Drilling operations telemetry framework

Abstract

A method can include receiving, using equipment at a site, data of a live operation, wherein the equipment includes a downhole tool on a tool string disposed in a wellbore in a geological environment and a mud pulse telemetry system; determining, using at least a portion of the data and a trained machine learning model, a control parameter for the mud pulse telemetry system; and controlling, using the control parameter, the mud pulse telemetry system.

Description

Cross-references to related applications

[0001] This disclosure claims priority to U.S. Provisional Application No. 63 / 558320, filed February 27, 2024, which is incorporated herein by reference in its entirety. Background Technology

[0002] A resource site can be an accumulation, pool, or group of pools of one or more resources (e.g., oil, gas, or a mixture of oil and gas) in a subsurface environment. A resource site may include at least one reservoir. The reservoir may be shaped in a manner capable of trapping hydrocarbons and may be covered by impermeable or sealed rock. A borehole may be drilled in the environment, which can be used to form a well for extracting hydrocarbons from the reservoir.

[0003] A drilling rig can be a system of components that can be operated to form boreholes in an environment, transport equipment into and out of the boreholes in the environment, etc. As an example, a drilling rig may include systems that can be used to drill boreholes and obtain information about the environment, about the well, etc. A resource site can be an onshore site, an offshore site, or an onshore and offshore site. A drilling rig may include components for performing operations onshore and / or offshore. A drilling rig may be, for example, ship-based, offshore platform-based, onshore, etc.

[0004] Site planning can be carried out in one or more phases, which may include an exploration phase aimed at identifying and assessing the environment (e.g., target traps, zones, etc.), which may include drilling one or more boreholes (e.g., one or more exploratory wells, etc.). Other phases may include evaluation, development, and mining phases. Summary of the Invention

[0005] A method may include: using on-site equipment to receive data from field operations, wherein the equipment includes downhole tools and a mud pulse telemetry system on a tool string disposed in a wellbore within a geological environment; using at least a portion of the data and a trained machine learning model to determine control parameters for the mud pulse telemetry system; and using the control parameters to control the mud pulse telemetry system. A system may include: a processor; a memory accessible by the processor; and processor-executable instructions stored in and executable in the memory to instruct the system to: use on-site equipment to receive data from field operations, wherein the equipment includes downhole tools and a mud pulse telemetry system on a tool string disposed in a wellbore within a geological environment; using at least a portion of the data and a trained machine learning model to determine control parameters for the mud pulse telemetry system; and using the control parameters to control the mud pulse telemetry system. One or more computer-readable storage media may include processor-executable instructions to instruct a computing system to: receive data from field operations using equipment at the site, wherein the equipment includes downhole tools and a mud pulse telemetry system mounted on a tool string in a wellbore within the geological environment; determine control parameters for the mud pulse telemetry system using at least a portion of the data and a trained machine learning model; and control the mud pulse telemetry system using the control parameters. Various other devices, systems, methods, etc., are also disclosed.

[0006] This summary is provided to introduce a series of concepts further described below in the detailed description. This summary is not intended to identify key or essential features of the claimed subject matter, nor is it intended to be used as an aid to limit the scope of the claimed subject matter. Attached Figure Description

[0007] The features and advantages of the described implementation can be more easily understood by referring to the following description in conjunction with the accompanying drawings.

[0008] Figure 1 Examples of equipment in geological environments are shown; Figure 2 Examples of equipment and well types are shown; Figure 3 Examples of mud pulse equipment and technology are shown; Figure 4 An example of a graph is shown; Figure 5 An example of a graph is shown; Figure 6 An example of a graph is shown; Figure 7 An example of the model is shown; Figure 8 An example of a portion of the model is shown; Figure 9 An example of a lexicalization architecture is shown; Figure 10A and Figure 10B Example components showing an example of a workflow; Figure 11 An example of the workflow is shown; Figure 12 An example of the system is shown; Figure 13 Examples of methods and systems are shown; and Figure 14 An example of a computing system is shown. Detailed Implementation

[0009] The following description includes the best mode currently contemplated for practical application of the described implementations. This description should not be construed as restrictive, but merely as an attempt to describe the general principles of the implementations. The scope of the described implementations should be determined with reference to the published claims.

[0010] Figure 1 An example of geological environment 120 is shown. Figure 1 In this context, the geological environment 120 may be a sedimentary basin comprising multiple layers (e.g., stratified), including reservoir 121 and intersecting, for example, fault 123 (e.g., or multiple faults). As an example, the geological environment 120 may be equipped with various sensors, detectors, actuators, etc. For example, equipment 122 may include a communication circuitry for receiving and transmitting information relative to one or more networks 125. Such information may include information associated with downhole equipment 124, which may be equipment used for information acquisition, assisting resource recovery, etc. Other equipment 126 may be located remotely from the well site and include sensing circuitry, detection circuitry, transmitting circuitry, or other circuitry. Such equipment may include storage and communication circuitry for storing and transmitting data, instructions, etc. As an example, one or more pieces of equipment may provide the measurement, collection, communication, storage, analysis, etc., of data (e.g., regarding one or more mined resources). As an example, one or more satellites may be provided for communication, data acquisition, and other purposes. For example, Figure 1 A satellite communicating with a network 125 that can be configured for communication is shown. Note that the satellite may additionally or alternatively include circuitry for imaging (e.g., space, spectrum, time, radiation, etc.).

[0011] Figure 1The geological environment 120 is also shown as optionally including equipment 127 and 128 associated with a well, which includes a substantially horizontal portion that may intersect with one or more fractures 129. For example, consider a well in a shale formation that may include natural fractures, artificial fractures (e.g., hydraulic fractures), or a combination of natural and artificial fractures. As an example, a well may be drilled for a laterally extending reservoir. In such examples, there may be lateral variations in properties, stresses, etc., where assessment of such variations can aid in planning, operations, etc., to develop the reservoir (e.g., via fracturing, injection, extraction, etc.). As an example, equipment 127 and / or 128 may include components, one or more systems, etc., for fracturing, seismic sensing, seismic data analysis, assessment of one or more fractures, injection, production, etc. As an example, equipment 127 and / or 128 may provide, for example, the measurement, collection, communication, storage, and analysis of data such as production data (e.g., regarding one or more extracted resources). As an example, one or more satellites may be provided for communication, data acquisition, and other purposes.

[0012] Figure 1 Examples of equipment 170 and equipment 180 are also shown. Such equipment (which may be a system of components) can be adapted to geological environment 120. Although equipment 170 and 180 are shown as land-based, various components can be adapted to marine systems.

[0013] Equipment 170 includes a platform 171, a derrick 172, a crane 173, a wire rope 174, a traveling block assembly 175, a winch 176, and a loading / unloading platform 177 (e.g., a second-level platform). As an example, the wire rope 174 can be controlled at least partially via the winch 176, causing the traveling block assembly 175 to travel vertically relative to the platform 171. For example, by winding in the wire rope 174, the winch 176 can move the wire rope 174 through the crane 173 and away from the platform 171 to raise the traveling block assembly 175; while by releasing the wire rope 174, the winch 176 can move the wire rope 174 through the crane 173 and lower the traveling block assembly 175 toward the platform 171. When the traveling block assembly 175 carries drill pipe (e.g., casing, etc.), tracking the movement of the traveling block 175 can provide an indication of how much drill pipe has been deployed.

[0014] A derrick can be a structure used to support an overhead crane and a traveling block that is at least partially operatively connected to the overhead crane via wire ropes. The derrick can be pyramidal in shape and provide a suitable strength-to-weight ratio. The derrick can be moved as a unit or in parts (e.g., it can be assembled and disassembled).

[0015] As an example, a winch may include a drum, brakes, a power supply, and various auxiliary devices. The winch can controllably release and wind up a wire rope. The wire rope can be wound onto a crane and connected to a traveling block to obtain mechanical advantages in a "pulley" or "roller" configuration. Releasing and winding up the wire rope allows the traveling block (e.g., and anything that may be suspended below it) to be lowered into or raised from the borehole. The release of the wire rope can be powered by gravity, while the winding of the wire rope can be powered by a motor, engine, etc. (e.g., an electric motor, a diesel engine, etc.).

[0016] As an example, a crane may include a pulley (e.g., Geneva) assembly located at or near the top of the derrick or rig, through which a wire rope passes. A traveling block may include a Geneva assembly that can move up and down within the derrick or rig via a wire rope passing through both the Geneva assembly of the traveling block and the Geneva assembly of the crane. The crane, traveling block, and wire rope can form a pulley system for the derrick or rig, enabling the handling of heavy loads (e.g., drill string, drill pipe, casing, liner, etc.) to be lifted out of or lowered into the borehole. As an example, the wire rope diameter can be from about one centimeter to about five centimeters, such as steel cable. By using Geneva assemblies, such wire ropes can carry heavier loads than a single-strand wire rope can support.

[0017] As an example, a derrickman can be a member of the drilling team working on a platform attached to a derrick or rig. The derrick may include a loading platform on which the derrickman can stand. As an example, such a loading platform may be located approximately 10 meters or higher above the rig. In an operation known as tripping (TOH), the derrickman may wear a harness that allows him to tilt outward from the loading platform (e.g., a second-level platform) to reach the drill pipe located at or near the center of the derrick or rig, and to wind a wire rope around the drill pipe and pull it back to its storage position (e.g., a finger beam), for example, until the time when it may be expected to lower the drill pipe into the borehole. As an example, the drilling rig may include automated drill pipe handling equipment, allowing the derrickman to control the machinery rather than handle the drill pipe manually.

