CN110500034B - Method for establishing neural network model, determining torsional pendulum drill string parameters and directionally drilling - Google Patents
Method for establishing neural network model, determining torsional pendulum drill string parameters and directionally drilling Download PDFInfo
- Publication number
- CN110500034B CN110500034B CN201910819123.0A CN201910819123A CN110500034B CN 110500034 B CN110500034 B CN 110500034B CN 201910819123 A CN201910819123 A CN 201910819123A CN 110500034 B CN110500034 B CN 110500034B
- Authority
- CN
- China
- Prior art keywords
- neural network
- network model
- data
- drilling
- drill string
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000005553 drilling Methods 0.000 title claims abstract description 146
- 238000000034 method Methods 0.000 title claims abstract description 102
- 238000003062 neural network model Methods 0.000 title claims abstract description 97
- 238000012360 testing method Methods 0.000 claims abstract description 34
- 238000012549 training Methods 0.000 claims abstract description 32
- 230000008569 process Effects 0.000 claims abstract description 29
- 238000005259 measurement Methods 0.000 claims description 18
- 238000013528 artificial neural network Methods 0.000 claims description 11
- 238000004088 simulation Methods 0.000 claims description 7
- 239000012530 fluid Substances 0.000 claims description 6
- 238000010200 validation analysis Methods 0.000 claims description 6
- 230000000712 assembly Effects 0.000 claims 1
- 238000000429 assembly Methods 0.000 claims 1
- 238000005516 engineering process Methods 0.000 abstract description 8
- 230000009286 beneficial effect Effects 0.000 abstract description 4
- 238000012795 verification Methods 0.000 description 9
- 230000007787 long-term memory Effects 0.000 description 4
- 238000005457 optimization Methods 0.000 description 4
- 230000006403 short-term memory Effects 0.000 description 4
- 230000015572 biosynthetic process Effects 0.000 description 3
- 230000004044 response Effects 0.000 description 3
- 238000012216 screening Methods 0.000 description 3
- 238000013473 artificial intelligence Methods 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 230000035515 penetration Effects 0.000 description 2
- 230000000306 recurrent effect Effects 0.000 description 2
- 239000004215 Carbon black (E152) Substances 0.000 description 1
- 241000512668 Eunectes Species 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 230000009918 complex formation Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000002349 favourable effect Effects 0.000 description 1
- 229930195733 hydrocarbon Natural products 0.000 description 1
- 150000002430 hydrocarbons Chemical class 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 239000007788 liquid Substances 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 230000015654 memory Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000002028 premature Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000002441 reversible effect Effects 0.000 description 1
- 239000011435 rock Substances 0.000 description 1
- 239000002002 slurry Substances 0.000 description 1
- 239000003381 stabilizer Substances 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
Images
Classifications
-
- E—FIXED CONSTRUCTIONS
- E21—EARTH OR ROCK DRILLING; MINING
- E21B—EARTH OR ROCK DRILLING; OBTAINING OIL, GAS, WATER, SOLUBLE OR MELTABLE MATERIALS OR A SLURRY OF MINERALS FROM WELLS
- E21B7/00—Special methods or apparatus for drilling
- E21B7/04—Directional drilling
Landscapes
- Life Sciences & Earth Sciences (AREA)
- Engineering & Computer Science (AREA)
- Geology (AREA)
- Mining & Mineral Resources (AREA)
- Physics & Mathematics (AREA)
- Environmental & Geological Engineering (AREA)
- Fluid Mechanics (AREA)
- General Life Sciences & Earth Sciences (AREA)
- Geochemistry & Mineralogy (AREA)
- Earth Drilling (AREA)
Abstract
The invention provides a method for establishing a sliding directional drilling neural network model, a method for determining torsional pendulum drill string control parameters and a directional drilling method. The model establishing method comprises the following steps: collecting historical operation data and establishing a database; establishing an initial neural network model based on a database; training an initial neural network model; and testing the accuracy of the model, if the accuracy is more than a preset value, determining the model is a sliding directional drilling neural network model, and if the accuracy is less than the preset value, continuing training. The control parameter determination method comprises the following steps: the method is adopted to establish a neural network model, and torsional pendulum drill string control parameters in the drilling process are determined based on the model. The directional drilling method comprises the steps of adopting the torsional pendulum drill string to control parameters during the drilling process, changing or maintaining the direction of the tool face of the downhole power drill tool, and controlling the track of the well hole to extend towards the target direction. The beneficial effects of the invention include: the method is simple; the neural network model is stable and reliable, and is beneficial to improving the automation level of the directional drilling technology.
