, which can easily lead to large error values ​​and low fit. To improve the accuracy and performance of the network, this invention first normalizes the logging-while-drilling data before constructing the BP neural network method. Therefore, this invention first constructs a logging data normalization model and method that incorporates prior physical knowledge.

[0023] For example, in step S2, during drilling, the drill bit performs work on the rock at the bottom of the well under the action of drilling pressure and torque, breaking the rock to obtain the rate of advance. Teale first proposed the concept of energy consumed per unit volume of rock breaking ( The rock-breaking efficiency is evaluated using the following expression: ; In the formula, The specific energy is mechanical, and based on prior knowledge of physics, its unit is Pa. Drilling pressure, in N; This refers to the cross-sectional area of ​​the wellbore, in meters (m²). 2 ; Rotary / top drive speed, in r / s; This refers to the output torque of the turntable / top drive, expressed in N·m. This is the mechanical drilling speed, measured in m / s.

[0024] The first term on the right side of the above equation represents the drilling pressure efficiency of the downhole drill bit; while the second term represents the total torque efficiency of the surface rotary table / top drive, i.e., the torque in the above equation. It is not the actual torque consumed by the drill bit, but rather the sum of the torque consumed by the drill bit and the torque consumed by the drill string, plus the rotational speed recorded by the logging system. The speed of the rotary table / top drive is not the drill bit speed, therefore the above formula cannot directly reflect the drilling conditions of the drill bit. Accurately obtaining the drill bit's rock-breaking torque is the key to solving the above problem.

[0025] Existing technologies no longer use surface torque data recorded by well logging systems, but instead propose calculating drill bit torque based on the integral relationship between drill bit torque and drill pressure. ; ; In the formula, This refers to the drill bit torque. The drill bit diameter is in meters (m). This is the radial distance from the drill bit center to the rock-breaking contact point in polar coordinates, in meters. In polar coordinates, the circumferential angle of the rock-breaking contact point around the central axis of the drill bit is expressed in radians. The effective drilling pressure of the drill bit, expressed in N; denoted as the dimensionless friction coefficient of the drill bit in shearing rock under drilling pressure. Due to variations in factors such as bottom hole lithology, drilling fluid parameters, and drill bit type, it remains difficult to use the above model to identify the drilling conditions of the drill bit breaking rock.

[0026] To address the shortcomings of existing purely theoretical calculation models, this invention proposes the following improvement method based on engineering practice. First, the logging data normalization model is modified to incorporate prior physical knowledge into the logging data normalization model: ; In the formula, This represents the actual rock-breaking energy utilization rate of the drill bit. Drilling pressure, unit: ; The cross-sectional area of ​​the wellbore is expressed in units of... ; Rotary / top drive speed, unit: ; For the output speed of downhole power drilling tools, the manufacturer will provide the relationship between its output speed and displacement, in units of... ; The real-time torque recorded by the logging system during normal drilling, in units of... ; The average torque recorded by the logging system during normal circulation and rotation of the drill bit as it is lifted from the bottom of the well, in units of... ; This refers to the mechanical drilling speed, measured in units of... ; This represents the actual rotational speed of the drill bit. This represents the actual rock-breaking torque of the drill bit.

[0027] The above formula normalizes multi-source logging data (drill pressure, rotational speed, torque, mechanical drilling speed, displacement, drill bit diameter, and drilling fluid parameters), achieving dimensionality reduction of the neural network model input parameters, which is beneficial to improving the computational efficiency of subsequent data analysis and intelligent processing. Compared with existing technologies, this invention can more directly reflect the drill bit's operating conditions.

[0028] For example, in step S1, the present invention proposes to combine the effective torque of the drill bit with drilling testing. The method for obtaining it. For example... Figure 2The main steps of this method are as follows, as shown in the records of normal drilling torque and rotational torque after the drill bit is lifted off the bottom of the well during actual drilling: ① The torque values ​​recorded by the surface logging system during normal rotary drilling can be read from the data before 10:20. It is easy to know Equal to drill string rotation torque and drill bit rock breaking torque The sum of (including the torque provided by downhole power drilling tools to the drill bit). In Figure 2 In the example shown, when the drilling pressure is 120 kN, The value fluctuates between 9 and 18 kN·m, and can be calculated using data exported from the backend. The average value is 13.5 kN·m.