[0018] As an example, tripping in and out of the borehole can refer to the actions of removing equipment from the borehole and / or running equipment into the borehole. As an example, equipment may include drill strings that can be retrieved from and / or run into or replaced in the wellbore. As an example, tripping in and out of the drill pipe may be performed when the drill bit has become dull or has otherwise ceased to drill efficiently and needs to be replaced.

[0019] Figure 2An example of a well site system 200 is shown (e.g., at a well site that may be located onshore or offshore). As shown, the well site system 200 may include: a mud tank 201 for storing mud and other materials (e.g., where the mud may be drilling fluid); a suction line 203 serving as the inlet of a mud pump 204 for pumping mud from the mud tank 201 to a vibratory hose 206; a winch 207 for hoisting one or more drilling wire ropes 212; a riser 208 for receiving mud from the vibratory hose 206; a square drill pipe hose 209 for receiving mud from the riser 208; one or more gooseneck pipes 210; a traveling block 211; and a crane 213 for carrying the traveling block 211 via one or more drilling wire ropes 212 (e.g., see...). Figure 1 Overhead crane 173); derrick 214 (see, for example, see...) Figure 1 The derrick 172; 218 or top drive 240; 219; rotary table 220; drill platform 221; bell-shaped sub 222; one or more blowout preventers (BOPs) 223; drill string 225; drill bit 226; casing head 227; and flow pipe 228 for carrying mud and other materials to, for example, mud tank 201.

[0020] exist Figure 2 In the example system, a wellbore 232 is formed in the underground formation 230 by rotary drilling; note that directional drilling can also be used in various example implementations.

[0021] like Figure 2 As shown in the example, drill string 225 is suspended within wellbore 232 and has drill string assembly 250, which includes drill bit 226 at its lower end. As an example, drill string assembly 250 may be a bottom hole assembly (BHA).

[0022] Well site system 200 can provide operation of drill string 225 and other operations. As shown, well site system 200 includes platform 211 and derrick 214 positioned above wellbore 232. As mentioned, well site system 200 may include rotary table 220 through which drill string 225 passes.

[0023] like Figure 2As illustrated in the example, well site system 200 may include a crisscross drill pipe 218 and associated components, or a top drive 240 and associated components. Regarding the crisscross drill pipe example, the crisscross drill pipe 218 may be a square or hexagonal metal / alloy rod with holes drilled to serve as a mud flow path. The crisscross drill pipe 218 can be used to transmit rotational motion from rotary table 220 to drill string 225 via a crisscross drill pipe insert 219, while allowing drill string 225 to be lowered or raised during rotation. The crisscross drill pipe 218 may pass through a crisscross drill pipe insert 219 that can be driven by rotary table 220. As an example, rotary table 220 may include a main insert operatively coupled to the crisscross drill pipe insert 219, such that rotation of rotary table 220 can rotate the crisscross drill pipe insert 219 and thus rotate the crisscross drill pipe 218. The square drill pipe filler 219 may include an internal profile that matches the external profile (e.g., square, hexagonal, etc.) of the square drill pipe 218; however, it has a slightly larger size so that the square drill pipe 218 can move freely up and down within the square drill pipe filler 219.

[0024] Regarding the example of the top drive, top drive 240 can provide functions performed by the kelly and rotary table. Top drive 240 can rotate drill string 225. As an example, top drive 240 may include one or more motors (e.g., electric motors and / or hydraulic motors) connected via suitable transmissions to a short section of tubing known as a hollow shaft, which can then be screwed into a protective joint or the drill string 225 itself. Top drive 240 can be suspended from traveling block 211, so the rotating mechanism can move freely up and down along derrick 214. As an example, top drive 240 may allow drilling to be performed using more single stands than the kelly / rotary table method.

[0025] exist Figure 2 In the example, mud tank 201 can store mud, which can be one or more types of drilling fluid. As an example, a wellbore can be drilled to extract fluids, inject fluids, or both (e.g., hydrocarbons, minerals, water, etc.).

[0026] exist Figure 2In the example, drill string 225 (e.g., including one or more downhole tools) may consist of a series of drill pipes threaded together to form a long tubing, with drill bit 226 at its lower end. When drill string 225 is advanced into the wellbore for drilling, before or at some point coinciding with drilling, mud may be pumped from mud tank 201 (e.g., or from another source) to a port of crisscross drill pipe 218, or, for example, to a port of top drive 240, via lines 206, 208, and 209 by pump 204. The mud may then flow through channels (e.g., or multiple channels) in drill string 225 and exit from a port located on drill bit 226 (see, for example, directional arrows). As the mud exits drill string 225 through the port in drill bit 226, it may then circulate upwards through an annular region (e.g., an open wellbore, casing, etc.) between the outer surface of drill string 225 and the surrounding wall, as indicated by the directional arrows. In this manner, the mud lubricates the drill bit 226 and carries heat energy (e.g., friction or other energy) and formation cuttings to the surface, where the mud (e.g., and the cuttings) can be returned to the mud tank 201 for example for recycling (e.g., by treatment to remove cuttings, etc.).

[0027] The drilling mud pumped into the drill string 225 by pump 204 forms a mud cake lining the wellbore after leaving the drill string 225. This mud cake, among other functions, reduces friction between the drill string 225 and the surrounding walls (e.g., the wellbore, casing, etc.). This reduction in friction facilitates the advancement or retraction of the drill string 225. During drilling operations, the entire drill string 225 can be retrieved from the wellbore and optionally replaced, for example, with a new or sharper drill bit, a smaller diameter drill string, etc. As mentioned, the act of retrieving the drill string from the wellbore or replacing the drill string within the wellbore is called tripping in and out. Depending on the direction of tripping in and out, tripping in and out can be referred to as tripping up, tripping outward, tripping down, or tripping inward.

[0028] As an example, consider drilling downwards, where, as the drill bit 226 of the drill string 225 reaches the bottom of the wellbore, the pumping of mud begins to lubricate the drill bit 226 to achieve the purpose of drilling and enlarging the wellbore. As mentioned, the mud can be pumped into the channels of the drill string 225 by pump 204, and while filling the channels, the mud can be used as a medium for transmitting energy (e.g., energy that can encode information, as in mud pulse telemetry).

[0029] As an example, mud pulse telemetry equipment may include a downhole device configured to realize changes in mud pressure to generate one or more acoustic waves, which can be used to modulate information. In such an example, information from downhole equipment (e.g., one or more modules of drill string 225) can be transmitted to a surface unit, which can then forward this information to other equipment for processing, control, etc.

[0030] As an example, telemetry equipment can operate by transmitting energy via the drill string 225 itself. For instance, consider a signal generator that transmits encoded energy signals to the drill string 225 and a repeater that can receive and relay such energy for further transmission of encoded energy signals (e.g., information, etc.).

[0031] As an example, drill string 225 may be equipped with: a telemetry device 252 including a rotatable drive shaft; a turbine impeller mechanically coupled to the drive shaft such that the drilling mud can rotate the turbine impeller; a modulator rotor mechanically coupled to the drive shaft such that the rotation of the turbine impeller causes the modulator rotor to rotate; a modulator stator mounted adjacent to or near the modulator rotor such that the rotation of the modulator rotor relative to the modulator stator generates pressure pulses in the drilling mud; and a controllable brake for selectively braking the rotation of the modulator rotor to modulate the pressure pulses. In such an example, an alternator may be coupled to the aforementioned drive shaft, wherein the alternator includes at least one stator winding electrically coupled to a control circuit to selectively short-circuit the at least one stator winding to electromagnetically brake the alternator and thereby selectively brake the rotation of the modulator rotor to modulate the pressure pulses in the drilling mud.

[0032] exist Figure 2 In one example, the wellhead control and / or data acquisition system 262 may include a circuit system for sensing pressure pulses generated by the telemetry equipment 252 and, for example, transmitting the sensed pressure pulses or information derived therefrom for processing, control, etc.

[0033] The component 250 shown in the example includes a logging-while-drilling (LWD) module 254 (e.g., an LWD tool), a measurement-while-drilling (MWD) module 256 (e.g., a MWD tool), an optional module 258, a rotary steerable system (RSS) and / or a motor 260, and a drill bit 226. Such components or modules may be referred to as tools, and the drill string may include multiple tools.

[0034] Regarding RSS, it relates to techniques used in directional drilling. Directional drilling involves drilling into the earth to form an skewed borehole, such that the borehole's trajectory is not vertical; instead, the trajectory deviates from the vertical along one or more sections of the borehole. As an example, consider a target located at a lateral distance from a surface location where a drilling rig might be stationed. In such an example, drilling could begin with a vertical section and then deviate from the vertical, such that the borehole aligns with and eventually reaches the target. Directional drilling can be achieved when a target is not reachable from a vertical position on the earth's surface, when there is material on the earth that could hinder drilling or other harmful elements (e.g., consider salt domes), when the formation extends laterally (e.g., consider relatively thin but laterally extending reservoirs), when multiple boreholes are to be drilled from a single surface borehole, and when a decompression well is desired, among other things.

[0035] One method of directional drilling involves mud motors; however, mud motors face several challenges depending on factors such as the rate of penetration (ROP) and the transfer of weight to the drill bit due to friction (e.g., weight on bit, WOB). A mud motor can be a positive displacement motor (PDM) that operates to drive the drill bit (e.g., during directional drilling). The PDM operates as drilling fluid is pumped through it, converting the hydraulic power of the drilling fluid into mechanical power that causes the drill bit to rotate.