Description
Technical Field
The invention relates to the technical field of oil and gas drilling engineering, in particular to a method for establishing a sliding directional drilling neural network model, a method for determining torsional pendulum drill string control parameters in a sliding directional drilling process and a method for sliding directional drilling.
Background
In the development of unconventional reservoirs, the most troublesome problem faced is deviation of the drilling trajectory, which, due to complex formation conditions, usually deviates from the pre-set drilling scheme. Sliding drilling is the main mode of trajectory control of a classical directional well or a horizontal well, and can change the direction of a well hole, increase the inclination and reduce the inclination. The whole sliding drilling system can be divided into an uphole part and a downhole part; the underground part mainly comprises a winch, a crown block, a traveling block, a top drive, a slurry pump and the like, wherein the crown block and the traveling block form a pulley block for lifting and lowering the drilling tool, and the top drive is driven by a motor to rotate a drill column so as to control a tool surface; the downhole part is a drilling tool assembly BHA (bottom hole assembly), and mainly comprises a drill pipe, a weighted drill pipe, a drill collar, a stabilizer, a MWD (measurement while drilling), a downhole power drilling tool, a drill bit and the like from top to bottom, wherein the MWD can measure the tool face angle of the downhole power drilling tool, convert data into pulse signals and upload the pulse signals to the ground through mud liquid for decoding.
In the conventional sliding directional drilling process, a drill string does not rotate but axially slides along a well wall, in the practical process, the inclination angle and the azimuth angle of a well hole are changed mainly by changing or maintaining the tool face of the underground power drilling tool, so that the aim of controlling the track of the well hole is fulfilled, and if the deviation problem of the track of the drilled well hole and the designed well hole is caused and the deviation problem needs to be caused or the azimuth is twisted, the tool face of the underground power drilling tool needs to be changed or maintained in a mode of manually controlling a top drive or a rotary table to rotate the drill string for a certain angle, so that the requirement of adjusting the track of the well hole. In the whole sliding drilling process, because a drilling tool does not rotate, static friction force accumulates on a drill column along with the extension of a horizontal section to generate a 'pressure supporting' phenomenon, the drill pressure cannot be effectively transmitted and even the drill column is bent, so that a series of problems of low drilling speed, low pure drilling time efficiency, premature damage of an underground pipe column and tools and the like are caused, even complex accidents such as drill sticking and the like are induced, the drilling operation risk is high, once an emergency occurs, a drill bit needs to be lifted away from the bottom of a well, a tool face angle is readjusted, directional drilling can be continued after the drill bit is stabilized, time is consumed, and the increase of the length of the horizontal section and the increase of the directional drilling speed are severely restricted.
In order to solve the problems, a new control system based on top drive for alternately swinging the drill string clockwise and anticlockwise on the ground, namely a drill string torsion swing system, is developed in recent years. The system applies forward and reverse torsion to the drill string by controlling the top drive, drives the drilling tool to rotate clockwise and anticlockwise alternately, releases friction resistance and torque of the drill string, and relieves the phenomenon of 'pressure support' caused by irrotation of the drill string, thereby improving the mechanical drilling speed and breaking through the limitation faced by sliding drilling. The treatment method has the following defects:
the underground working conditions of actual drilling are very complex, various parameters are full of uncertainty, the response of a tool face angle is influenced by various factors such as stratum lithology, bit pressure, drilling speed and the like, great interference and uncertainty are brought to the dynamic control of the tool face angle, a group of proper control parameters are difficult to select in advance to meet various complex underground working conditions, repeated adjustment is often needed depending on the experience of drillers and directional engineers, the efficiency is low, and the response speed of the tool face is slow.
Secondly, the torque set value, the movement speed, the holding time and other factors of the wellhead torsional pendulum drill string influence the torque transmission of the drill string, and an accurate mathematical formula or a mechanical model does not exist. The drill string torsional pendulum system is only a top drive executing mechanism and does not have an automatic control function.
Disclosure of Invention
In view of the deficiencies in the prior art, it is an object of the present invention to address one or more of the problems in the prior art as set forth above. For example, one of the objectives of the present invention is to provide a method for building a neural network model for sliding directional drilling, a method for determining control parameters of a torsional pendulum drill string during sliding directional drilling, and a method for sliding directional drilling, so as to improve the automation level of the directional drilling technology.