[0029] For example, Table 1 shows some data obtained from well logging during normal drilling of a horizontal well. The average torque from 10:30 to 10:32 (a longer period can be read as needed) is 24.50 kN·m.

[0030] Table 1. Some drilling parameters automatically recorded by the logging system during normal drilling of Well X.

[0031] ② At 10:20, the drill string was raised 5m on site. After that, the drill bit was removed from the bottom of the well, and the drilling pressure dropped to 0, but the rotational speed remained the same as in step ① (75 RPM). The torque recorded by the logging system at this time was read. In this step, since the drill bit no longer consumes torque, at well depths of several thousand meters... This can be considered as being related to step ① Same. Figure 2 In the case shown, The range is between 8 and 9 kN·m. The average value is 8.5 kN·m.

[0032] For example, Table 2 shows some drilling parameters automatically recorded by the logging system when the drill bit is lifted from the bottom of the well and the drilling is being rotated and circulated for rock clearing. The average torque during this period is 18.10 kN·m. It is easy to see that this value corresponds to the value in formula (3). .

[0033] Table 2. Some drilling parameters automatically recorded by the logging system during normal rotating circulation and rock clearing when the drill bit of Well X is lifted from the bottom of the well.

[0034] ③ Lower the drill bit back to the bottom of the well, increase the drilling pressure to the normal value, and adjust the effective rock-breaking torque at the drill bit. It can be calculated using the following formula. In the given example, Equal to 5 kN·m. The innovative aspect of this invention is: ; In the formula, To achieve effective rock-breaking torque, For torque and, The torque recorded by the logging system; like Figure 2 As shown, since step ① maintains normal rotary drilling, the non-productive time generated by step ② is only 10 minutes. It should be noted that during horizontal well drilling, the drill string needs to be frequently moved up and down under pump start-up and drill string selection conditions to improve the rock-carrying efficiency of the horizontal section; that is, step ② is basically unnecessary to design specifically. Therefore, the method proposed in this invention can obtain drill bit torque without changing existing drilling technology and logging equipment.

[0035] In another example, since the well depth exceeds 4700 meters at this point, and the drill bit and drill string only rise 5 meters, the torque consumed by the drill string in both time periods can be considered equal. Therefore, the average rock-breaking torque of the drill bit during the 10:30-10:32 period can be deduced to be 24.50 - 18.10 = 6.40 (kN·m). The drill bit is lowered back to the bottom of the well, and the drilling pressure is increased to the normal value to continue drilling. The average torque consumed by the drill string rotation during the subsequent 200 meters of drilling is... Continuing with a value of 18.10 kN·m will introduce approximately 200 / 4700 = 4.3% error, which meets engineering requirements; alternatively, multiple tests can be conducted to establish... The functional relationship with well depth. The actual rock-breaking torque of the drill bit afterwards is based on the real-time torque during normal drilling. With average torque Calculate the difference.

[0036] It should be noted that during horizontal well drilling, the drill string needs to be moved up and down frequently under pump start-up and drill string selection conditions to improve the rock-carrying efficiency of the horizontal section, meaning that step ② is basically not required. Therefore, the method proposed in this invention can obtain the drill bit torque without changing the existing drilling process and logging equipment.

[0037] although It will increase with well depth, but the daily footage in deep horizontal sections is limited and can be ignored for a period of time compared to well depths of several thousand meters. The changes; in addition, if necessary, the above tests can be repeated daily to update. The value obtained from equation (4) Substituting into equation (3) yields the normalized model for the rock-breaking efficiency of the drill bit used in this invention. The above method has been applied in actual drilling in the field and has been recognized and welcomed by engineering and technical personnel.

[0038] After each well is completed, the construction unit will collect and archive logging data and well history data (including drill bit exit photos and damage descriptions) according to a standardized format. Drill bit stick-slip vibration is a contributing factor to "core" damage to the drill bit (e.g., Figure 3 As shown in the example of a drill bit exhibiting "core-penetration" damage due to stick-slip conditions (where the crown and inner conical teeth broke off, resulting in severe wear and loss of the drill bit body, making further drilling impossible), well history data from adjacent wells was collected to identify and plot the corresponding logging data at the time of the stick-slip incident. Figure 4 Logging data and when drill bit stick-slip occurs MSE b The data is shown in the chart. From Figure 4 It can be seen that after 10:30, the drill bit experienced stick slip, and the torque fluctuated between 6-16 kN·m, changing from the normal 7-11 kN·m. The method constructed in this invention, which combines drilling testing... The model normalized the multi-source parameter data and amplified the fluctuations in the original logging data (from the normal 200-1000 ksi to fluctuations between 50-2000 ksi, with the mean increasing from 600 ksi to 1200 ksi and the fluctuation frequency decreasing to less than half of the normal well section). This can reflect the real-time downhole drill bit rock breaking conditions and provide a reliable basis for identifying drill bit sticking and slipping.