[0036] As an example, PDM can operate in a combined rotary mode, where surface equipment (e.g., rotary table, top drive, etc.) is used to rotate the drill string bit by rotating the entire drill string, and drilling fluid is used to rotate the drill string bit. In such an example, the surface RPM (SRPM) can be determined using surface equipment, and the downhole RPM of the mud motor can be determined using various factors related to drilling fluid flow, mud motor type, etc. As an example, in combined rotary mode, the bit RPM can be determined or estimated as the sum of the SRPM and the mud motor RPM, assuming that the SRPM and the mud motor RPM are in the same direction.

[0037] As an example, when the drill string is not rotated from the ground to drive the drill bit in a specific cutting direction, the PDM mud motor can operate in a so-called slippery mode. In such examples, the drill bit RPM can be determined or estimated based on the mud motor's RPM. As an example, during slippery mode, the drill string can be oscillated by ground equipment, for example, to make the drill string oscillate in clockwise and counterclockwise directions, which can, for example, help reduce the risk of stuck drill bit.

[0038] RSS (Rotating Steering Module) can be used in directional drilling where there is continuous rotation from surface equipment, which can mitigate slippage of the directional motor (e.g., PDM). RSS can be deployed during directional drilling (e.g., deviated, horizontal, or extended wells). RSS can be designed to minimize its interaction with the wellbore wall, which can help maintain wellbore quality. RSS can be designed to apply a relatively consistent lateral force similar to a stabilizer that rotates with the drill string or orients the drill bit in the desired direction while rotating continuously at the same rate of revolutions per minute as the drill string.

[0039] LWD module 254 can be housed in a suitable type of drill collar and can contain one or more selected types of logging tools. It will also be understood that more than one LWD and / or MWD module can be employed. When referring to the location of a module, by way of example, it can refer to the module at the location of LWD module 254, MWD module 256, etc. An LWD module can include capabilities for measuring, processing, and storing information, as well as for communicating with surface equipment. In the example shown, LWD module 254 can include a seismic measurement device.

[0040] MWD module 256 can be housed in a suitable type of drill collar and may include one or more devices for measuring the characteristics of drill string 225 and drill bit 226. As an example, MWD module 256 may include equipment for generating electricity, for example, to power various components of drill string 225. As an example, MWD module 256 may include telemetry equipment 252, for example, where a turbine impeller can generate electricity through the flow of mud; it should be understood that other power sources and / or battery systems may be used to power various components. As an example, MWD module 256 may include one or more measuring devices of the following types: drill pressure measuring device, torque measuring device, vibration measuring device, impact measuring device, stick-slip measuring device, direction measuring device, and inclination angle measuring device.

[0041] Figure 2 Some examples of the types of wells that can be drilled are also shown. For example, consider 272, 274, 276, and 278, respectively.

[0042] As an example, drilling operations may include directional drilling, where, for example, at least a portion of the well includes a curved axis. For instance, consider a radius defining the curvature, where the angle of inclination relative to the vertical direction can vary up to an angle between approximately 30 degrees and approximately 60 degrees, or, for example, an angle of approximately 90 degrees or possibly greater than approximately 90 degrees.

[0043] As an example, a directional well may include several shapes, each of which may be designed to meet specific operational requirements. As an example, the drilling process may be based on information and performed when the information is communicated to the drilling engineer. As an example, the inclination angle and / or direction may be modified based on information received during the drilling process.

[0044] As an example, borehole deflection can be achieved in part by using one or more of an RSS, a downhole motor, and / or a turbine. Regarding motors, for example, the drill string can include a positive displacement motor (PDM).

[0045] As an example, the system could be a guidance system and include equipment for performing methods such as geological guidance. As an example, the guidance system could include a PDM or turbine located on the lower part of the drill string, just above the drill bit, where a bend can be mounted. As an example, above the PDM, a MWD (Multi-Wave Detection) and / or LWD (Low-Wave Detection) device could be mounted, which provides real-time or near-real-time data of interest (e.g., inclination angle, direction, pressure, temperature, actual drill weight, torque stress, etc.). Regarding the latter, the LWD device could enable the transmission of various types of data of interest to the surface, including, for example, geological data (e.g., gamma-ray logging, resistivity, density, and sonic logging, etc.).

[0046] The coupling of sensors that provide real-time or near-real-time information about well trajectory with one or more well logs, for example, characterizing formations from a geological perspective, can enable geoguided methods. Such methods can include navigating subsurface environments, for example, to follow a desired route to reach one or more desired targets.

[0047] As an example, a drill string may include: an azimuth density neutron (ADN) tool for measuring density and porosity; a MWD tool for measuring tilt angle, azimuth angle, and impact; a compensated dual resistivity (CDR) tool for measuring resistivity and gamma-ray related phenomena; one or more variable diameter stabilizers; one or more bend joints; and a geological guidance tool that may include a motor and, optionally, equipment for measuring and / or responding to one or more of tilt angle, resistivity, and gamma-ray related phenomena.

[0048] As an example, geological steering can include intentional directional control of the wellbore based on the results of downhole geological logging measurements, in a manner aimed at keeping the directional wellbore within a desired area, zone (e.g., oil-producing layer), etc. As an example, geological steering can include guiding the wellbore to keep it within a specific section of the reservoir, for example, to minimize gas and / or water breakthroughs, and for example, to maximize economic production from the well including the wellbore.

[0049] Refer again Figure 2The well site system 200 may include one or more sensors 264 operatively coupled to a control and / or data acquisition system 262. As an example, the one or more sensors may be at a surface location. As an example, the one or more sensors may be at a downhole location. As an example, the one or more sensors may be located at one or more remote locations not within approximately one hundred meters of the well site system 200. As an example, the one or more sensors may be located at a compensation well site, wherein the well site system 200 and the compensation well site are located in a common oil and gas field (e.g., an oil field and / or a gas field).

[0050] As an example, one or more of the sensors 264 may be provided for tracking the movement of the drill pipe, tracking the movement of at least a portion of the drill string, etc.

[0051] As an example, system 200 may include one or more sensors 266 that can sense signals and / or transmit signals to fluid conduits, such as drilling fluid conduits (e.g., drilling mud conduits). For example, in system 200, one or more sensors 266 may be operatively coupled to a portion of riser 208 through which mud flows. As an example, a downhole tool may generate pulses that can travel through the mud and be sensed by one or more of the sensors 266. In such examples, the downhole tool may include associated circuitry, such as, for example, coded circuitry that can encode signals to reduce the need for transmission. As an example, surface circuitry may include decoding circuitry to decode encoded information transmitted at least partially via mud pulse telemetry. As an example, surface circuitry may include encoder and / or decoder circuitry, and downhole circuitry may include encoder and / or decoder circuitry. As an example, system 200 may include a transmitter that can generate signals that can be transmitted downhole via mud (e.g., drilling fluid) as a transmission medium.

[0052] As an example, one or more portions of the drill string may become stuck. The term "stuck" can refer to one or more different degrees of inability to move or remove the drill string from the borehole. As an example, in a stuck condition, it may be possible to rotate the drill string or lower it back into the borehole, or, for example, in a stuck condition, it may be impossible to move the drill string axially in the borehole, although some rotation may be possible. As an example, in a stuck condition, it may be impossible to move at least a portion of the drill string both axially and rotationally.

[0053] The term "stuck" can refer to a portion of the drill string that cannot be rotated or moved axially. As an example, a condition known as "differential stuck" can be a situation where the drill string cannot move along the axis of the borehole (e.g., rotate or reciprocate). Differential stuck can occur when high contact forces caused by low reservoir pressure, high wellbore pressure, or both are applied to a sufficiently large area of ​​the drill string. Differential stuck can have both time and economic costs.

[0054] As an example, adhesion force can be the product of the pressure differential between the wellbore and the reservoir and the area over which the pressure differential acts. This means that applying a relatively low pressure differential (Δp) over a large working area can have the same effect on stuck pipe as applying a high pressure differential over a small area.

[0055] As an example, a situation known as "mechanical stuck" can be one in which the movement of the drill string is restricted or prevented using mechanisms other than differential pressure stuck. Mechanical stuck can be caused by one or more of the following: debris in the wellbore, abnormal wellbore geometry, cement, keyways, or cuttings buildup in the annulus.

[0056] Various types of data associated with field operations can be 1-D sequence data. For example, data on one or more of the drilling system, downhole conditions, formation properties, and surface mechanics can be measured as single-channel or multi-channel time series data.

[0057] Mud pulse telemetry (MPT) can provide the transmission of information related to downhole parameters and / or conditions that may relate to formation characteristics traversed along the wellbore. As an example, LWD technology can employ an MPT system. As another example, an MPT system can use a communication channel formed by the drilling fluid column in the hole of the drill pipe to transmit signals, for example, using a downhole transmitter and a surface receiver (e.g., and / or vice versa).

[0058] As an example, the equipment at the site may include dual telemetry control equipment that can provide an uplink (e.g., transmission of sensor data and / or other information to the surface) and a downlink, wherein the downlink can, for example, provide commands to downhole equipment (e.g., switching between telemetry modes, configuring telemetry, commanding one or more directional drilling tools, commanding one or more data acquisition tools, etc.). As an example, the downlink may include a temporary interruption of drilling operations to allow operation of one or more mud pumps at the surface to generate pulses that are detectable using one or more downhole receivers. While mud pumps are mentioned, one or more other techniques may be employed. For example, consider a downlink system including at least one mud pump for pumping drilling fluid from a drilling fluid storage tank to the drilling system via a riser, wherein a return line is provided for returning the fluid to the storage tank. In such examples, a drilling fluid modulator may be in fluid communication with the riser or return line. As an example, a flow modulator may be placed on a bypass line. As an example, the downlink system may include a flow restrictor that can provide resistance to high-pressure mud flow and thus help protect the flow modulator. In such examples, the downlink system may include a shunt, which can help reduce turbulence in the mud flow.