In order to achieve the above object, an aspect of the present invention provides a method for building a sliding directional drilling neural network model, which may include the steps of: collecting operation data of a drilled and produced oil and gas well and establishing a database, wherein the operation data comprises geological parameters and historical directional operation data; establishing an initial neural network model based on a database; training an initial neural network model; and testing the accuracy of the trained neural network model, if the accuracy is higher than a preset value, the trained neural network model is a sliding directional drilling neural network model, and if the accuracy is lower than the preset value, continuing to train the neural network model until the accuracy reaches the preset value.
In one exemplary embodiment of the method of establishing a sliding directional drilling neural network model of the present invention, the historical directional operation data may include historical wiggle drill string control parameters, historical logging data, and historical measurement while drilling data.
In an exemplary embodiment of the method of establishing a neural network model for sliding directional drilling of the present invention, the historical directional operation data may further comprise artificial empirical data, which may comprise at least one of a drilling assembly, drilling fluid performance parameters, and a wellbore trajectory.
In an exemplary embodiment of the method of building a sliding directional drilling neural network model of the present invention, after building the database and before building the initial neural network model, the method may further comprise the steps of: the data in the database is screened to remove duplicate, erroneous, and invalid data.
In an exemplary embodiment of the method of building a sliding directional drilling neural network model of the present invention, after building the database and before building the initial neural network model, the method further comprises the steps of: and extracting the data characteristics of the historical oriented operation.
In an exemplary embodiment of the method for establishing a sliding directional drilling neural network model of the present invention, the database may comprise a ratio of 5-8: 1-3: 1-2 of a training set, a validation set and a test set.
In one exemplary embodiment of the method of building a sliding directional drilling neural network model of the present invention, the type of the initial neural network model may comprise an artificial neural network model.
In one exemplary embodiment of the method of modeling a sliding directional drilling neural network of the present invention, the testing may comprise simulation testing.
The invention further provides a method for determining control parameters of the torsional pendulum drill string in the sliding directional drilling process. The method may comprise the steps of: collecting operation data of a drilled and produced oil and gas well and establishing a database, wherein the operation data comprises geological parameters, historical torsional pendulum drill string control parameters, historical logging data and historical measurement while drilling data, or the operation data comprises the geological parameters, the historical torsional pendulum drill string control parameters, the historical logging data, the historical measurement while drilling data and artificial experience data; establishing an initial neural network model based on a database; training an initial neural network model; testing the accuracy of the trained neural network model, if the accuracy is above a preset value, the trained neural network model is the selected neural network model, and if the accuracy is less than the preset value, continuing to train the neural network model until the accuracy reaches above the preset value; substituting geological parameters, logging data and measurement while drilling data obtained in the field sliding directional drilling process into the selected neural network model under the condition that the operation data does not include artificial experience data to obtain torsional pendulum drill string control parameters; and substituting the geological parameters, path data, measurement while drilling data and the artificial experience data acquired in the field sliding directional drilling process into the selected neural network model under the condition that the operation data comprise the artificial experience data to obtain the control parameters of the torsional pendulum drill string.
Yet another aspect of the present invention provides a method of sliding directional drilling, which may include: in the directional drilling process, the method is adopted to determine a torsional pendulum drill string control parameter in the sliding directional drilling process, change or maintain the direction of a tool face of a downhole power drill tool and control the extension of a well track to a target direction.
Compared with the prior art, the beneficial effects of the invention can include: the method is simple and easy to implement, the established network control model in the sliding directional drilling process is stable and reliable, the optimal directional control parameters of the torsional pendulum drill string can be obtained, and the method is favorable for improving the automation level of the directional drilling technology.
Drawings
The above and other objects and features of the present invention will become more apparent from the following description taken in conjunction with the accompanying drawings, in which:
FIG. 1 is a schematic flow chart of the method of constructing a neural network model for sliding directional drilling according to the present invention;
fig. 2 identifies a flow diagram in an example of the invention.
Detailed Description
Hereinafter, a method of a sliding directional drilling neural network model (also referred to as a network control model) and a method of determining a wiggle drill string control parameter during sliding directional drilling according to the present invention will be described in detail with reference to the accompanying drawings and exemplary embodiments.
Due to the complexity and uncertainty of the actual drilling underground working condition, the dynamic control of the tool face angle is greatly disturbed, and the control parameters selected in advance are difficult to meet various complex working conditions. Because factors such as a wellhead torsional pendulum drill string movement torque set value, a torsional pendulum drill string movement speed, holding time and the like have no accurate data support on the influence of the drill string torque transmission, a drill string swing system of the sliding directional drilling generally does not have an automatic control function; furthermore, the difficulty in controlling the face of a sliding steerable drilling tool is that the downhole drill string is similar to a flexible beam of large slenderness ratio, there is a great deal of contact and friction with the borehole wall, and the interaction of different bits with the rock is complex.