[0039] For example, by reviewing well history data, the output speed of the downhole power drilling tool is as follows when the drill bit diameter is 215.9 mm and the displacement is 30 L / min. N m The value is 80 RPM. At this point, the values ​​of all variables on the right side of formula (3) are obtained. Substituting these values ​​into formula (3) yields the rock-breaking efficiency of the drill bit at the corresponding time and well depth. .

[0040] For example, step S3 includes: importing logging data from multiple drill bit stick-slip operations. Computational model, analysis The characteristic parameters such as amplitude, frequency, and time-domain rate of change will The data is passed to the first layer of the network (input layer), where each input variable corresponds to an input neuron; After weighting and nonlinear transformation at the input layer, the data is passed to the next layer of the network (hidden layer or output layer); the second step is repeated until the data is passed to the last layer of the network (output layer), and the network output is calculated; the output is compared with the actual label, and the error is calculated.

[0041] A real-time identification model for horizontal well bit stick-slip conditions based on logging data was constructed following neural network modeling steps including training and testing. The reliability of the stick-slip prediction model is further verified by adjusting the threshold values ​​of each characteristic parameter. This will then... The computational model and the real-time identification model of stick-slip conditions of horizontal well drill bits based on logging data are embedded in the logging system to enable real-time identification of stick-slip conditions during drilling and to provide prompts, thus providing support for timely manual intervention and avoiding drill bit damage.

[0042] Specifically, the following steps are included: (1) such as Figure 5 The prediction results of the neural network model when using multi-source data as input vectors are shown. The true value of the well section under normal working conditions is recorded as 1, and the true value of the well section where stick-slip occurs is recorded as 2. The number of hidden layer nodes is set to 4. The training algorithm is selected as the trainscg algorithm, and the excitation functions are selected as tansing and purelin. Multiple parameters such as mechanical drilling rate, torque, and drilling pressure are used as input vectors and imported into the BP neural network for training and testing. Through training and testing, it was found that the error is large and the fitting degree is extremely small, and the recognition accuracy is relatively low.

[0043] (2) For example Figure 6 only The neural network model prediction results, when used as input vectors, show that the true value of the well section under normal operating conditions is denoted as 1, and the true value of the well section where stick-slip occurs is denoted as 2. The number of hidden layer nodes is set to 4, the training algorithm is selected as the trainscg algorithm, and the activation functions are selected as tansing and purelin. The parameters are used as input vectors for training and testing. The predicted values ​​of the test set have a high degree of fit with the true values ​​and small errors. It can completely identify normal working conditions and has a high recognition accuracy.

[0044] (3) Comparing steps (1) and (2), it can be found that, with Neural network models achieve higher prediction accuracy when a single parameter is used as an input vector.

[0045] (4) such as Figure 7 The impact of hidden layers and their number of nodes on training results is shown, and the optimal number of nodes can be selected by changing the number of hidden layer nodes. Higher fit and smaller error values ​​result in higher accuracy. Using the Trainlm / LMs (Levenberg-Marguardt) algorithm with a maximum iteration count of 1000, neural network training was performed with MSE inputs ranging from 4 to 20 hidden layer nodes. Based on the fit and error values, it was found that a hidden layer node count of 5 resulted in the highest fit and the smallest error.

[0046] (5) Furthermore, with the maximum number of iterations fixed at 1000, the Trianscg and Traind algorithms were replaced to optimize the number of hidden layer nodes. The results were compared with those of the Trainlm / LMs algorithm. Figure 8The training results for different algorithms are shown. It can be seen that the optimal number of hidden layer nodes is basically 5 for all algorithms. With the Traingd algorithm, fitting failure occurs earlier as the number of hidden layer nodes increases. Therefore, the Trianscg algorithm is selected as the optimal algorithm, and the optimal number of hidden layer nodes is determined to be 5.