[0059] As explained, drilling fluids can be used to lubricate and cool downhole drill string components, carry cuttings from the bottom of the wellbore to the surface, and balance static pressure in the formation. As an example, an MPT system can use encoded drilling fluid pressure signal pulses generated using downhole equipment, which are propagated through the drilling fluid (e.g., mud) to surface equipment, where they can be detected and decoded for interpretation; note that a downlink, where available, can provide mud pulse transmissions that can be detected by the downhole equipment. In various cases, the bandwidth requirement for uphole transmissions to the surface can be greater than the bandwidth requirement for downhole transmissions to downhole tools. For example, uphole transmissions can be used for one or more types of sensor data that can be acquired and transmitted and / or stored and transmitted later at a desired frequency. Such data can be evaluated to determine one or more conditions that can provide improved control over field operations (e.g., drilling operations). In various cases, the downlink can be a control command, which (e.g., a command relatively lightweight in magnitude compared to sensor data)

[0060] As an example, an MPT system can utilize valves to controllably restrict the flow of drilling fluid, which in turn generates pressure waves that can travel within a continuous column of drilling fluid (e.g., within the borehole of the drill string). In such examples, the pressure waves can travel at varying speeds, depending on one or more factors, such as, for example, the properties of the drilling fluid. As an example, controlled downhole pressure pulse variations can be used to modulate the amplitude, phase, and / or frequency of mud pulse signals. As an example, surface-received signals can respond to pulse variations associated with downhole measurement data (e.g., LWD data, etc.).

[0061] Figure 3 Examples of portions of MPT systems 310, 320, and 330 that can be used to generate mud pulses are shown. For example, systems 310, 320, and 330 can use techniques for mud pulse pressure pulses to generate one or more flows, such as a positive pulse for each MPT system 310, a negative pulse for each MPT system 320 (e.g., annular pressure relief telemetry), and a continuous wave (e.g., mud alarm) wave for each MPT system 330.

[0062] As an example, a telemetry system may include links; for instance, consider a two-link system with three nodes, where the middle node (e.g., an intermediate node) can be considered a repeater. As another example, the framework can provide planning and / or control operations for single-link and / or multi-link telemetry systems.

[0063] As an example, in a telemetry system, the pressure pulses generated by the lift valve pulse generator can form discrete pressure waves, including positive and negative pulses (annular pressure relief telemetry); the alarm pulse generator can generate periodic continuous wave pulses; and the rotating oscillating shear valve pulse generator can generate discrete pulse and / or continuous wave signals. Such mud pulsar sequences generated can be identified by ground signal processing equipment.

[0064] The transmitted pulse signals (including information) can be encoded using one or more of various techniques, such as pre-encoding before they propagate to the surface for detection and decoding. Surface pressure transmitters often detect signals weakened by signal strength degradation (attenuation), erosion, pressure drops across the wellbore, and / or one or more other signal attenuation effects. As explained, the received information may include, for example, logging data, pressure parameters, temperature, drill bit orientation, deviation, etc. As explained, logging data may include various formation conductivity, acoustic and nuclear properties, porosity, induction, pressure gradient, etc. To improve drilling operations, MPT systems may include various features to recover the transmitted signals with acceptable fidelity.

[0065] As an example, one or more types of MPT system signal degradation may occur. For instance, mud channel signals may experience pressure fluctuations and uncertainties caused by various MPT system components that potentially alter signal properties during data transmission. Noise sources causing pressure fluctuations may include one or more of the following: mud pumps, pulsation dampers, surface piping, pressure transmitter location, drill string components, mud properties, well depth, etc. In various cases, the mud channel through which pulse signals travel as they propagate to the surface can be relatively harsh, necessitating complex surface signal detection and extraction. Mud pulse signal property distortions that can degrade signal transmission and data rates can be related to downhole signal strength, signal attenuation, surface-induced noise (electrical noise), and signal reflections caused by surface piping. Drilling noise, drill string movement noise within the wellbore, attenuation, and mud pump noise can lead to unpredictable and complex adjustable signal degradation. Depending on mud channel conditions and the severity of signal distortion, telemetry systems can be configured to be robust and flexible downhole and / or at the surface to achieve the maximum possible data rate.

[0066] As an example, MPT systems can utilize frequencies ranging from approximately 1 Hz to approximately 20 Hz (e.g., applicable to various LWD MPT systems). Mud pump noise can interfere with these frequencies in various situations. Furthermore, as the MPT system signal propagates along the wellbore through the drilling mud, the signal can be attenuated and dispersed. For example, consider underbalanced drilling mud, which can lead to viscous dissipation and frictional energy loss at the wellbore wall. Variations in the compressibility of the drilling mud (e.g., due to gas introduction, gas release, etc.) can increase the risk of formation damage, which can affect the shape of the signal pulse and complicate decoding. Factors affecting mud pulse signal attenuation can include wellbore depth, drilling mud type, number of joints in the drill string, drill string inner diameter, and signal operating frequency. The attenuation effect can be different across a range of frequency components; for example, lower frequency components may experience less attenuation than higher frequencies. Good control of various noise sources in the 1 Hz to 20 Hz band can help transmit MPT data at the lowest possible frequency to minimize signal attenuation effects. In some high data rate exploration and drilling operations, signal attenuation can be less burdensome than the various noise sources and their distribution across the available channel transmission band.

[0067] In various scenarios, pulsars in MPT telemetry can generate signals that also propagate in the opposite direction, causing downlink signals to be reflected at the drill bit and then combined to produce constructive or destructive interference signals. Signal reflections caused by one or more of the following—mud pumps, pulsation dampers, drill string joints and diameter changes, and dispersion and filtering of certain frequencies within the mud channel—can be detected at one or more ground signal detectors. The final signal traveling up the drill string can depend on the mud velocity, the position of the pulsar in the drill collar, the operating frequency, and the details of the BHA forming the main waveguide. Signal reflections can occur at one or more impedance mismatch locations in the drill string; therefore, equipment and / or operation can be designed to reduce such mismatches. In various scenarios, multiple receivers can be used, for example, based on transient impulse responses in the uplink and downlink channels, to employ multi-echo suppression.

[0068] As mentioned, formation properties can affect MPT (Medium-to-Temperature Transmission). For example, the amount of solid particles in the formation, the volume fraction of free gas, and the compressibility of the mud channel can affect mud pulse velocity and signal attenuation. Changes in mud density and compressibility, with increasing mud solids and free gas content, can decrease mud pulse velocity, which can add complexity to signal transmission speed. As an example, an initial mud pulse propagation velocity model might treat the drilling fluid as a single phase; however, drilling fluid can be a multiphase fluid that can include clay, formation cuttings of various sizes, barite, and free gas mixed with water or oil, which can affect the mud bulk modulus; note that water-based mud velocities tend to be higher than oil-based mud velocities. In various situations, if gas inrush is not promptly counteracted, the instability effects can escalate into a harmful blowout.

[0069] Sources of random noise in the well can include pressure fluctuations caused by drill bit vibration, drill motor stall, and the interaction between the drill bit and the formation being drilled. Mechanical drill rig vibration and electrical noise coupled into the electrical wiring that carries electrical signals from sensors to signal receivers can degrade telemetry signal detection and extraction. Due to its low noise spectrum, this type of noise can present as band-limited white Gaussian noise. Although random noise frequencies may tend to be relatively small, they can affect certain signal bands, resulting in relatively large random pressure amplitude fluctuations and reducing the downhole signal-to-noise ratio (SNR).

[0070] As explained, various signal attenuations can degrade the quality of ground-received signals and complicate telemetry signal detection and extraction. As an example, a framework could be designed to provide improved MPT, particularly improved recovery of transmitted signals for LWD-enhanced operations.

[0071] As an example, the framework can implement machine learning models to configure one or more parameters of the MPT system. For instance, consider a machine learning model that could be or include a deep ensemble residual network for hybrid MPT planning and operation.

[0072] As an example, a framework can be implemented to select telemetry parameters when planning operations. Such a framework could include a Deep-Ens-ResNet to model information bit confidence and combine the information bit confidence with the results of a physics-based model to achieve a hybrid approach (e.g., a combination of machine learning-based and physics-based methods). As an example, a Deep-Ens-ResNet could be trained using historical data, allowing the trained model to output estimated bit confidence and the uncertainty surrounding the estimate, which can be used (whether performed by a human or machine) for decision-making to determine when to trust the model's predictions.

[0073] As an example, bit confidence can be used as a metric to characterize transmission quality (e.g., on a bit-by-bit basis). For instance, a high bit confidence can be considered a high transmission quality; conversely, a low bit confidence can be considered a low transmission quality. As an example, one or more thresholds can be used to categorize transmission quality using one or more bit confidence metrics.

[0074] As explained, MPT can be implemented to transmit downhole acquired data (e.g., LWD data and / or MWD data) to the surface using pressure pulses in the mud system. Measurements can be converted into amplitude or frequency modulation modes of the mud pulses. In various cases, the MPT system can be used bidirectionally, for example, to transmit commands from the surface to the downhole (e.g., downlink). As an example, measurements performed downhole can be stored in solid-state memory for a certain period of time and later transmitted to the surface (e.g., and / or retrieved at the surface when equipment is pulled out of the wellbore). Data transmission can involve digitally encoding the data and transmitting the encoded data to the surface as pressure pulses in the mud system.