In order to solve the problem that the underground working condition of the drilling is complex, the uncertainty of various parameters has influence on the control of the tool face angle, improve the automation level of the directional drilling technology and improve the self-adaptive capacity under different environments, the invention provides a method for determining the control parameters of a torsional pendulum drill column in the sliding directional drilling process, and the method can be realized based on a neural network model.
The invention provides a method for establishing a sliding directional drilling neural network model.
In an exemplary embodiment of the method of the present invention for modeling a sliding directional drilling neural network, as shown in fig. 1, the method may comprise the steps of:
s01: collecting operation data of the drilled and produced oil and gas wells, and establishing a database based on the collected data. The operational data includes geological parameters and historical directional operational data. Historical operational data may include twisted pendulum drill string, logging, and measurement while drilling data. Wherein the twisted drill string data (i.e., the twisted drill string control parameters) may include at least one of an input torque, an output torque, a maximum torque hold time, a twisted drill string movement speed, and a torque swing range; the logging data may include at least one of formation pressure, formation fracture pressure, torque, and rate of penetration; the measurement-while-drilling data may include well depth, well deviation, azimuth, tool face, and the like. Further, the historical operational data may also include empirical operational data given by drilling personnel in drilled hydrocarbon well operations, and the empirical operational data may include at least one of a selected drilling tool assembly, drilling fluid performance parameters, and wellbore trajectory.
S02: an initial neural network model is established based on the database. The initial neural network model may include an artificial neural network model, such as a long-term and short-term memory neural network model. The artificial neural network utilizes a specific neural network model and a directional operation data set, can linearize a complex problem through continuous updating iterative training, solves the problems of high dimensionality, separable nonlinearity and the like of sliding directional data, and simultaneously learns the experience of directional data of drilling experts. The long-term and short-term memory neural network model can avoid the problem that the gradient generated by a common recurrent neural network when learning the data rule disappears by improving the recurrent neural network structure, and has higher convergence speed and higher generalization. Establishing the initial neural network model may include: and selecting a proper deep learning framework and constructing a long-term and short-term memory network module. And constructing a long-term and short-term memory network layer with a certain depth, splicing a proper full connection layer, preliminarily drawing out hyper-parameters according to the data structure and the network characteristics, and training and adjusting through the existing data set.
S03: and training the initial neural network model. This step can be based on anaconda platform, and training of neural network model is carried out in computer through command line. The training process mainly comprises forward propagation and backward propagation to adjust proper weight and bias items, and finally the model with the highest accuracy is obtained and stored.
S04: and verifying the trained neural network model, wherein the verification method can comprise simulation test, namely performing simulation test on the trained model by using the torsional pendulum drill string, the logging and measurement while drilling data of the test set. And under the condition that the accuracy of the test is smaller than the preset value, continuously training the initial neural network model until the accuracy reaches above the preset value. If the accuracy of the test is above the preset value, the trained model is a qualified neural network model (namely, a sliding directional drilling neural network model) which passes the test. The predetermined value may be 99%.
In this embodiment, the sliding directional drilling of the present invention may have a control system of the top drive, i.e., a drill string sway system.
In the present embodiment, the steps S02 and S03 may be performed by a computer, which has higher speed and accuracy.
In this embodiment, the method may further include the steps of: after the database is built, the data in the database is screened. The screening comprises the step of eliminating repeated, wrong and invalid data in the database so as to reduce the interference of the data on the training of the neural network model and improve the efficiency and the accuracy of the training of the neural network model.
In this embodiment, after the database is established, the method may further include the steps of: and extracting data characteristics in the sliding directional drilling operation process.
The data in the sliding directional drilling operation process comprises data on a torsional pendulum drill string system, a logging system and a measurement while drilling system, and the purpose of extracting data characteristics is to establish a network control model and an intelligent decision algorithm in the dynamic directional drilling process to obtain optimal torsional pendulum drill string directional control parameters, so that the automation level of the directional drilling technology is improved.
In this embodiment, the data in the database may be divided into a trial set (also referred to as a training set), a validation set and a test set, wherein the training set and the validation set may be applied in step S03, and the test set may be applied in step S04. Wherein the training set can be used to train model parameters; the verification set is used for verifying whether the trained model is available; the test set is the final data used to test the accuracy of the model. More specifically, the training set is used for training the model, different models are trained by trying different methods and ideas and using the training set, then the optimal model is selected by using an effective verification method through the verification set, the performance of the model on the verification set is improved through continuous iteration, and finally the performance of the model is evaluated through the test set.