[0047] (6) For example Figure 9 The impact of the maximum number of iterations on the training results is shown. Under the optimal algorithm and the optimal number of nodes, only the maximum number of iterations was changed for testing. The maximum number of iterations was selected to be 100-1500 for training the BP artificial neural network. After testing, it was found that the optimal maximum number of iterations was 1000.

[0048] Under optimal model parameters (maximum number of iterations of 1000 and number of hidden layer nodes of 5), the Trainscg algorithm was used for 10-fold cross-validation training and testing. The results are shown in Table 3. It can be seen that when using different subsets of the dataset as the test dataset, the differences in goodness of fit, relative error, and time consumption are very small, indicating that the BP artificial neural network model is stable and efficient. Furthermore, the average goodness of fit is 0.94, and the average relative error is 1.3%, indicating that the BP artificial neural network has good automatic stick-slip recognition capabilities.

[0049] Table 3. Results of 10-fold cross-validation

[0050] Finally, The computational model and the real-time identification model of stick-slip conditions of horizontal well drill bits based on logging data are embedded in the logging system to enable real-time identification of stick-slip conditions during drilling and to provide prompts, thus providing support for timely manual intervention and avoiding drill bit damage.

[0051] Compared with manual judgment and verification, the present invention is more accurate in identifying inefficient working conditions; compared with the method of installing downhole vibration sensors, the method proposed in this invention does not change the existing process technology and does not increase additional costs.

[0052] Example 2: The present invention provides a real-time identification system for stick-slip conditions of horizontal well drilling bits, the system comprising: The module for obtaining the true rock-breaking energy utilization rate of the drill bit is used to normalize multi-source logging data using a logging data normalization model that incorporates prior physical knowledge, thereby obtaining the true rock-breaking energy utilization rate of the drill bit. The multi-source logging data includes drilling pressure, torque, and mechanical drilling speed. The effective rock-breaking torque acquisition module is used to acquire the effective rock-breaking torque at the drill bit at different times and well depths under normal drilling pressure during drilling tests by combining the method of acquiring the effective torque of the drill bit during drilling tests. The acquired effective rock-breaking torque at the drill bit at different times and well depths under normal drilling pressure is then input into the logging data normalization model that combines physical prior knowledge to obtain multiple logging data of the drill bit sticking during drilling tests, including the true rock-breaking energy utilization rate of the drill bit at the corresponding time and well depth. The module for constructing a real-time identification model of horizontal well bit stick-slip condition based on logging data is used to analyze the amplitude, frequency, and time-domain variation rate characteristic parameters of the actual rock-breaking energy utilization rate of the drill bit at the corresponding time and well depth. Following the training and testing neural network modeling steps, a real-time identification model of horizontal well bit stick-slip condition based on logging data is constructed, and the real-time identification of horizontal well bit stick-slip condition is performed.

[0053] To further illustrate the effects of the embodiments of the present invention, the following experiments were conducted.

[0054] like Figure 10 The method for identifying and controlling bit stick-slip during actual drilling is shown in this paper: During the actual drilling process of a certain well, the stick-slip was identified using the method proposed in this paper. The significant increase in mean and amplitude, coupled with a decrease in frequency, triggered a drill bit stick-slip warning. Subsequently, to address the stick-slip condition, drilling parameters were adjusted starting at 02:40 (reducing drilling pressure), resolving the stick-slip condition, preventing drill bit damage, and completing the drilling trip. Figure 10 As shown, The normalization and amplification effects on logging data are evident.

[0055] Figure 11 The drill bit that emerged from the well after the actual drilling identification and control of the stick-slip is shown in the above-mentioned well-emerging photo. It can be seen that although the crown tooth and inner cone tooth of the drill bit showed stick-slip damage symptoms, the drill bit damage did not develop into "core removal" failure after the drilling parameters were adjusted to suppress the stick-slip condition, which confirms the practical value of the present invention.