[0075] As an example, pressure pulses can be generated by the rotor / stator pair of an MWD module (e.g., an MWD tool) when the rotor controllably (e.g., periodically) impedes flow through the stator. As another example, planning can be carried out using physics-based models designed to model the attenuation and reflection of pressure waves in the drilling fluid as they travel through the pipe from the downhole modulator to the surface. Such models can be used to estimate the appropriate telemetry bands to be used prior to normal drilling operations. For example, telemetry estimation parameters can be considered that may include modes (e.g., telemetry methods for power-on), frequency (Hz), and bit rate (transmission speed in bits per second (bps)).

[0076] In various telemetry operations, the basic operating parameters that need to be selected may include: carrier frequency (e.g., the center frequency of the telecommunication signal transmitted around it); and symbol rate (e.g., the transmission speed of the telecommunication system, which may be closely related to the bandwidth occupancy of the transmitted signal). These parameters can be carefully selected to make the drilling of the section as smooth and reliable as possible. For this purpose, several operating aspects, including attenuation, channel, and noise, can be considered. Regarding attenuation, as drilling progresses and the tool reaches deeper within the wellbore (e.g., along the measured depth and / or actual vertical depth), the communication signal becomes more attenuated. This attenuation can be managed such that, at the end of the section, the signal energy at the surface is high enough to be received without error. Regarding the channel, the speed of telemetry signals tends to be relatively slow, for example, traveling at approximately 1000 m / s in mud. This relatively low propagation speed, combined with changes in acoustic impedance along the propagation path, can produce severe signal distortion due to multiple reflections in the propagation channel. Regarding noise, it can pose a variety of challenges to MPT, as explained, which can be caused by one or more sources (e.g., mud pump noise, drilling noise, electrical noise, etc.).

[0077] The ability to send information through a communication channel can be limited by a maximum transmission rate, which can be called the bandwidth of the channel. According to information theory, even in noisy channels with relatively low bandwidth, it is possible to achieve essentially perfect, error-free communication by keeping the transmission rate within the bandwidth of the channel and by using error correction schemes: the transmission of additional bits makes it possible to extract data from a noisy signal (see, for example, Shannon, CE (1948), “A Mathematical Theory of Communication”, Bell System Technical Journal, 27, 379-423).

[0078] Figure 4 Example graph 400 shows the relationship between bit error rate (BER) and so-called signal-to-noise ratio (SNR). In the case of telecommunications systems, the ratio of signal energy to noise measured at the ground can be used to establish the SNR, or more precisely, the ratio of energy per bit to noise power spectral density (Eb / N0), which defines the system's performance. Figure 4 The graph 400 shows the relationship between the bit error rate (BER) and the surface area ratio (SNR) of the QPSK modulation technique. As can be seen, below approximately 8 dB, the system reliability is affected by 2 x 10⁻⁶ SNR. -4 The significant impact of the error probability. Although in Figure 4The example mentions QPSK, but one or more other techniques can be used, which may include one or more other phase shift keying (PSK) techniques. PSK can be defined as a digital modulation process that transmits data by changing (e.g., modulating) the phase of a constant-frequency carrier. As an example, modulation can be achieved by changing the sine and cosine inputs at precise times. As examples, one or more of amplitude shift keying (ASK, e.g., OOK, etc.), frequency shift keying (FSK), PSK, orthogonal frequency division multiplexing (OFDM), trellis-coded modulation (TCM), etc., can be utilized.

[0079] As an example, during operation, if the error rate is less than 1e -3 In such cases, the workflow might require updating the carrier frequency and bit rate in an effort to improve the SNR. In such examples, this update could be achieved via a flow-rate downlink, but this is often a time-consuming and expensive operation (e.g., by introducing non-productive time (NPT)). As explained, the downlink could involve temporarily halting drilling and / or one or more other operations. Furthermore, the downlink can be combined with the uplink, for example, in the case of transmitting the uplink to acknowledge receipt of the downlink.

[0080] Downlink may be undesirable or otherwise limited for one or more reasons. As explained, when the error rate falls below a threshold (e.g., low SNR), it may be desirable to adjust one or more telemetry parameters. However, in various cases where telemetry parameters can be better selected before the need for mud pulse telemetry, there may be less need for downlink (e.g., no need at all).

[0081] To limit the risk of low SNR during drilling intervals, planning can be utilized to provide estimates of the most likely available telemetry parameters. As explained, drilling of the wellbore can occur within intervals, where intervals may gradually decrease in size (e.g., wellbore diameter) as the wellbore becomes longer. Therefore, the drill string can be pulled out (POOH) at the end of the interval where the BHA and / or one or more other components are selected for drilling the next interval. In such methods, telemetry parameters can be programmed at the surface, for example, before the drill string is run in (RIH). While the need for downlink can be reduced or eliminated through proper planning, one or more techniques can provide improved telemetry parameters that can be transmitted via downlink when necessary, despite potential costs (e.g., time, resources, etc.). In general, proper planning can provide improved data quality, improved data volume, less interruption in interval drilling, etc.

[0082] Regarding planning, as an example, a physics-based model telemetry planner can be implemented, which can encapsulate years of theoretical and experimental knowledge related to the behavior of acoustic waves in drilling mud. Such a planner can be able to use numerical simulation techniques (such as, for example, finite element numerical simulation) to simulate the propagation of telemetry signals to the ground, where the physics-based equations are discretized to account for physical phenomena such as attenuation and reflection. However, such planners may sometimes not be implemented or implemented infrequently because they may lack accuracy and / or require several hours of interaction to collect input data.

[0083] Figure 5 Example graphs 510, 520, and 530 are shown comparing the real signals (SPT1 and SPT2 (signal pressure transmitters)) with the predicted signals from a planner (such as a planner based on the aforementioned numerical simulation technique). Although the predicted signals follow the global trend of the real signals, the errors tend to be relatively high. In the graphs, SPT1 is shown in purple, SPT2 in red, and the predicted signal strength is shown in green.

[0084] Several reasons could explain why this type of mismatch can be observed during operation. For example, consider slight model biases, changes in operating conditions, etc. In general, planners can be very sensitive to the accuracy of the input to the planner model, where small mismatches can produce large variations in the predicted values. As an example, to improve planning, the output of a physics-based model planner can be supplemented with historical datasets, such as those recorded in the same or similar locations, in an effort to fill gaps in input inaccuracies by utilizing historical data.

[0085] As an example, the framework can employ a machine learning model of bit confidence trained on historical data. In such examples, the framework can provide supplementary physics-based models, such as a physics-based model planner. As explained, the framework can provide implementations of hybrid approaches that leverage one or more machine learning-based models and one or more physics-based models.

[0086] As an example, a metric called bit confidence can be used to describe the quality factor related to the reliability of telemetry decoding. For QPSK modulation, the log-likelihood ratio (LLR) bit confidence can be defined as follows: in d m This represents, for example, the Euclidean distance between the estimated symbol and the constellation center of the symbol in a multidimensional space (e.g., a multidimensional chart, a graph, etc.). m (These can be 1, 2, 3, etc.); Note the subscript.k It can be used to distinguish between bit information and symbol information.

[0087] As an example, bit confidence metrics can be at their maximum at high SNR, where symbols are close to the center of a matching constellation. As explained, distance can be used to characterize the difference between what was received and what should have been received (e.g., whether in simulation and / or in a real system).

[0088] Figure 6 Example plot 600 shows simulation results estimating bit confidence versus BER (e.g., the relationship between BER probability and the percentage of samples at different bit confidence levels). As can be seen, the percentage of points above a given threshold decreases significantly as BER deteriorates. Figure 6 In the example, the graphs indicate bit confidence metrics greater than 50, greater than 75, and greater than 90. As shown, the percentage of samples with various BER levels increases with lower bit confidence. As explained, bit confidence can be used as a metric characterizing transmission quality (e.g., via BER, etc.). Graph 600 can be viewed as a reliability graph indicating the degradation of bit quality.

[0089] Figure 7 An example model with inputs and outputs for performance prediction (e.g., bit confidence prediction) is shown, where, for example, the inputs may include various parameters, such as one or more of telemetry parameters, operational parameters, drilling rig parameters, etc. Such models can be used, for example, to predict the reliability (e.g., performance quality, etc.) of a telemetry system through bit confidence prediction. As shown, various parameters can be utilized, including parameters of the geographical, geological, equipment, operational, etc. Regarding operations performed at one or more sites (e.g., drilling site, well site, etc.), such parameters may be available in one or more databases.

[0090] Regarding models used for performance prediction, they can be designed to utilize extensive historical data covering various job types and operations, including details of telemetry patterns (e.g., data rates and carrier frequencies), rig parameters (e.g., location), and attributes related to operational parameters used for the job (e.g., mud parameters, BHA specifications, etc.). As an example, machine learning methods can explore the potential of various features as inputs for training one or more machine learning models (e.g., considering feature engineering, etc.). As explained, the trained machine learning model (e.g., an ML model) can be implemented by a framework for predicting the performance of a telemetry system. As an example, the output can be used to customize the input, for example, in an iterative loop where telemetry parameters can be customized (e.g., optimized); note that the various parameters can be fixed (e.g., location) or otherwise unsuitable for customization or substantial customization (e.g., planned wellbore section sizes, etc.).