The test set, the verification set and the test set are regarded as a whole 10, and the three parts can be divided into 5-8 parts in proportion: 1-3: 1-2, the division aims at providing a dynamic division proportion, and a user can select a reasonable proportion according to experience and actual needs so as to achieve the purposes of fully training, verifying and evaluating the model. Dividing the data set according to the proportion can improve the application speed of the model. If the division is not good, the application deployment of the model is greatly influenced, and even the work done can be short of one step. Through training of a large amount of data and optimization selection of an algorithm, the iterative neural network model is continuously updated, and finally a stable control algorithm can be obtained, so that the self-adaptive capacity under different environments is realized.
The neural network model established by the invention is an iterative neural network model which is continuously updated through training of a large amount of data and algorithm optimization selection, and has self-adaptive capacity under different environments.
The invention further provides a method for determining control parameters of the torsional pendulum drill string in the sliding directional drilling process.
The determination method may adopt the method in the above exemplary implementation to establish a sliding directional drilling neural network model, and then obtain the torsional pendulum drill string control parameter based on the model and by combining the relevant data obtained in the field sliding directional drilling process.
In an exemplary embodiment of the method for determining a control parameter of a twisted drill string during sliding directional drilling according to the present invention, the method for determining a control parameter of a twisted drill string during sliding directional drilling (also referred to as a twisted drill string directional control parameter) may comprise the steps of:
steps S01-S04 identical to those in the exemplary embodiment of the method of establishing a sliding directional drilling neural network model.
S05: and obtaining a torsional pendulum drill string control parameter based on the established neural network model and basic data acquired in the field sliding directional drilling process. Basic data can be input into the established model, and then the torque control parameters can be obtained. The base data may include, among other things, wellbore trajectory, well depth structure, drilling tool assembly, friction coefficient, drilling fluid density, and other construction parameters. The underlying data is data other than the wiggle drill string data. Torque control parameters will also affect the control of the toolface, and if the torque is not selected properly, the toolface will be stable, which may result in directional drilling failure.
And under the condition that the historical operation data collected in the step S01 comprises the torsional pendulum drill string, the logging data and the measurement while drilling data, inputting geological parameters, logging data and measurement while drilling data acquired in the field sliding directional drilling process into the established neural network model to obtain torsional pendulum drill string control parameters.
And under the condition that the historical operation data collected in the step S01 include the torsional pendulum drill string, logging and measurement while drilling data and the empirical operation data given by the drilling staff, inputting geological parameters, path data, measurement while drilling data and the empirical operation data given by the drilling staff, which are acquired in the field sliding directional drilling process, into the established neural network model to obtain torsional pendulum drill string control parameters.
The neural network model established by the invention can determine the control parameters of the torsional pendulum drill string in real time in field production, and the neural network model established by the invention is applied to directional drilling, so that the automatic control technology of the directional drilling technology can be improved.
The invention also provides a sliding directional drilling method based on artificial intelligence, which comprises the steps of determining a torsional drill string control parameter during sliding directional drilling by adopting the method in the second exemplary embodiment during drilling, changing or maintaining the tool face direction of a downhole power drill and controlling the extension of a borehole track to a target direction.
In order that the above-described exemplary embodiments of the invention may be better understood, further description thereof with reference to specific examples is provided below.
As shown in fig. 2, a sliding directional drilling method based on artificial intelligence may include the steps of:
(1) collecting data and establishing a database;
(2) establishing a neural network model based on a database;
(3) training the neural network model;
(4) and verifying the neural network model and optimizing the neural network model to obtain an intelligent decision algorithm, and obtaining the optimal directional control parameter of the torsional pendulum drill string according to the intelligent decision algorithm.
Further, in step (1), the data in the database is derived from drilled and produced oil and gas wells, and the data at least comprises geological parameters and historical directional operation data, such as well depth, formation characteristics and the like.
Further, the historical directional operation data includes at least data on a twisted pendulum drill string system, a logging system, and a measurement-while-drilling system. Such as input torque, output torque, hook load, pump pressure, rate of penetration, weight-on-bit friction, angle, maximum torque hold time, etc.
Further, the historical directional operation data at least comprises empirical operation data given by drilling engineers in drilled and produced oil and gas well operations, such as drilling tool selection, drilling fluid performance parameters and the like.