[0056] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications, equivalent substitutions and improvements made by those skilled in the art within the scope of the technology disclosed in the present invention and within the spirit and principles of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A method for real-time identification of stick-slip conditions in horizontal well drilling bits, characterized in that, The method includes the following steps: S1, based on logging-while-drilling technology, acquires multi-source data on drill pressure, rotational speed, torque, and mechanical drilling speed, and constructs a method for obtaining the effective rock-breaking torque of the drill bit combined with logging-while-drilling testing, reading the torque value recorded during normal rotary drilling. The torque recorded when the drill string is lifted off the bottom of the well and kept running at the original rotation speed. The effective rock-breaking torque at the drill bit under different drilling times and well depths was obtained. S2. Construct a normalized model for logging data. Input the effective rock-breaking torque of the drill bit under different time and well depth conditions and logging data into the normalized model to calculate the true rock-breaking energy utilization rate. The variation pattern with time and well depth; S3, Analysis of the actual rock-breaking energy utilization rate at the corresponding time and well depth. In addition to multi-source logging data, the dataset was divided into training and test sets in a 7:3 ratio. The data was normalized using prior physical knowledge, and a backpropagation (BP) neural network was constructed for parameter optimization, training, and validation. This led to the development of a real-time identification model for horizontal well bit stick-slip conditions based on logging data. The threshold values ​​of characteristic parameters such as amplitude, frequency, and time-domain rate of change are used to determine whether a stick-slip condition is present, enabling real-time identification and control of stick-slip conditions in horizontal well drill bits.

2. The method for real-time identification of stick-slip conditions of horizontal well drilling bits according to claim 1, characterized in that, In step S1, acquiring multi-source data on drilling pressure, rotational speed, torque, and mechanical drilling speed includes: S101, Read the torque during normal rotary drilling and , Equal to drill string rotation torque and effective rock-breaking torque sum; S102. Raise the drill string, the drill bit leaves the bottom of the well, the drilling pressure drops to 0 but the rotational speed remains unchanged, and read the torque recorded by the logging system at this time. ; With step S101 The values ​​are the same; S103. Lower the drill bit back to the bottom of the well, increase the drilling pressure to the normal value, and set the effective rock-breaking torque at the drill bit. It is obtained by calculation using the following formula; ； In the formula, To achieve effective rock-breaking torque, For torque and, The torque recorded by the logging system; Torque recorded by the logging system The progress varies with well depth, but the daily footage in the horizontal section is limited, and short-run cleaning is required to obtain data at multiple well depths. And then obtain through linear interpolation The functional relationship between the well depth and the depth is expressed in the above formula. It is a function of well depth; the obtained By incorporating a normalized model of logging data that incorporates prior physical knowledge, the rock-breaking efficiency of the drill bit at the corresponding time and well depth can be obtained. .

3. The method for real-time identification of stick-slip conditions in horizontal well drilling bits according to claim 1, characterized in that, In step S2, the normalization model for logging data is constructed as follows: ； In the formula, This represents the actual rock-breaking energy utilization rate of the drill bit. Drilling pressure, unit: ; The cross-sectional area of ​​the wellbore is expressed in units of... ; Rotary / top drive speed, unit: ; For the output speed of downhole power drilling tools, the manufacturer will provide the relationship between its output speed and displacement, in units of... ; The real-time torque recorded by the logging system during normal drilling, in units of... ; The average torque recorded by the logging system during normal circulation and rotation of the drill bit as it is lifted from the bottom of the well, in units of... ; This refers to the mechanical drilling speed, measured in units of... ; This represents the actual rotational speed of the drill bit. This represents the actual rock-breaking torque of the drill bit.

4. The method for real-time identification of stick-slip conditions of horizontal well drilling bits according to claim 1, characterized in that, In step S3, constructing a real-time identification model for stick-slip conditions of horizontal well drill bits based on logging data includes: S301. The actual rock-breaking energy utilization rate of the drill bit at the corresponding time and well depth is transmitted to the first input layer of the neural network, with each input variable corresponding to an input neuron. S302. The actual rock-breaking energy utilization rate of the drill bit at the corresponding time and well depth is weighted and nonlinearly transformed in the input layer and then transmitted to the next hidden layer or output layer of the network. Repeat the above steps until the data is transmitted to the last output layer of the network and the output result of the network is calculated. S303. Compare the output results with the actual labels, and use MAE and R2 to calculate the average relative error and goodness of fit between the output value and the true value; divide the dataset into training set and test set in a 7:3 ratio, and combine physical prior knowledge to construct a real-time identification model for horizontal well bit stick-slip conditions based on logging data according to the neural network modeling steps.

5. The method for real-time identification of stick-slip conditions in horizontal well drilling bits according to claim 4, characterized in that, After constructing a real-time identification model for stick-slip conditions of horizontal well drill bits based on logging data, if the output value of the model is 2, it indicates a stick-slip condition. When the deviation between the model's predicted value and the actual value exceeds a set threshold, it is necessary to compare the characteristic parameters of the amplitude and frequency of the actual rock-breaking energy utilization rate of the drill bit with the changes in time and well depth with the threshold to determine whether the condition is stick-slip and further verify the reliability of the real-time identification model for stick-slip conditions of horizontal well drill bits based on logging data.