[0091] As mentioned, ML models can be or include ResNet or residual neural networks as deep learning types, where weight layers can learn residual functions from the inputs of reference layers. Residual networks can be networks with skip connections that can perform, for example, identity mappings by merging layer outputs through addition. As an example, additionally or alternatively, one or more other types of models can be utilized. As an example, consider one or more of the dense layers of fully connected neural networks, gradient-based models (e.g., XGBoost, Light Gradient Boosting Machine (LGBM), CATBoost, etc.). Deep learning types of models can be utilized in various situations where sufficient amounts of data (e.g., sufficiently representative) can be obtained from various telemetry operations.

[0092] Figure 8 An example of a portion of a deep residual network is shown, specifically residual block 800. Such blocks can be used in neural networks that provide computer vision; note that the framework can utilize such blocks with tabular data rather than raster (e.g., pixel) images. As an example, the framework can provide data presented as one or more images (e.g., via encoding the data in a way that can be represented by pixels).

[0093] As an example, for ResNet, tabular data can be represented as follows: ResNet(x) = Prediction(ResNetBlock(…(ResNetBlock(Linear(x))))) ResNetBlock(x) = x + Dropout(Linear(Dropout(ReLU(Linear(BatchNorm(x)))))) Prediction(x) = Linear(ReLU(BatchNorm(x))) Figure 9 An example of an architecture 900 for encoding categorical features is shown. Specifically, architecture 900 shows components of a feature word morphemer. As shown in the figure, in Figure 9 In the examples, the feature word metafactor can handle category input (e.g., see...). 1 and 2) Furthermore, it can also handle numerical inputs. As an example, a feature morphist can be implemented to process the input... Transform into embedding T ∈ R (k x d) As explained, inputs about categories can be transformed into values. As an example, a lexer can operate by, for instance, splitting a categorical input (e.g., text) into lexes according to a set of rules. These lexes can then be converted into numbers, which can be used to construct tensors as input to the model. As explained, additional inputs to the model can also be added by the lexer. As an example, various types of inputs to the model can be categorical inputs and / or numerical inputs.

[0094] As an example, the framework could leverage deep ensemble methods, such as architectures with ResNet-like features. In such examples, deep ensemble methods could include initializing a training set with random weights. A neural network (e.g., consider a set of ResNet models, etc.). These networks can include two outputs: the mean and the mean. and variance By treating the observed values ​​as samples from a Gaussian distribution with predicted mean and variance, the workflow can include minimizing the negative log-likelihood criterion: As an example, the ensemble can be considered as a uniformly weighted Gaussian mixture model with the following mean and variance: Figure 10A and Figure 10BAn example of workflow 1000 is shown, along with various blocks and model architectures for a deep integration of ResNet models (e.g., a ResNet model ensemble that could be called Deep-Ens-ResNet). As illustrated, additionally or alternatively, input features may include a variety of features such as, for example, telemetry-related variables (e.g., tool characteristics, patterns, frequencies, drilling rates, surface pipe dimensions, sensor locations, etc.), mud properties (e.g., pressure, flow rate, density, viscosity, etc.), operational parameters (e.g., section diameter, bottom depth, inclination angle, etc.), and one or more other inputs.

[0095] exist Figure 10A and Figure 10B In the example, the workflow illustrates residual blocks as comprising various components that may include batch normalization, dense layers, rectified linear units (ReLU), dropouts, dense layers, and dropout components, where feedforward (e.g., skipping) may be utilized. As explained, a residual block can be a skip connection block designed to learn the residual function of the reference layer input (e.g., rather than learning an unreferenced function); note that the ResNet architecture includes one or more residual blocks. As explained, the ResNet architecture can use one or more skip connections to address the vanishing gradient problem. As an example, the ResNet architecture can stack multiple identity mappings (e.g., convolutional layers that do nothing initially), skip those layers, and reuse activations from previous layers, where skipping can accelerate initial training by compressing the network into fewer layers. As an example, during retraining, all layers can be expanded, allowing the remainder of the network (e.g., the residual portion) to explore more feature space of the input.

[0096] like Figure 10A and Figure 10B As shown in the example, a feature word meta-transformer can be utilized. Feature word meta-transformers can be customized as needed; note that the example feature word meta-transformer shows a combination of three numerical features and two categorical features. As shown, the feature word meta-transformer can transform the input (x) into an embedding (T). The input embedding can be fed into a dense layer, followed by multiple residual blocks, and then another dense layer, where each of the residual blocks can, for example, include various components, such as those explained above. As shown, the tabular ResNet model can generate outputs such as the bit confidence mean and bit confidence variance values.

[0097] like Figure 10A and Figure 10BAs shown in the example, a Deep-Ens-ResNet architecture can be formed by multiple assembly components, such as an ensemble of multiple table ResNet models. In such an example, the input can be fed into the ensemble, where the output about the mean and variance can be generated by averaging the individual outputs of each of the multiple table ResNet models.

[0098] The number of ResNet models in the ensemble can be customized (e.g., consider more than two and less than twenty). Figure 10A and Figure 10B In the example, individual ResNet models can take the same features (e.g., input); however, they can be initialized with different weights at the start of training, which can thus provide differently weighted (e.g., trained) ResNet models that can provide diverse outputs. In such approaches, given the diversity of initial weights among individual ResNet models, each of the ResNet models can have a common architecture but capture aspects of the data slightly differently.

[0099] Figure 11 An example workflow 1100 for a hybrid approach to control telemetry is illustrated. As shown, workflow 1100 includes receiving input from a trained ML model and input from a physics-based engine, allowing weighting of the outputs to generate an updated ranking of telemetry control values, optionally along with uncertainty. For example, once Deep-Ens-Net has been trained on historical data, such models can be used to obtain estimates of bit confidence and its uncertainty, which can be combined with the output of the physics-based engine to rank telemetry parameters for use in controlling an MPT system. As shown in workflow 1100, the output can be generated by using predictions from both the machine learning-based model and the physics-based model. In such approaches, the combination can account for the uncertainty of the estimated bit confidence by giving smaller weights to machine learning scores that may have higher uncertainty values.

[0100] exist Figure 11 In the example, the parameter ranking provided by the physics-based engine is based on physics parameter inputs. Note that the inputs to the physics-based engine can include various parameters of MPT (e.g., signal frequency, data rate, etc.). The physics-based engine can output MPT parameters that are expected to be reliable for systems including MPT equipment. For example, the output can be or include one or more of modulation mode, data rate, and frequency (e.g., physics-based recommendations).

[0101] Regarding the operational parameters input to the ML model, these can be operational parameters used for the operation of a telemetry system (e.g., within a physical environment characterized by physical parameter inputs from a physics-based engine). As an example, various parameters can be input to the ML model (e.g., see...). Figure 10A and Figure 10B (Examples). For example, consider parameters regarding current and / or past MPT performance, rig location, mud used during operations, etc.

[0102] exist Figure 11 In the examples, physics-based outputs and data-driven outputs can be combined to improve MPT performance, regardless of whether it is via planning and / or control (e.g., segment-to-segment control, real-time control, etc.). As an example, the output can include a list of parameter sets for MPT (e.g., data rate 1, frequency 2; data rate 2, frequency 1; etc.). As explained, a physics-based engine can output such sets in a ranked order, which can effectively recommend one set over another. As an example, a field engineer can consider the top three parameter sets (e.g., MPT parameters). As illustrated, an ML model can generate performance predictions (e.g., bit confidence). As an example, an ML model can provide evaluations of more than three ranked parameter sets, such as, for example, the top ten parameter sets. The output of the ML model can be used to weight a selected number of ranked sets output by the physics-based model, such that an updated ranking can be generated at least in part based on the bit confidence predicted by the ML model, along with, for example, uncertainty.

[0103] As an example, one approach could include leveraging a trained residual network model to predict the mean and variance of bit confidence (digital bits, not drill bits), where the residual network model could include residual blocks characterized by skip connections. Such approaches can have the ability to mitigate the risk of experiencing the vanishing gradient problem, especially in the context of deep neural networks. As explained, the model can be trained using a negative log-likelihood loss function to output the mean and variance.

[0104] As an example, deep ensemble networks can be used for robust estimation of uncertainty. For instance, consider an ensemble of residual neural networks initialized randomly or otherwise to achieve estimation of uncertainty. In such examples, the output mean and variance can be aggregated for the final prediction.

[0105] As explained, the framework can implement a hybrid approach that combines machine learning-based model outputs and physics-based model outputs in a way that takes into account uncertainties (e.g., in either or both outputs).

[0106] Figure 12An example of system 1200 is shown, which includes a frame 1210, a drilling rig control system 1230, and one or more downhole tools 1250. In such an example, the drilling rig control system 1230 can provide information transmission to the frame (e.g., via wired and / or wireless connections), whereby the frame 1210 generates one or more control parameters for telemetry. In such an example, telemetry can be used for transmissions to and / or from one or more of the one or more downhole tools 1250, wherein such telemetry can be controlled at least in part based on one or more of the one or more control parameters. In such an example, system 1200 can provide for the planning and / or execution of operations, wherein, for example, the frame 1210 can be implemented in a real-time manner for one or more field operations, such as, for example, determining optimal telemetry control parameters for one or more sections of a wellbore being drilled or to be drilled. For example, the frame 1210 can be invoked if there are changes in the drilling fluid, and / or if there are changes in one or more of the section specifications, BHA equipment, etc. As explained, various factors can affect telemetry, especially mud pulse telemetry. Such factors may be known a priori before field operations and / or may be known once at least one or more field operations have begun and / or been completed, which may, for example, lead to one or more changes to the planning (e.g., drilling planning that specifies the equipment, technology, etc. for drilling one or more sections of the wellbore).