Furthermore, after the data is collected in the step (1), the data in the database is subjected to screening operation and the extraction of data characteristics in the sliding directional drilling operation process. The interference of invalid data on the training of the neural network model is reduced by screening the data (error, invalid and repeated data), and the efficiency and accuracy of the training of the neural network model are improved.
Further, the database is divided into a test set, a validation set and a test set in proportion.
Further, the proportion range of the test set, the verification set and the test set is 5-8: 1-3: 1-2, preferably 7: 3: 2.
further, the neural network model in the step (2) at least comprises an artificial neural network model and a long-short term memory neural network model.
Further, the verifying the neural network model in the step (4) at least includes a simulation test.
Further, if the simulation test accuracy rate reaches more than 99%, the neural network model passes verification; and (4) if the simulation test accuracy is lower than 99%, optimizing the neural network model and then repeating the step (3) and the step (4).
In summary, the advantages of the method for establishing a neural network model for sliding directional drilling, the method for determining control parameters of a torsional pendulum drill string and the directional drilling method of the present invention may include:
(1) the method is simple, convenient, easy and efficient.
(2) According to the invention, a network control model and an intelligent decision algorithm are established around directional operation process data cleaning and feature extraction, and a network control model in the sliding directional drilling process is established, so that the optimal directional control parameters of the torsional pendulum drill string can be obtained, and the automation level of the directional drilling technology is improved.
(3) In the conventional directional drilling process, a group of proper control parameters needs to be selected in advance, but various complex underground working conditions are difficult to meet, and the repeated adjustment is often needed depending on the driller experience, so that the problems of low efficiency and slow response speed of a tool face exist. The invention can solve the problems and improve the self-adaptive capacity under different environments by establishing the network control model.
(4) In the neural network model training process, the database is divided into a test set, a verification set and a test set according to proportion, and the iterative neural network model is continuously updated through training of a large amount of data and algorithm optimization selection to obtain a stable and reliable control algorithm, so that the self-adaptive capacity under different environments is realized.
(5) The invention adopts a strategy of combining manual operation experience and a deep learning method, utilizes a high-level computer to carry out big data processing and neural network model training, and can realize intelligent decision of control parameters and automatic control of a directional tool surface.
(6) The invention can realize the optimization of the oil reservoir value, not only can improve the drilling efficiency, but also is beneficial to improving the automatic control level of the drill string torsional pendulum drill string system, has qualitative improvement on the drilling result of the directional well, and has great development prospect.
Although the present invention has been described above in connection with exemplary embodiments, it will be apparent to those skilled in the art that various modifications and changes may be made to the exemplary embodiments of the present invention without departing from the spirit and scope of the invention as defined in the appended claims.
Claims (7)
1. A method of modeling a neural network for sliding directional drilling, the method comprising the steps of:
collecting operation data of a drilled and produced oil and gas well and establishing a database, wherein the operation data comprises geological parameters and historical directional operation data;
establishing an initial neural network model based on a database;
training an initial neural network model;
testing the accuracy of the trained neural network model, if the accuracy is above a preset value, the trained neural network model is a sliding directional drilling neural network model, and if the accuracy is less than the preset value, continuing to train the neural network model until the accuracy reaches above the preset value;
the historical directional operation data comprises historical torsional pendulum drill string control parameters, historical logging data and historical measurement while drilling data, and the historical directional operation data further comprises artificial experience data, wherein the artificial experience data comprises at least one of drilling tool assemblies, drilling fluid performance parameters and well tracks;
the database comprises the following components in proportion of 5-8: 1-3: 1-2 of a training set, a validation set and a test set.
2. The method of building a sliding directional drilling neural network model of claim 1, wherein after building the database and before building the initial neural network model, the method further comprises the steps of:
the data in the database is screened to remove duplicate, erroneous, and invalid data.
3. The method of building a sliding directional drilling neural network model of claim 1, wherein after building the database and before building the initial neural network model, the method further comprises the steps of: and extracting the data characteristics of the historical oriented operation.
4. The method of building a sliding directional drilling neural network model of claim 1, wherein the type of initial neural network model comprises an artificial neural network model.
5. The method of modeling a sliding directional drilling neural network of claim 1, wherein the testing comprises simulation testing.