6. The method for real-time identification of stick-slip conditions in horizontal well drilling bits according to claim 4, characterized in that, Step S3, which involves constructing a real-time identification model for stick-slip conditions of horizontal well drill bits based on logging data, specifically includes the following steps: (1) Record the true value of the well section under normal working conditions as 1, the true value of the well section where stick-slip occurs as 2, set the number of hidden layer nodes to 4, select the trainscg algorithm for training, select tansing and purelin for excitation functions, and import multiple parameters such as mechanical drilling speed, torque and drilling pressure as input vectors into the BP neural network for training and testing. (2) Record the true value of the well section under normal working conditions as 1, and the true value of the well section where stick-slip condition occurs as 2. Set the number of hidden layer nodes to 4, select the trainscg algorithm for training, select tansing and purelin for excitation functions, and select multiple parameters such as mechanical drilling speed, torque and drilling pressure as input vectors for training and testing. (3) Compare steps (1) and (2), using the MSEb single-item parameter as the input vector; (4) Select the optimal number of nodes by changing the number of hidden layer nodes; use the Trainlm / LMs algorithm, select the maximum number of iteration steps as 1000, and input MSE to train the neural network with a hidden layer node count of 4-20; (5) Fix the maximum number of iterations to 1000, select Trianscg as the optimal algorithm, and optimize the number of hidden layer nodes; (6) Under the optimal algorithm and the optimal number of nodes, only the maximum number of iterations is changed for testing. The maximum number of iterations is selected as 100-1500 for training the BP artificial neural network.

7. The method for real-time identification of stick-slip conditions in horizontal well drilling bits according to claim 1, characterized in that, In step S3, the actual rock-breaking energy utilization rate at the corresponding time and well depth of the drill bit and the real-time identification model of stick-slip condition of horizontal well drill bit based on logging data are embedded into the logging system to identify stick-slip condition in real time during drilling and provide prompts.

8. A real-time identification system for stick-slip conditions of drill bits in horizontal well drilling, characterized in that, The system implements the real-time identification method for stick-slip conditions of horizontal well drilling bits as described in any one of claims 1-7, and the system includes: A module for calculating effective rock-breaking torque of the drill bit is constructed. This module integrates a method for obtaining the effective rock-breaking torque of the drill bit with drilling tests, yielding the effective rock-breaking torque at different times and well depths. The rock-breaking torque is then fitted as a function of the well depth, allowing for real-time prediction of the rock-breaking torque under subsequent drilling conditions. The fitting function between the drill bit's rock-breaking torque and well depth is iteratively corrected based on subsequent drilling test results. The module inputs the obtained effective rock-breaking torque at different times and well depths under normal drilling pressure into a logging data normalization model incorporating prior physical knowledge. This model obtains multiple logging data points during drilling tests where the drill bit experiences stick-slip, including the actual rock-breaking energy utilization rate at the corresponding time and well depth. A module for calculating the true rock-breaking energy utilization rate at the drill bit; constructing a logging data normalization model to normalize multi-source logging data such as effective rock-breaking torque, top drive speed, displacement, and drilling pressure of the drill bit to obtain the true rock-breaking energy utilization rate of the drill bit; A real-time identification model and method for horizontal well bit stick-slip conditions based on logging data is proposed. By combining well history prior data, the evolution data of the drill bit rock breaking energy utilization rate with time and well depth under normal drilling and stick-slip conditions are obtained. Then, according to the training and testing neural network modeling steps, a real-time identification model for horizontal well drilling bit stick-slip conditions based on logging data is constructed, and real-time identification and control of horizontal well bit stick-slip conditions are carried out.

9. The real-time identification system for stick-slip conditions of horizontal well drilling bits according to claim 8, characterized in that, The system is mounted on a computer device, which includes at least one processor, a memory, and a computer program stored in the memory and running on the at least one processor, wherein the processor executes the computer program to perform the above-mentioned functions.

10. The real-time identification system for stick-slip condition of horizontal well drilling bits according to claim 8, characterized in that, The system is mounted on a computer-readable storage medium that stores a computer program that, when executed by a processor, can perform the above-mentioned functions.

Images