[0107] As an example, the framework can provide performance evaluation for an MPT system. For instance, a hybrid planner can generate MPT parameter values ​​for configuring the MPT system used for transfer during drilling operations in sections of the wellbore. In such examples, one or more error metrics can be calculated, which can be in real time. As an example, one or more error metrics in operation can be compared with one or more expected error metrics that can be generated by the framework. In such examples, if a deviation occurs between the performance in operation and the expected performance, the system can request a replanning of the MPT parameters, which can include assessing whether the deviation is caused by the MPT system (e.g., MPT equipment) and / or one or more other factors (e.g., changes in mud density, reservoir fluid influx into the wellbore, etc.). In such approaches, the framework can provide detection of one or more problems that may occur during drilling operations and potentially affect MPT performance. As an example, operation data can be added to one or more databases, which can provide additional training for one or more ML models and / or modification of one or more physics-based models.

[0108] Figure 13An example of method 1300 is shown, which includes: a receiving block 1310 for receiving data from field operations using equipment at the site, wherein the equipment includes downhole tools and a mud pulse telemetry system set on a tool string in a wellbore given in the geological environment; a determining block 1320 for determining control parameters of the mud pulse telemetry system using at least a portion of the data and a trained machine learning model; and a control block 1330 for controlling the mud pulse telemetry system using the control parameters.

[0109] Figure 13 Various computer-readable medium (CRM) blocks 1311, 1321, and 1331 are also shown. Such blocks may include instructions executable by one or more processors, which may be one or more processors of a computing framework, system, computer, etc. The computer-readable medium may be a non-signal, non-carrier, and non-transitory computer-readable storage medium. For example, a computer-readable medium may be a physical memory component capable of storing information in a digital format.

[0110] exist Figure 13 In the example, system 1390 includes one or more information storage devices 1391, one or more computers 1392, one or more networks 1395, and instructions 1396. Regarding the one or more computers 1392, each computer may include one or more processors (e.g., processing cores) 1393 and memory 1394 for storing instructions 1396, which may be executed, for example, by at least one of the one or more processors. As an example, the computer may include one or more network interfaces (e.g., wired or wireless), one or more graphics cards, display interfaces (e.g., wired or wireless), etc. System 1390 may be specifically configured to perform... Figure 13 Method 1300 includes one or more parts.

[0111] Regarding the types of machine learning models, consider one or more of the following: Support Vector Machine (SVM) models, k-Nearest Neighbor (KNN) models, ensemble classifier models, neural network (NN) models, etc. As examples, machine learning models can be deep learning models (e.g., deep Boltzmann machines, deep belief networks, convolutional neural networks, stacked autoencoders, etc.), ensemble models (e.g., random forests, gradient boosting machines, bootstrap aggregation, AdaBoost, stacked generalization, gradient boosting regression trees, etc.), neural network models (e.g., radial basis function networks, perceptrons, backpropagation, Hopfield networks, etc.), regularization models (e.g., ridge regression, minimum absolute shrinkage and selection operators, elastic networks, minimum angle regression), rule-based system models (e.g., stereotyped, one-rule, zero-rule, repeated incremental pruning to produce error reduction), and regression models (e.g., linear regression, ordinary least squares regression, stepwise regression, multivariate adaptive regression splines, local estimation scatter plot smoothing, log-odds regression, etc.). Bayesian models (e.g., Naive Bayes, Average Dependency Estimator, Bayesian Belief Network, Gaussian Naive Bayes, Multinomial Naive Bayes, Bayesian Networks), decision tree models (e.g., Classification and Regression Trees, Iterative Dichotomy 3, C4.5, C5.0, Chi-square Automatic Interaction Detection, Decision Stumps, Conditional Decision Trees, M5), dimensionality reduction models (e.g., Principal Component Analysis, Partial Least Squares Regression, Salmon Map, Multidimensional Scaling, Projective Tracking, Principal Component Regression, Partial Least Squares Discriminant Analysis, Mixed Discriminant Analysis, Quadratic Discriminant Analysis, Regularized Discriminant Analysis, Flexible Discriminant Analysis, Linear Discriminant Analysis, etc.), instance models (e.g., k-Nearest Neighbors, Learned Vector Quantization, Self-Organizing Map, Locally Weighted Learning, etc.), clustering models (e.g., k-means, k-median, Expectation Maximization, Hierarchical Clustering, etc.), etc.

[0112] As an example, computational frameworks with libraries, toolboxes, etc. (such as those in the MATLAB framework (MathWorks, Inc., Natick, MA)) can be used to build machine learning models (ML models). The MATLAB framework includes toolboxes that provide supervised and unsupervised machine learning algorithms, including Support Vector Machines (SVM), boosting and bagged decision trees, k-nearest neighbors (KNN), k-means, k-centroids, hierarchical clustering, Gaussian mixture models, and hidden Markov models. Another MATLAB framework toolbox is the Deep Learning Toolbox (DLT), which provides a framework for designing and implementing deep neural networks using algorithms, pre-trained models, and apps. DLT provides convolutional neural networks (ConvNet, CNN) and long short-term memory (LSTM) networks for performing classification and regression on image, time series, and text data. DLT includes features for building network architectures such as generative adversarial networks (GANs) and Siamese networks using custom training loops, shared weights, and automatic differentiation. DLT provides model exchange to various other frameworks.

[0113] As examples, the TENSORFLOW framework (Google LLC, Mountain View, California) can be implemented; it is an open-source software library for dataflow programming that includes a symbolic mathematics library and can be implemented for machine learning applications that may include neural networks. As an example, the CAFFE framework can be implemented; it is a DL framework developed by the Berkeley AI Research Center (BAIR) (University of California, Berkeley, California). As another example, consider utilizing the Python programming language SCIKIT platform (e.g., scikit-learn). As an example, frameworks such as the APOLLO AI framework (APOLLO.AI GmbH, Germany) can be utilized. As an example, frameworks such as the PYTORCH framework (Facebook AI Research Lab (FAIR) of Facebook, Inc., Menlo Park, California) can be utilized.

[0114] As an example, training methods can include various actions that can manipulate the dataset to train the ML model. As another example, the dataset can be split into training and testing data, where the testing data can provide evaluation. Methods can include parameters and cross-validation of optimal parameters, which can be provided for model training.

[0115] The TENSORFLOW framework can run on multiple CPUs and GPUs, with optional CUDA (NVIDIA Corp., Santa Clara, California) and SYCL (The Khronos Group Inc., Beaverton, Oregon) extensions for general-purpose computing on graphics processing units (GPUs). TENSORFLOW is available on 64-bit Linux, macOS (Apple Inc., Cupertino, California), Windows (Microsoft Corp., Redmond, Washington), and mobile computing platforms including those based on Android (Google LLC, Mountain View, California) and iOS (Apple Inc.).

[0116] TENSORFLOW computation can be represented as a stateful data flow graph; note that the name TENSORFLOW derives from the operations performed by this type of neural network on a multidimensional data array. Such an array can be referred to as a "tensor".

[0117] As an example, one approach may include: using equipment at the site to receive data from field operations, wherein the equipment includes downhole tools and a mud pulse telemetry system on a tool string located in a wellbore within the geological environment; using at least a portion of the data and a trained machine learning model to determine control parameters for the mud pulse telemetry system; and using the control parameters to control the mud pulse telemetry system.

[0118] In such examples, determining the mean digital bit confidence level may include using a trained machine learning model. In such examples, determining the variance of the digital bit confidence level may include using a trained machine learning model.

[0119] As an example, trained machine learning models may include ensemble models. In such examples, the mean and variance of the ensemble model can provide aggregate mean and aggregate variance.

[0120] As an example, one approach could include predicting the variance of digital bit confidence by including individual models of the ensemble model that are randomly initialized. For instance, consider a method for training that includes individual models of the ensemble model that are initialized differently.

[0121] As an example, one approach may include determining whether to generate predictions using a trained machine learning model by generating predictions using a physics-based model and by combining predictions at least in part based on uncertainties in the predictions. In such examples, the combination may include weighting the predictions of the physics-based model based at least in part on the predictions of the trained machine learning model. As an example, the physics-based model may at least model the decay of mud pulses.

[0122] As an example, one approach could include determining which modulations can be considered. For instance, considering modulations that could include PSK modulation (e.g., QPSK modulation, etc.).

[0123] As an example, control parameters may include one or more of telemetry mode, telemetry data rate, and telemetry frequency.

[0124] As an example, a trained machine learning model may include at least one residual block. In such an example, at least one residual block may include at least one residual block characterized by one or more skip connections.

[0125] As an example, the negative log-likelihood loss function can be used to train a trained machine learning model.

[0126] As an example, tabular data can be used to train a trained machine learning model, where the tabular data includes data from field operations performed at other locations. As another example, a feature lexer can be used to train a trained machine learning model that converts at least a portion of the tabular data from category lexers to numbers.

[0127] As an example, one method may include controlling a mud pulse telemetry system to transmit sensor data acquired by a downhole tool via generating mud pulses in drilling fluid disposed in a wellbore, wherein the tool string includes a hole and wherein drilling fluid is disposed in the hole.

[0128] As an example, a system may include: a processor; a memory accessible by the processor; and processor-executable instructions stored in and executable in the memory to instruct the system to: receive data from field operations using equipment at the site, wherein the equipment includes downhole tools and a mud pulse telemetry system on a tool string located in a wellbore in the geological environment; determine control parameters for the mud pulse telemetry system using at least a portion of the data and a trained machine learning model; and control the mud pulse telemetry system using the control parameters.