6. A method of determining a control parameter for a twisted pendulum drill string during sliding directional drilling, the method comprising the steps of:
collecting operation data of a drilled and produced oil and gas well and establishing a database, wherein the operation data comprises geological parameters, historical torsional pendulum drill string control parameters, historical logging data and historical measurement while drilling data, or the operation data comprises the geological parameters, the historical torsional pendulum drill string control parameters, the historical logging data, the historical measurement while drilling data and artificial experience data;
establishing an initial neural network model based on a database;
training an initial neural network model;
testing the accuracy of the trained neural network model, if the accuracy is above a preset value, the trained neural network model is the selected neural network model, and if the accuracy is less than the preset value, continuing to train the neural network model until the accuracy reaches above the preset value;
substituting geological parameters, logging data and measurement while drilling data obtained in the field sliding directional drilling process into the selected neural network model under the condition that the operation data does not include artificial experience data to obtain torsional pendulum drill string control parameters; substituting geological parameters, path data, measurement while drilling data and artificial experience data acquired in the field sliding directional drilling process into the selected neural network model under the condition that the operation data comprise artificial experience data to obtain torsional pendulum drill string control parameters;
wherein the artificial empirical data comprises at least one of a drilling assembly, drilling fluid performance parameters, and a wellbore trajectory;
the database comprises the following components in proportion of 5-8: 1-3: 1-2 of a training set, a validation set and a test set.
7. A method of sliding directional drilling, the method comprising: during sliding directional drilling, determining a torsional drill string control parameter by the method of claim 6, changing or maintaining the direction of a tool face of a downhole power tool, and controlling the extension of a borehole trajectory to a target direction.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910819123.0A CN110500034B (en) | 2019-08-30 | 2019-08-30 | Method for establishing neural network model, determining torsional pendulum drill string parameters and directionally drilling |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910819123.0A CN110500034B (en) | 2019-08-30 | 2019-08-30 | Method for establishing neural network model, determining torsional pendulum drill string parameters and directionally drilling |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110500034A CN110500034A (en) | 2019-11-26 |
| CN110500034B true CN110500034B (en) | 2021-03-19 |
Family
ID=68590836
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910819123.0A Active CN110500034B (en) | 2019-08-30 | 2019-08-30 | Method for establishing neural network model, determining torsional pendulum drill string parameters and directionally drilling |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110500034B (en) |
Families Citing this family (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113129157A (en) * | 2019-12-30 | 2021-07-16 | 中石化石油工程技术服务有限公司 | Underground stuck-drill fault real-time early warning method suitable for shale gas long water section |
| CN111176113B (en) * | 2019-12-31 | 2022-11-22 | 长安大学 | Drilling tool stress optimization control method based on long-time memory neural network |
| CN111520123B (en) * | 2020-03-27 | 2021-03-02 | 中国石油化工股份有限公司石油工程技术研究院 | Mechanical drilling speed prediction method, device and equipment |
| CN111810113A (en) * | 2020-08-11 | 2020-10-23 | 中国石油天然气集团有限公司 | Rotary steering automatic drilling method based on deep learning |
| CN112000651A (en) * | 2020-08-11 | 2020-11-27 | 中国石油天然气集团有限公司 | Sliding orientation data processing method |
| CN111946258B (en) * | 2020-08-11 | 2022-08-02 | 中国石油天然气集团有限公司 | GRU-based sliding orientation intelligent control method |
| CN111794741B (en) * | 2020-08-11 | 2023-08-18 | 中国石油天然气集团有限公司 | Method for realizing sliding directional drilling simulator |
| CN112065281B (en) * | 2020-09-10 | 2022-06-03 | 中国石油天然气股份有限公司 | Drilling tool combination structure parameter determination method based on composite drilling proportion target |
| CN112360341B (en) * | 2020-09-27 | 2021-12-24 | 电子科技大学 | Machine learning-based sliding directional drilling tool face state identification method |
| CN113090188B (en) * | 2021-04-26 | 2022-12-20 | 西安石油大学 | Intelligent drilling real-time borehole trajectory measurement method and system based on machine learning |
| CN113464120B (en) * | 2021-09-06 | 2021-12-03 | 中国石油集团川庆钻探工程有限公司 | Tool face state prediction method and system, and sliding directional drilling method and system |