[0129] As an example, one or more computer-readable storage media may include processor-executable instructions to instruct a computing system to: receive data from field operations using equipment at the site, wherein the equipment includes downhole tools and a mud pulse telemetry system on a tool string located in a wellbore in the geological environment; determine control parameters for the mud pulse telemetry system using at least a portion of the data and a trained machine learning model; and control the mud pulse telemetry system using the control parameters.

[0130] As an example, a computer-readable medium (CRM) can be used in part to implement a method as, for example, a module, block, etc., which includes information such as instructions suitable for execution by one or more processors (or processor cores) to instruct a computing device or system to perform one or more actions. As an example, a single medium may be configured with instructions to at least partially allow the execution of various actions of the method. As an example, the computer-readable medium (CRM) may be a non-carrier computer-readable storage medium (e.g., a non-transitory medium). As an example, a computer program product may include instructions suitable for execution by one or more processors (or processor cores), wherein the instructions can be executed to implement at least a portion of one or more methods.

[0131] According to the implementation scheme, one or more computer-readable media may include computer-executable instructions that are output by an instruction computing system for controlling a process. For example, such instructions may be provided as output to a sensing process, injection process, drilling process, extraction process, extrusion process, pumping process, heating process, etc.

[0132] In some implementations, one or more methods can be performed by a computing system. Figure 14 An example of system 1400 is shown, which may include one or more computing systems 1401-1, 1401-2, 1401-3 and 1401-4, which may be operatively coupled via one or more networks 1409, which may include wired networks and / or wireless networks.

[0133] As an example, the system may include an arrangement of individual computer systems or distributed computer systems. Figure 14 In the example, computer system 1401-1 may include one or more modules 1402, which may be or include, for example, processor-executable instructions that are executable to perform various tasks (e.g., receiving information, requesting information, processing information, simulating, outputting information, etc.).

[0134] As an example, a module may execute independently or in cooperation with one or more processors 1404, which are operatively coupled to one or more storage media 1406 (e.g., via wired, wireless, etc.). As an example, one or more of the processors 1404 may be operatively coupled to at least one of one or more network interfaces 1407. In such examples, computer system 1401-1 may, for example, transmit and / or receive information via one or more networks 1409 (e.g., consider one or more of the Internet, private networks, cellular networks, satellite networks, etc.). As shown, one or more other components 1408 may be included in computer system 1401-1.

[0135] As an example, computer system 1401-1 may receive information from and / or transmit information to one or more other devices, which may be or may include, for example, one or more of computer systems 1401-2. The devices may be located in a physical location different from the physical location of computer system 1401-1. As an example, the location may be, for example, a processing facility location, a data center location (e.g., a server farm), a drilling rig location, a well site location, a downhole location, etc.

[0136] As an example, a processor may be or may include a microprocessor, a microcontroller, a processor module or subsystem, a programmable integrated circuit, a programmable gate array, or another control or computing device.

[0137] As an example, storage medium 1406 may be implemented as one or more computer-readable or machine-readable storage media. As an example, storage may be distributed within and / or across multiple internal and / or external housings of the computing system and / or additional computing systems.

[0138] As an example, one or more storage media may include one or more different forms of memory, including: semiconductor memory devices such as dynamic or static random access memory (DRAM or SRAM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), and flash memory; magnetic disks such as fixed disks, floppy disks, and removable disks; other magnetic media, including magnetic tape; optical media such as optical discs (CDs) or digital video discs (DVDs), Blu-ray discs, or other types of optical storage; or other types of storage devices.

[0139] As an example, one or more storage media may be located in a machine that runs machine-readable instructions, or at a remote site from which machine-readable instructions can be downloaded over a network for execution.

[0140] As an example, various components of a system (such as a computer system, for example) can be implemented in hardware, software, or a combination of both (e.g., including firmware), including one or more signal processing circuits and / or application-specific integrated circuits.

[0141] As an example, the system may include a processing device that may be or include a general-purpose processor or a special-purpose chip (e.g., a chipset), such as an ASIC, FPGA, PLD, or other suitable device.

[0142] As an example, the device may be a mobile device including one or more network interfaces for transmitting information. For example, the mobile device may include a wireless network interface (e.g., operable via IEEE 802.11, ETSI GSM, BLUETOOTH, satellite, etc.). As an example, the mobile device may include components such as a main processor, memory, display, display graphics circuitry (e.g., optionally including touch and gesture circuitry), SIM slot, audio / video circuitry, motion processing circuitry (e.g., accelerometer, gyroscope), wireless LAN circuitry, smart card circuitry, transmitter circuitry, GPS circuitry, and battery. As an example, the mobile device may be configured as a cellular phone, tablet computer, etc. As an example, a mobile device may be used to implement a method (e.g., wholly or partially). As an example, a system may include one or more mobile devices.

[0143] As an example, the system can be a distributed environment, such as a so-called "cloud" environment, in which various devices, components, etc., interact for purposes such as data storage, communication, and computation. As an example, a device or system may include one or more components for transmitting information via one or more of the Internet (e.g., in the case of communication via one or more Internet protocols), cellular networks, satellite networks, etc. As an example, a method can be implemented in a distributed environment (e.g., wholly or partially as a cloud-based service).

[0144] As an example, information can be input from a display (e.g., consider a touchscreen), output to a display, or both. As an example, information can be output to a projector, laser device, printer, etc., so that the information can be viewed. As an example, information can be output stereoscopically or holographically. Regarding printers, consider 2D or 3D printers. As an example, a 3D printer can include one or more materials that can be output to construct 3D objects. For example, data can be provided to a 3D printer to construct a 3D representation of underground strata. As an example, layers (e.g., horizons, etc.) can be constructed in 3D, geological bodies, etc., can be constructed in 3D. As an example, wellbores, fractures, etc. (e.g., as positive structures, as negative structures, etc.) can be constructed in 3D.

[0145] Although only a few examples have been described in detail above, those skilled in the art will readily understand that many modifications can be made to the examples. Therefore, all such modifications are intended to be included within the scope of this disclosure as defined in the appended claims. In the claims, the device plus function clauses are intended to cover structures described herein that perform the enumerated functions, and not only structural equivalents, but also equivalent structures. Thus, although nails and screws may not be structural equivalents because nails use a cylindrical surface to hold wooden parts together, while screws use a helical surface, nails and screws can be equivalent structures in the context of fastening wooden parts.

Claims

1. A method comprising: Using on-site equipment to receive data from field operations, said equipment includes downhole tools and mud pulse telemetry systems on a tool string set in the wellbore in the geological environment; The control parameters of the mud pulse telemetry system are determined using at least a portion of the data and a trained machine learning model. as well as The control parameters are used to control the mud pulse telemetry system.

2. The method of claim 1, wherein the determination includes: The trained machine learning model is used to predict the confidence level of the mean digit bit.

3. The method of claim 2, wherein the determination includes: The trained machine learning model is used to predict the variance of the confidence level of digital bits.

4. The method of claim 3, wherein the trained machine learning model includes an ensemble model.

5. The method of claim 4, wherein the mean and variance of the ensemble model include aggregated mean and aggregated variance.

6. The method of claim 4, wherein the variance of predicting the confidence level of the digital bit comprises: The individual models of the ensemble model are randomly initialized.

7. The method of claim 1, wherein the determination comprises: The trained machine learning model is used to generate predictions; Use physics-based models to generate predictions; And the predictions are combined based at least in part on the uncertainties in the predictions.

8. The method of claim 7, wherein the combination comprises: The predictions of the physics-based model are weighted at least in part based on the predictions of the trained machine learning model.

9. The method of claim 7, wherein the physics-based model at least models the attenuation of mud pulses.

10. The method of claim 1, wherein the determination takes modulation into account.

11. The method of claim 10, wherein the modulation comprises QPSK modulation.

12. The method of claim 1, wherein the control parameters include one or more of telemetry mode, telemetry data rate, and telemetry frequency.

13. The method of claim 1, wherein the trained machine learning model comprises at least one residual block.

14. The method of claim 13, wherein the at least one residual block comprises at least one residual block characterized by one or more skip connections.

15. The method of claim 1, wherein the trained machine learning model is trained using a negative log-likelihood loss function.

16. The method of claim 1, wherein the trained machine learning model is trained using tabular data, wherein the tabular data includes data from field operations performed at other sites.

17. The method of claim 16, wherein the trained machine learning model is trained using a feature morphemer, which morphemes at least a portion of the tabular data from category terms to numbers.

18. The method of claim 1, wherein the control comprises: The mud pulse telemetry system is controlled to transmit sensor data acquired by the downhole tool via generating mud pulses in the drilling fluid disposed in the wellbore, wherein the tool string includes a hole and the drilling fluid is disposed in the hole.

19. A system comprising: processor; A memory that can be accessed by the processor; Processor-executable instructions, which are stored in the memory and can be executed to instruct the system: Using on-site equipment to receive data from field operations, said equipment includes downhole tools and mud pulse telemetry systems on a tool string set in the wellbore in the geological environment; The control parameters of the mud pulse telemetry system are determined using at least a portion of the data and a trained machine learning model. as well as The control parameters are used to control the mud pulse telemetry system.

20. One or more computer-readable storage media, including processor-executable instructions to instruct a computing system: Using on-site equipment to receive data from field operations, said equipment includes downhole tools and mud pulse telemetry systems on a tool string set in the wellbore in the geological environment; The control parameters of the mud pulse telemetry system are determined using at least a portion of the data and a trained machine learning model. as well as The control parameters are used to control the mud pulse telemetry system.

Images