| CN114215501A (en) * | 2022-02-23 | 2022-03-22 | 西南石油大学 | Control method for stable platform in rotary steering system |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104141482A (en) * | 2014-06-28 | 2014-11-12 | 中国石油集团渤海钻探工程有限公司 | Directional sliding drilling monitoring while drilling control device and control method of directional sliding drilling monitoring while drilling control device |
| CN106050216A (en) * | 2016-06-02 | 2016-10-26 | 中国石油天然气集团公司 | Top drive twisting drag reduction method and device capable of improving sliding drilling efficiency |
| CN106437513A (en) * | 2016-09-26 | 2017-02-22 | 中国石油大学(华东) | Complex structural well friction resistance reduction and power drill tool face adjusting method |
| CN107013163A (en) * | 2017-04-10 | 2017-08-04 | 青岛金江源工业装备有限公司 | A kind of directional drilling, the electric rotary disc device and its control method of anti-backing pressure |
| CN109869131A (en) * | 2019-03-26 | 2019-06-11 | 敖江昵 | A kind of micro- compound deflecting method and device of directional well |
-
2019
- 2019-08-30 CN CN201910819123.0A patent/CN110500034B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104141482A (en) * | 2014-06-28 | 2014-11-12 | 中国石油集团渤海钻探工程有限公司 | Directional sliding drilling monitoring while drilling control device and control method of directional sliding drilling monitoring while drilling control device |
| CN106050216A (en) * | 2016-06-02 | 2016-10-26 | 中国石油天然气集团公司 | Top drive twisting drag reduction method and device capable of improving sliding drilling efficiency |
| CN106437513A (en) * | 2016-09-26 | 2017-02-22 | 中国石油大学(华东) | Complex structural well friction resistance reduction and power drill tool face adjusting method |
| CN107013163A (en) * | 2017-04-10 | 2017-08-04 | 青岛金江源工业装备有限公司 | A kind of directional drilling, the electric rotary disc device and its control method of anti-backing pressure |
| CN109869131A (en) * | 2019-03-26 | 2019-06-11 | 敖江昵 | A kind of micro- compound deflecting method and device of directional well |
Non-Patent Citations (2)
| Title |
|---|
| 神经网络钻井参数优化;赵毅;《中国优秀硕士学位论文全文数据库》;20180815(第08期);B019-8 * |
| 赵毅.神经网络钻井参数优化.《中国优秀硕士学位论文全文数据库》.2018,(第08期),B019-8. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110500034A (en) | 2019-11-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110500034B (en) | Method for establishing neural network model, determining torsional pendulum drill string parameters and directionally drilling | |
| CA2789219C (en) | System and method for optimizing drilling speed | |
| CA2977943C (en) | Model generation for real-time rate of penetration prediction | |
| RU2643057C2 (en) | Managing wellbore exploitation with use of uncertainty calculations | |
| US9995129B2 (en) | Drilling automation using stochastic optimal control | |
| AU2013406722B2 (en) | Controlling a bottom hole assembly in a wellbore | |
| CN110500081B (en) | Automatic drilling method based on deep learning | |
| NO174305B (en) | Procedure for ae predetermining a drill bit's drilling path or ae deriving an instruction of anisotropy index for the drill bit in directional drilled wells | |
| CN113090188B (en) | Intelligent drilling real-time borehole trajectory measurement method and system based on machine learning | |
| CN109557584A (en) | A kind of horizontal drilling Real-time Seismic geological syntheses guidance method | |
| CN116451013B (en) | Deep stratum rock in-situ drillability grade value prediction method | |
| CN107861917A (en) | Method of the multivariate data combined calculation well to stratigraphic boundary distance in horizontal well | |
| Etesami et al. | A semiempirical model for rate of penetration with application to an offshore gas field | |
| CN115628009A (en) | Intelligent borehole trajectory control method based on reinforcement learning | |
| NO20210101A1 (en) | Automated production history matching using bayesian optimization | |
| CN114764844B (en) | Shale gas drilling optimization design method and device | |
| US10240414B2 (en) | Regulating downhole fluid flow rate using an multi-segmented fluid circulation system model | |
| CN112901070B (en) | Method for tracking horizontal well while drilling and control method for controlling drilling direction of drill bit | |
| Kuswana et al. | Challenges and Considerations in A Tripping Operation: The Need for An Advanced Control | |
| US20230175383A1 (en) | System and method for automated identification of mud motor drilling mode | |
| Kolmer et al. | ROP Optimization of Lateral Wells in SW Oklahoma: Artificial Neural Network Approach | |
| Mayibeki | Drilling Optimization of a Caney Shale Well Using Offset Well Drilling Data | |
| Ma et al. | Deep and horizontal drilling technologies for natural gas | |
| Nascimento | Modelamento matemático para otimização de perfuração em seções de pré-pal: um foco em operações no oceano Atlântico sul | |
| Lecourtier | Interactive drilling for fast track oilfield development: proceedings of the seminar held in Rueil-Malmaison, November 9, 1999 |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |