Monitoring a drilling rig

Abstract

A method of monitoring a drilling rig 100 comprising receiving audio samples (406, Fig 4) and vibration samples (408, Fig 4). The audio samples represent a sound produced by the drilling rig. The vibr

Description

The present disclosure relates to a method and apparatus for monitoring a drilling rig. Drilling rigs are used in the oil and gas, mining, geothermal and construction industries to create boreholes in rock. A number of metrics exist to quantify drilling operations performed by a drilling rig, such as the speed of rotation of a drill string or the weight on a drill bit. However, many drilling rigs lack the sensors needed to measure those metrics. Without the ability to measure such metrics, the success of a drilling operation relies upon the skill of a human operator who observes the drilling rig and decides how best to control the rig in order to carry out the drilling operation. Summary In accordance with a first aspect of the present disclosure, a method of monitoring a drilling rig is provided. The method comprises receiving audio samples, the audio samples representing a sound produced by the drilling rig. The method further comprises receiving vibration samples, the vibration samples representing a vibration produced by the drilling rig. The method further comprises processing the audio samples and the vibration samples to identify a state of the drilling rig and / or to determine one or more metrics of an operation performed by the drilling rig. Audio data and vibration data can be collected using transducers (e.g., microphones and accelerometers) that are easily retrofitted to an existing drilling rig, without requiring physical modifications to the drilling rig itself. The method disclosed herein thus allows monitoring of the state of an existing drilling rig that does not have built-in monitoring capabilities. Alternatively or additionally, the method disclosed herein allows an operation performed using such an existing drilling rig to be monitored. The method and apparatus disclosed herein may find particular use in the mining industry. In general, drilling rigs used in the mining industry tend to have very limited instrumentation, and tend not to have the ability to monitor or analyse metrics of operations being performed by the drilling rig. The method and apparatus disclosed herein allows such drilling rigs to be easily and non-invasively upgraded to have monitoring capabilities. The present disclosure is not limited solely for use in the mining industry. The techniques described herein can be applied to drilling rigs used in the oil and gas industry, such as for oil or gas exploration and / or oil or gas extraction. Alternatively or in addition, the techniques described herein can be applied to drilling rigs used for: drilling boreholes for extracting water from below ground; drilling boreholes for extracting geothermal energy; and / or drilling boreholes for piles of buildings and other structures. Other fields of application of the techniques described herein may become apparent to the skilled person after reading the present disclosure. The method disclosed herein may be wholly or partially implemented using a processor. The processor may be incorporated into an apparatus for monitoring a drilling rig, which is referred to herein as a “monitoring apparatus”. The phrase “sound produced by the drilling rig” is intended to encompass sounds resulting from the drilling rig performing an operation, even if those sounds are not produced by a component of the drilling rig itself. For example, a sound produced by a drilling fluid (also known as drilling mud) may be regarded as a sound produced by the drilling rig. However, some states of the drilling rig may be identifiable and some metrics may be determinable using sounds produced only by the drilling rig itself (e.g., using sounds of the drilling rig’s motor, drill bit or mud pump etc.). Likewise, the phrase “vibration produced by the drilling rig” is intended to encompass vibrations resulting from the drilling rig performing a particular operation, even if those vibrations do not occur in a component of the drilling rig itself. For example, vibration of the drill floor or the borehole may be regarded as a vibration of the drilling rig. However, some states of the drilling rig may be identifiable and some metrics may be determinable using vibrations of the drilling rig itself (e.g., using vibrations of the drilling rig’s derrick, motor, drill bit, drill string, rotary table, kelly drive, top drive or mud pump etc.). By using a microphone to capture audio samples and an accelerometer to capture vibration samples, the monitoring apparatus is able to observe a wider frequency range than would be possible using a microphone alone or an accelerometer alone. Moreover, the monitoring apparatus is able to identify states and / or to determine metrics that cannot be identified or determined using audio alone or vibration alone. For instance, the processing of vibration samples allows the identification of states and / or the determination of metrics that are characterised by frequency components below the audio frequency range (e.g., below 20 Hz). The processing of vibration samples allows the identification of states and / or the determination of metrics that are characterised by frequency components in the audio frequency range (e.g., between 20 Hz and 20 kHz). The audio samples and the vibration samples may represent sound and vibration during a common time period. That is to say, the audio samples and the vibration samples may have been captured during exactly the same time period, or during a different but overlapping time period. The term “metric” refers to a measurable property of an operation performed by the drilling rig. The metric need not be a property that is directly controllable. The “state” of a drilling rig refers to the type of operation being performed by the drilling rig at a particular moment in time, as will become clearer from the examples described below. The state of the drilling rig and / or the one or more metrics of the operation may be displayed to a user e.g., in the form of a dashboard showing the current state of a drilling rig and / or the current value of each metric. Alternatively or in addition, the state of the drilling rig and / or the one or more metrics of the operation may be stored, e.g., in the form of a log recording the state of the drilling rig and / or the values of the metrics over a period of time. Alternatively or in addition, the state of the drilling rig and / or the one or more metrics of the operation may be transmitted to a data processing device, whereupon they may undergo further processing or analysis. Optionally, identifying the state of the drilling rig includes identifying that the drilling rig is: performing a drilling operation; performing a tripping operation; performing a core extraction operation; performing a borehole cleaning operation; performing a borehole stabilising or sealing operation; performing a surveying operation; performing a wireline logging operation; undergoing drill bit replacement; undergoing addition, removal or replacement of one or more sections of a drill string; undergoing maintenance; or inactive. The monitoring apparatus may be configured to identify any one or more of these states. The foregoing list of states is provided purely by way of example, and the monitoring apparatus could be configured to identify further states if the need arises. Each state of the drilling rig is characterised by a sound, a vibration or a combination thereof. For example, a drilling operation may be characterised by a sound and vibration whose frequency spectrum has a peak power between 60 and 120 Hz, corresponding to the speed of rotation of the drill bit. A core extraction operation may be characterised by a sound and vibration whose frequency spectrum has a peak power at a frequency higher than 120 Hz, corresponding to the speed of a motor of a wireline core extraction device. A tripping operation may be characterised by a sound and vibration whose frequency spectrum has a peak power at yet another frequency, corresponding to the speed of a motor used to raise or lower the drill string. The addition or replacement of a section of a drill string may be characterised by a sporadic occurrence of relatively high power noise with a broad frequency spectrum, where such noise is caused by clamping adjacent sections of a drill string to one another. Inactivity of the drilling rig may be characterised by a sound and vibration whose frequency spectrum has no identifiable peak power. Thus, each state of the drilling rig can be identified by processing the audio samples and the vibration samples to identify particular features that characterise a particular state. The drilling rig is said to be performing a drilling operation when a drill string is rotating and / or when a drill bit is cutting a substrate. A number of different types of drilling operations may be uniquely identified, including: pilot hole drilling, in which a hole with a relatively small diameter is drilled before a borehole drilling operation is performed; borehole drilling, in which a general purpose drill bit (e.g., a roller cone bit) is used to drill a borehole; core drilling, in which a hollow drill bit is used to cut a rock specimen from the borehole for analysis; and / or reaming, in which the diameter of a borehole is enlarged. The drilling rig is said to be performing a tripping operation when the drill string is being raised from the borehole and / or lowered into the borehole. The drilling rig is said to be performing a core extraction operation when a specimen is being raised from the borehole to the surface. The drilling rig is said to be performing a borehole cleaning operation when a drilling fluid is circulated within the borehole to remove cuttings. The drilling rig is said to be performing a borehole stabilising or sealing operation when it is performing any task intended to strengthen the walls of the borehole, to reduce or prevent ingress of fluid into the borehole, and / or to reduce or prevent egress of fluid out of the borehole. Non-limiting examples of borehole stabilising or sealing operations include grouting, and installing casing or liners in the borehole. The drilling rig is said to be performing a surveying operation when the direction, depth and / or diameter of the borehole is being measured. The drilling rig is said to be performing a wireline logging operation when tools (other than the drill string) are lowered into the borehole to collect data. The drilling rig is said to be undergoing drill bit replacement when any type of drill bit (including, but not limited to, a roller cone bit or a hollow bit for core drilling operation) is being replaced. The drilling rig is said to be inactive when it is not performing any operation. It should be appreciated that other states of the drilling rig are possible, depending on the capabilities of a particular drilling rig. For example, some drilling rigs may be able to perform other operations in addition to drilling, tripping and core extraction. Conversely, some drilling rigs may not be able to perform core extraction operation. Optionally, the one or more metrics include at least one metric relating to a drilling operation performed by the drilling rig. The at least one metric relating to a drilling operation optionally includes any one or more of: rotational speed of a drill string of the drilling rig; weight on a drill bit of the drilling rig; a depth of the drill bit; and / or a rate of penetration. When the metric is the rotational speed of a drill string, processing the audio samples and the vibration samples may include performing a frequency analysis of the samples. For example, a Fourier transform of the audio samples and / or the vibration samples may be performed to generate a frequency spectrum representing one or more frequencies present in the sound and / or vibration. The frequency spectrum may be analysed to identify the rotational speed of the drill string. When the metric is the depth of the drill bit, processing the audio samples may include identifying the addition, removal and / or replacement of one or more sections of a drill string. The length of the drill string may be inferred from the total number of sections present in the drill string following the addition, removal and / or replacement of a section. The depth of the drill bit may then be estimated by multiplying the length of the drill string (expressed as a number of sections) by the length of each section. Optionally, the one or more metrics include at least one metric relating to a tripping operation performed by the drilling rig. Optionally, the one or more metrics include at least one metric relating to a core extraction operation performed by the drilling rig. Optionally, the one or more metrics include one or more of: a starting time of the operation performed by the drilling rig; an end time of the operation performed by the drilling rig time; and / or a duration of the operation performed by the drilling rig. Optionally, the one or more metrics include at least one metric relating to inactivity of the drilling rig. The at least one metric relating to inactivity of the drilling rig optionally includes any one or more of: a starting time of a period of inactivity of the drilling rig; an end time of the period of inactivity of the drilling rig; and / or a duration of the period of inactivity of the drilling rig. Optionally, processing the audio samples and the vibration samples to identify a state of the drilling rig comprises: processing the audio samples, the vibration samples and / or features derived therefrom using a trained classifier, the trained classifier being configured to classify input data as being indicative of a particular state of the drilling rig- Optionally, processing the audio samples and the vibration samples to determine one or more metrics of an operation performed by the drilling rig comprises: processing the audio samples, the vibration samples and / or features derived therefrom using a trained classifier, the trained classifier being configured to estimate a value of the one or more metrics. Optionally, the method further comprises: operating the drilling rig in a predetermined manner in which the state of the drilling rig and / or one or more metrics of the operation performed by the drilling rig are known; receiving audio samples and vibration samples when the drilling rig is operating in the predetermined manner; and labelling the audio samples and vibration samples received when the drilling rig was operating in the predetermined manner with the known state of the drilling rig and / or the one or more known metrics of the operation. The audio samples, the vibration samples and the labels can be used as training data to train a classifier. By acquiring training data in this manner, the classifier can be trained upon a particular drilling rig with a particular configuration (e.g., number and placement) of microphones and accelerometers. Optionally, the method further comprises training a classifier to identify a state of the drilling rig by using the audio samples and vibration samples received when the drilling rig was operating in the predetermined manner, and / or features derived therefrom, and the known state of the drilling rig as training data. The training data may thus be used to train the classifier to classify audio samples, vibration samples and / or features derived therefrom as being indicative of a particular state of the drilling rig. The classifier may be trained using a supervised learning process or any other suitable type of learning process. Optionally, the method further comprises training a classifier to estimate a value of one or more metrics of the operation performed by the drilling rig by using the audio samples and vibration samples received when the drilling rig was operating in the predetermined manner, and / or features derived therefrom, and the one or more known metrics of the operation as training data. The training data may thus be used to train the classifier to classify audio samples, vibration samples and / or features derived therefrom as being indicative of a particular state of the drilling rig. The classifier may be trained using a supervised learning process or any other suitable type of learning process. Optionally, the method further comprises determining an optimised value of one or more parameters of the drilling rig and / or a drilling process. The term “parameter” refers to a controllable property of the drilling rig or drilling process. The property may be controlled by controlled manually or automatically. In the case of a property that is controlled manually, the property may be controlled by a person working with the drilling rig such as a driller (i.e., a person who operates the drilling rig), a derrickman (i.e., a person who handles the drill string), a roughneck (i.e., a person working on the drill floor) and / or by a tool pusher (i.e., a person who oversees the drilling rig). In the case of a property that is controlled automatically, the monitoring apparatus disclosed herein may provide an input to the drilling rig, wherein the input causes a change in the value of the parameter. The optimised value may be determined so as to improve the efficiency, productivity, safety or some other aspect of the drilling rig and / or its operation. The monitoring apparatus disclosed may automatically determine the optimised value of the parameter. The monitoring apparatus can then automatically set the parameter to the optimised value automatically. Alternatively or additionally, the monitoring apparatus may output a human-readable indication of the optimised value, such that a person working with the drilling rig can manually set the parameter to the optimised value. The optimised value may be determined by processing the audio samples and / or the vibration samples. Alternatively or in addition, the optimised value may be based on the based on the state of the drilling rig and / or the one or more metrics of the operation performed by the drilling rig. The term “drilling process” refers to one or more operations performed by a drilling rig. A drilling process usually comprises a plurality of operations that are performed in a particular order. Optionally, the method further comprises controlling the drilling rig and / or the drilling process using the optimised value of the one or more parameters of the drilling rig and / or drilling process. The efficiency, productivity, safety or some other aspect of the drilling rig and / or its operation can be improved by controlling the drilling rig and / or the drilling process using the optimised value. Certain operations of the method disclosed herein may be implemented manually. In particular, operations involving manipulation of a drill string or a drill bit may be performed manually, as is customary in the field of drilling. In accordance with a second aspect of the present disclosure, an apparatus for monitoring a drilling rig is provided. The apparatus comprises a processor configured to perform any of the methods disclosed herein. Optionally, the apparatus further comprises one or more microphones. Optionally, wherein the one or more microphones include a directional microphone. A directional microphone can avoid noise from other components of the drilling rig being captured, without the need for filtering or other signal processing operations. Optionally, the directional microphone is oriented towards a portion of the drilling rig that produces a sound that characterises a state of the drilling rig and / or a metric of an operation performed by the drilling rig. Optionally, the one or more microphones include an omnidirectional microphone. An omnidirectional microphone allows sounds produced by a relatively wide area of the drilling rig to be captured. Optionally, at least one microphone of the one or more microphones comprises a magnet for detachably attaching the microphone to the drilling rig. Optionally, the apparatus further comprises one or more accelerometers. Optionally, at least one accelerometer of the one or more accelerometers comprises a magnet for detachably attaching the accelerometer to the drilling rig. The monitoring apparatus can be provided independently, such that it can be retrofitted to an existing drilling rig. Alternatively, the monitoring apparatus can be provided together with a drilling rig. In accordance with a further aspect of the present disclosure, a drilling rig is provided. The drilling rig comprises an apparatus for monitoring a drilling rig as disclosed herein. In accordance with a further aspect of the disclosure, a processor-readable medium is provided. The processor-readable medium comprises processor-executable instructions which, when executed by a processor, cause an apparatus comprising the processor to perform any of the methods disclosed herein. The processor-readable medium may be non-transitory (such as a disk or a memory device) or may be transitory (such as a signal). In accordance with a further aspect of the disclosure, a computer program is provided. The computer program comprises processor-executable instructions which, when executed by a processor, cause an apparatus comprising the processor to perform any of the methods disclosed herein. Brief Description of the Drawings Embodiments will now be described, purely by way of example, with reference to the accompanying drawings, in which like features are denoted by like reference signs, and in which: Figure 1 is a schematic diagram of an example of a drilling rig comprising a monitoring apparatus in accordance with the present disclosure; Figure 2 is a schematic diagram of an example of a monitoring apparatus in accordance with the present disclosure; Figure 3 is a schematic diagram of an example digital processing device of the monitoring apparatus shown in Figure 2; Figure 4 is a flow chart of an example of a method of monitoring a drilling rig in accordance with the present disclosure; and Figure 5 is a schematic diagram of an example arrangement for processing audio samples and vibration samples in accordance with the present disclosure. Detailed Description Figure 1 is a schematic diagram of a non-limiting example of a drilling rig 100 comprising a monitoring apparatus 200 in accordance with the present disclosure. Figure 1 is not to scale, and does not show the exact placement of components of the drilling rig 100 or the monitoring apparatus 200. Figure 1 does not depict a specific type of drilling rig. The drilling rig 100 is arranged to drill into a substrate 110. The substrate 110 may be a body of rock, for example. The drilling rig 100 comprises a drill string 120 and a drill bit 122. The drill string 120 comprises a plurality of tubular sections fitted end to end. The drill bit 122 is attached to an end of the drill string 120. The drill bit 122 is a core drill bit for extracting a specimen 114 of the substrate 110 in the example shown in Figure 1, but the present disclosure can be implemented using other types of drill bit. The drilling rig 100 further comprises a motor 130. The motor 130 may be an internal combustion engine (e.g., a diesel engine) or an electric motor. The motor 130 is mechanically coupled to the drill string 120, such that operation of the motor 130 causes the drill string 120 to rotate about its longitudinal axis. The motor 130 is coupled to the drill string 120 by a rotary table 132 in the example shown in Figure 1, but other coupling mechanisms are possible within the scope of the present disclosure. The drilling rig may comprise a bearing 134 to support the drill string 120 as it rotates. Rotation of the drill string 120 causes the drill bit 122 to cut the substrate 110, forming a borehole 112. As the drill bit 122 cuts the substrate 110, the drill string 120 and drill bit 122 penetrate into the substrate 110. The length of the drill string 120 can be extended as it penetrates deeper into the substrate 110 by adding additional tubular sections to the end of the drill string 120 that is opposite the drill bit 122. To allow a specimen 114 of the substrate 110 to be extracted, the drilling rig 100 further comprises a wireline 124 and a second motor 140. The wireline 124 extends through a bore 121 within the drill string 122. Operation of the second motor 140 pulls the specimen 114 of the substrate 110 from the drill bit 122 and along the bore 121. The monitoring apparatus 200 comprises one or more (in this case, three) microphones 202. A first microphone 202a is positioned near, and oriented towards, the motor 130, so as to capture a sound produced by the motor 130. A second microphone 202b is positioned near, and oriented towards, the second motor 130, so as to capture a sound produced by the second motor 140. A third microphone 202c is positioned near, and oriented towards, the drill string 120, so as to capture a sound produced by the drill string 120. In this example, the first and second microphones 202a, 202b are directional microphones, whereas the third microphone 202c is an omnidirectional microphone. The monitoring apparatus may comprise a different number of microphones. The positions of the microphones described above are given purely by way of example, and microphones can be positioned elsewhere so as to capture sounds produced by any other component of the drilling rig 100. The microphones can be any combination of directional and omnidirectional microphones, depending on the sound that is to be captured. The optimal distance of each microphone 202 from the source of a sound can be found through trial and error. If a microphone is placed too close to the source of a sound, there is a risk that the microphone will produce a clipped signal. Conversely, the sound of interest may be difficult to distinguish from background noise if the microphone is placed too far from the source of the sound. As a rough guide, each microphone 202 is typically positioned around one to five metres from the source of the sound of interest. The exact distance will depend upon the sensitivity of the microphone, the amplitude of the sound of interest and the amplitude of noise. Any or all of the microphones 202 may comprise a magnet 203 to allow the microphone to be detachably attached to a metal component of the drilling rig 100 (e.g., to a housing of a motor 130, 140, or to a derrick). Other means of attaching the microphones 202 to the drilling rig 100 are possible. The monitoring apparatus 200 further comprises one or more (in this case, three) accelerometers 204. A first accelerometer 204a is attached to the bearing 134, so as to capture a vibration produced when the drill string 120 rotates. A second accelerometer 204b is attached to the rotary table 132, so as to capture a vibration produced when the rotary table 132 causes the drill string 120 to rotate. A third accelerometer 204c is attached to the drill string 120, so as to capture a vibration of the drill string 120. The monitoring apparatus may comprise a different number of accelerometers, and the accelerometers can be positioned to capture vibrations produced by any component of the drilling rig. The positions of the accelerometers described above are given purely by way of example, and the accelerometers can be positioned elsewhere so as to capture vibrations produced by any other component of the drilling rig. Any or all of the accelerometers 204 may comprise a magnet 203 to allow the accelerometer to be detachably attached to a metal component of the drilling rig 100 (e.g., to the rotary table 203, bearing 134, or drill string 120). Other means of attaching the accelerometers 204 to the drilling rig 100 are possible. Figure 2 is a schematic diagram of the monitoring apparatus 200 shown in Figure 1. The monitoring apparatus 200 comprises one or more microphones 202, one or more accelerometers 204, an analogue front end 210, one or more analogue-to-digital converters 220, and a digital processing device 300. Each microphone 202 is configured to convert sound produced by the drilling rig 100 into an analogue electrical signal. Similarly, each accelerometer 204 is configured to convert a vibration produced by the drilling rig 100 into an analogue electrical signal. The outputs of the microphone(s) 202 and accelerometer(s) 204 are electrically connected to the input of the analogue front end 210. The analogue front end 210 is configured to condition the analogue electrical signals produced by the microphone(s) 202 and accelerometer(s) 204. For example, the analogue front end 210 may comprise a low-pass filter to remove high-frequency noise and / or to prevent aliasing by the analogue-to-digital converter(s) 220. Alternatively or in addition, the analogue front end 210 may comprise an amplifier to amplify or attenuate the analogue electrical signals produced by the microphone(s) 202 and accelerometer(s) 204, so as to use the full dynamic range of the analogue-to-digital converter(s) 220. In general, the analogue front end 210 has a separate signal path for each analogue signal to be conditioned. The analogue front end 210 has a plurality of analogue outputs, each corresponding to the signal produced by a respective microphone 202 or accelerometer 204. The outputs of the analogue front end 210 are electrically connected to the analogue-to-digital converter(s) 220. The analogue-to-digital converter(s) 220 are configured to convert each analogue electrical signals produced by the analogue front end 210 into a respective sequence of digital values, known as samples. The analogue-to-digital converter(s) 220 thus output one or more sequences of audio samples, where each sequence of audio samples corresponds to the sound incident on a respective microphone 202. Similarly, the analogue-to-digital converter(s) 220 output one or more sequences of vibration samples, where each sequence of vibrations samples corresponds to the vibration measured by a respective accelerometer 204. The monitoring apparatus 200 may comprise a respective analogue-to-digital converter 220 for each analogue electrical signal that is to be digitised. Alternatively, the monitoring apparatus 200 may comprise fewer analogue-to-digital converters 220 (and may comprise only one analogue-to-digital converter 220) by multiplexing multiple outputs of the analogue front end 210 onto the inputs of each analogue-to-digital converter 220. The audio samples and vibration samples produced by the analogue-to-digital converter(s) 220 are provided as inputs to the digital processing device 300. The output(s) of the analogue-to-digital converter(s) 220 are coupled to the input(s) of the digital processing device 300 by a wired connection, by a wireless connection, or by a combination of both wired and wireless connections. In a wired implementation, the outputs of the analogue-to-digital converters 220 are electrically connected to the inputs of the digital processing device 300 by an electrically conducting wire or cable. In a wireless implementation, the outputs of the analogue-to-digital converters 220 are coupled to the inputs of the digital processing device 300 by a radio frequency communication link. The radio frequency communication link can be implemented using any suitable communication protocol, such as Bluetooth™ or IEEE 802.11 (Wi-Fi ™). The digital processing device 300 shown in Figure 2 will now be described with reference to Figure 3. The digital processing device 300 comprises a processor 310, a memory 320, and a digital input / output (I / O) interface 330. The digital processing device 300 may optionally further comprise a user interface 340 and / or a network interface 350. In Figure 3, dashed lines are used to illustrate optional features of the digital processing device 300. The processor 310 can be any suitable type of data processing device, such as a microprocessor, microcontroller or application specific integrated circuit (ASIC). The processor 310 is communicatively coupled to the memory 320. The memory 320 can include a volatile memory, a non-volatile memory, or both volatile and non-volatile memories. The memory 320 stores a monitoring program 322. The monitoring program 322 includes processor-executable instructions that, when executed by the processor 310, cause the processor 310 to perform the method described below with reference to Figure 4. The memory 320 optionally stores one or more classifiers 324, training data 326 and / or log data 328. If present, the one or more classifiers 324 may be used to process the audio samples and the vibration samples to identify a state of the drilling rig 100 and / or to determine one or more metrics of an operation performed by the drilling rig, as described in more detail below with reference to block 410 of Figure 4. Alternatively or additionally, the one or more classifiers 324, if present, may be used to determine an optimised value of one or more parameters of the drilling rig 100 and / or drilling process, as described in more detail below with reference to block 412 of Figure 410. If present, the training data 326 is used to train the one or more classifiers 324, as described in more detail below with reference to block 404 of Figure 4. Alternatively, the one or more classifiers 324 may be pre-trained, such that there is no need for the memory 320 to store training data 326. The log data 328 comprises data indicative of the state of the drilling rig 100 at one or more previous moments in time and / or one or more metrics of an operation performed by the drilling rig 100 at one or more previous moments in time. The digital I / O interface 330 receives audio samples and vibration samples from the analogue-to-digital converter 220. In implementations in which the monitoring apparatus 200 is capable of controlling the drilling rig 100, the digital I / O interface 330 may also communicate one or more control signals to the drilling rig 100. The digital I / O interface 330 may optionally be configured to receive data from other sensors (not shown in the drawings) on or near the drilling rig 100. The user interface 340 comprises a display 342 and a keyboard 344. The display 342 is configured to output a visual indication of the state of the drilling rig 100 and / or one of more metrics of an operation performed by the drilling rig 100. The display 342 can be any suitable type of output device. For example, the display 342 may be a liquid crystal display (LCD) screen or an organic light-emitting diode (OLED) screen. The keyboard 344 comprises a plurality of buttons, which a user can use to input information for use by the monitoring program 322. For example, a user (such as a driller or tool pusher) can use the keyboard 344 to select which metric(s) are shown on the display 342. The display 342 and keyboard 344 may be integrated with one another in the form of a touchscreen. The network interface 340 allows the monitoring apparatus 200 to send data to and / or receive data from one or more other devices using a communication network. For example, the network interface 340 can be used to send data indicative of the state of the drilling rig 100 and / or one or more metrics of an operation performed by the drilling rig 100 to a remote computer via a communication network. As another example, the network interface 340 can be used to receive updates to the one or more classifiers 324 from a remote computer via a communication network. The communication network may be a local area network, a wide area network or a combination thereof. The communication network may include any type of wired and / or wireless data network, and may include the Internet. In some implementations, the network interface 340 may be used to implement a wired or wireless communication link that couples one or more outputs of the analogue-to-digital converter(s) 220 to the digital processing device 300. The operation of the monitoring apparatus 200 will now be described with reference to Figure 4, which illustrates a method 400 of monitoring a drilling rig 100. In Figure 4, dashed lines are used to illustrate optional features of the method 400. The method 400 may begin at block 402, at which the monitoring apparatus 200 collects training data 326. To collect the training data 326, the drilling rig 100 is operated in a predetermined manner in which the state of the drilling rig 100 and / or one or more metrics of the operation performed by the drilling rig 100 are known. For example, the motor 130 is activated, such that the drill string 120 rotates; in this case, the known state of the drilling rig 100 is that the drilling rig is performing a drilling operation. The rotational speed of the drill string 120 may be measured using a stroboscope or other suitable instrument, which may or may not be a part of the drilling rig 100 itself; in this case, the known metric of the drilling operation is that the rotational speed of the drill string 120 is the value measured by the stroboscope. As another example, the second motor 140 is activated, such that a specimen 114 of the substrate 110 is pulled along the bore 121; in this case, the known state of the drilling rig 100 is that the drilling rig is performing a core extraction operation. As yet another example, both motors 130, 140 are deactivated; in this case, the known state of the drilling rig 100 is that the drilling rig is inactive, and the known metric of the drilling operation is that the rotational speed of the drill string 120 is zero. While the drilling rig 100 is operating in the predetermined manner, audio samples and vibration samples are received by the monitoring apparatus 200. The audio samples and / or vibration samples are labelled with the known state of the drilling rig 100 and / or the metric(s) of the operation, thereby creating the training data 326. For example, labelling the samples may include adding metadata to a file comprising the samples, where the metadata describes the known state of the drilling rig 100 and / or contains value(s) of the known metric(s). The drilling rig 100 may be operated several times in each of several different manners, so as to build a corpus of training data 326. When the training data 326 has been collected at block 402, the method 400 proceeds to block 404. At block 404, the classifier(s) 324 are trained using the training data 326. For example, supervised learning (or any other suitable machine learning technique) can be used to train a classifier 324 to identify a particular state of the drilling rig 100. As another example, supervised learning technique (or any other suitable machine learning technique) can be used to train a classifier 324 to estimate a value of one or more metrics of an operation performed by the drilling rig. When the classifier(s) have been trained at block 404, the method 400 proceeds to block 406. Block 404 may be performed by the monitoring apparatus 200, or it may be performed by another computing device with more processing resources. Optionally, a classifier 324 may be trained to determine an optimised value of one or more parameters of the drilling rig 100 and / or one or more parameters of the drilling process. In one implementation, a human expert (such as an experienced driller) observes the drilling rig when it is being operated at block 402, and labels the samples with a recommendation on how a parameter of the drilling rig 100 and / or drilling process should be optimised. For example, the human expert may label the samples with a recommendation to increase the rotational speed of the drill string 120 to a particular value. Using the recommendation, supervised learning (or any other suitable machine learning technique) is used to train a classifier 324 to output an optimised value of one or more parameters of the drilling rig 100 and / or one or more parameters of the drilling process. In another implementation, reinforcement learning is used to determine an optimised value of one or more parameters of the drilling rig 100 and / or one or more parameters of the drilling process, without involvement of a human expert. If the classifier(s) 324 have been previously trained, or if the monitoring apparatus 200 does not include any classifiers 324, blocks 402 and 404 can be omitted. In this case, the method begins at block 406. At block 406, the monitoring apparatus 200 receives audio samples. The audio samples represent a sound produced by the drilling rig 100, and are captured using the microphone(s) 202. The method 400 then proceeds to block 408. At block 408, the monitoring apparatus 200 receives vibration samples. The vibration samples represent a vibration produced by the drilling rig 100, and are captured using the accelerometer(s) 204. The method 400 then proceeds to block 410. At block 410, the monitoring apparatus 200 processes the audio samples and vibration samples that were received at blocks 406 and 408, respectively. For example, the audio samples and vibration samples may be input to the classifier(s) 324 that were trained at block 404. The classifiers 324 may process the samples and generate one or more outputs indicative of a state of the drilling rig 100 and / or one or more metrics of an operation performed by the drilling rig 100. The processing of audio samples and vibration samples is described in more detail below with reference to Figure 5. When the samples have been processed at block 410, the method 400 may return to block 406, or proceed to either block 412 or block 414. At block 412, the monitoring apparatus 200 outputs the state of the drilling rig 100 and / or one or more metrics of the drilling operation. For example, the monitoring apparatus 200 may display the current state of the drilling rig and / or the current value of each metric on the display 342. Alternatively or additionally, the monitoring apparatus 200 may record the current state of the drilling rig and / or the current value of each metric in the log data 328. Alternatively or additionally, the monitoring apparatus 200 may use the network interface 350 to transmit the current state of the drilling rig and / or the current value of each metric to a remote computing device. The remote computing device may receive the current states of a plurality of drilling rigs and / or the current values of various metrics of a plurality of drilling rigs. The remote computing device may further provide a dashboard whereby the operation of several drilling rigs can be monitored and analysed. The method 400 may then return to block 406 or proceed to block 414. At block 414, the monitoring apparatus 200 determines an optimised value of one or more parameters of the drilling rig 100 and / or an optimised value of one or more parameters of a drilling process performed using the drilling rig 100. The optimised value(s) may be determined using the classifier(s) 324 that were trained at block 404. For example, audio samples and / or vibration samples may be input to the classifier(s) 324, and the classifiers 324 may process the samples and generate one or more outputs indicative of the optimised value(s). Alternatively or additionally, the state of the drilling rig 100 and / or the metric(s) that were determined at block 410 may be input to the classifier(s) 324, and the classifiers 324 may process the state and / or metric(s) to generate one or more outputs indicative of the optimised value(s). When the optimised value(s) have been determined at block 414, the method 400 may return to block 406, or proceed to either block 414 or block 416. At block 416, the monitoring apparatus 200 outputs an optimised value of one or more parameters of the drilling rig 100 and / or an optimised value of one or more parameters of a drilling process performed using the drilling rig 100. For example, the monitoring apparatus 200 may display the optimised value(s) on the display 342. Alternatively or additionally, the monitoring apparatus 200 may record the optimised value(s) in the log data 328. Alternatively or additionally, the monitoring apparatus 200 may use the network interface 350 to transmit the optimised value(s) to a remote computing device. The method 400 may then return to block 406 or proceed to block 418. At block 418, the monitoring apparatus 200 controls the drilling rig 100 based on the optimised value(s) of the parameters that were determined at block 414. For example, the monitoring apparatus 200 can control an input of the drilling rig 100 so as to set a parameter of the drilling rig and / or a parameter of the drilling process to the optimised value. In more detail, the digital I / O interface 330 may be used to output an electrical signal which, when received by an input of the drilling rig 100, sets the value of one or more parameters to the optimised value(s). The method 400 may then return to block 406 or proceed to block 416. The sequence of operations shown in Figure 4 is merely exemplary. Any of the tasks shown in method 400 may be performed in a different order that achieves substantially the same result. For example, blocks 406 and 408 may be performed simultaneously, such that the digital processing device 300 receives the audio samples and vibration samples at the same time. Figure 5 is a schematic diagram of an example arrangement for processing audio samples and vibration samples in accordance with the present disclosure. The arrangement shown in Figure 5 is implemented in the digital processing device 300, and performs block 410 (and, optionally, block 414) of the method 400 shown in Figure 4. As shown in Figure 5, audio samples 500 are supplied as inputs to one or more classifiers 324. The audio samples 500 are also supplied as inputs to an audio feature extractor 502 and a Fast Fourier Transform (FFT) module 520. The audio feature extractor 502 is configured to calculate one or more features of the audio samples 500. For example, the audio feature extractor 502 may calculate the mel-frequency cepstral coefficients, the zero-crossing rate and / or the energy of an audio signal represented by the audio samples. The FFT module 520 is configured to calculate the frequency spectrum of time series data. Thus, when the audio samples 500 are input to the FFT module 520, the frequency spectrum of the audio signal represented by the audio samples is calculated by the FFT module 520. The features calculated by the audio feature extractor 502 and the frequency spectrum calculated by the FFT module 520 are supplied as inputs to the one or more classifiers 324. Vibration samples 510 are supplied as inputs to the one or more classifiers 324. The vibration samples 510 are also supplied as inputs to a vibration feature extractor 512 and the Fast Fourier Transform (FFT) module 520. The vibration feature extractor 512 is configured to calculate one or more features of the vibration samples 510. For example, the vibration feature extractor 512 may be configured to calculate the zero-crossing rate and / or the energy of a vibration represented by the vibration samples. The FFT module 520 calculates the frequency spectrum of the vibration represented by the vibration samples 510. The features calculated by the vibration feature extractor 512 and the frequency spectrum calculated by the FFT module 520 are supplied as inputs to the one or more classifiers 324. The classifiers 324 may comprise one or more artificial neural networks 560. At least one of the artificial neural networks 560 may be a deep neural network. Each deep neural network comprises a plurality of nodes arranged as an input layer, a plurality of hidden layers, and an output layer. Any or all of the inputs to the classifiers 324 may be connected to a respective node of the input layer. Each node of the input layer is connected to at least one node of a first hidden layer of the plurality of hidden layers. Each node of the first hidden layer is connected to at least one node of a second hidden layer of the plurality of nodes. Each node of the second hidden layer is connected to at least one node of a further hidden layer or the output layer, depending on how many hidden layers are present of the plurality of nodes. In any case, each node of a last hidden layer of the plurality of hidden layers is connected to at least one node of the output layer. Each connection between nodes associated with a weight, and the weights are determined during block 404 of the method 400. The nodes of the output layer may be associated with classification outputs. That is, a certain node (or a certain combination of nodes) of the output layer may be associated with a particular state 530 of the drilling rig 100, a particular value (or range of values) of a metric 540, or a particular optimised value of a parameter 550. The classifiers 324 may comprise one or more Bayesian networks 570. Each Bayesian network 570 comprises a plurality of nodes arranged as a directed acyclic graph. Each leaf node of the Bayesian network 570 may represent a state of the drilling rig 100. Each non-leaf node of the Bayesian network 570 may represent a random variable upon which a state of the drilling rig 100 is dependent. Any or all of the inputs to the classifiers 324 may be associated with a respective non-leaf node of a Bayesian network 570. Alternatively or additionally, any or all of the outputs of the artificial neural networks 560 may be associated with a respective non-leaf node of a Bayesian network 570. Each leaf node of the Bayesian network 570 may be connected to an output of the classifiers 324, and may indicate whether the drilling rig 100 is in a particular state 530. The probability of the drilling rig 100 being in a particular state for a given set of values of the random variables represented by the leaf nodes is determined during block 404 of the method 400. Each Bayesian network 570 may thus be trained to identify a particular state of the drilling rig based on the audio samples 500, the vibration samples 510 and features derived therefrom. The output of one classifier 324 may feed into the input of another classifier, so as to perform a particular classification task. For example, a first classifier may be trained to identify whether the motor 130 is operating, a second classifier may be trained to identify when a clamp is attached to the drill string 210, and a third classifier may be trained to identify that the drilling rig 100 is in a particular state based on the outputs of the first and second classifiers. In this example, the first, second and third classifiers may be either an artificial neural network or a Bayesian network. It may be possible to determine some metrics without using the classifiers 324. For 5 example, the rotational speed of the drill string 120 can be determined from the FFT of certain audio and / or vibration samples. The method and apparatus for monitoring a drilling rig disclosed above use both audio samples and vibration samples to identify a state of the drilling rig and to determine the 10 metrics of an operation performed by the drilling rig. Some states of a drilling rig can be identified by processing audio samples alone or using vibration samples alone. Similarly, some metrics of certain operations can be determined using audio samples alone or using vibration samples alone. Therefore, the present disclosure extends to methods and apparatuses for monitoring a drilling rig in which either audio samples or 15 vibration samples are received and processed to identify a state of the drilling rig and / or to determine one or more metrics of an operation performed by the drilling rig. It will be understood that the invention has been described above purely by way of example, and that modifications of detail can be made within the scope of the claims.

Claims

1. A method of monitoring a drilling rig, the method comprising:receiving audio samples, the audio samples representing a sound produced by the drilling rig;receiving vibration samples, the vibration samples representing a vibration produced by the drilling rig; andprocessing the audio samples and the vibration samples to identify a state of the drilling rig and / or to determine one or more metrics of an operation performed by the drilling rig.

2. A method in accordance claim 1, wherein identifying the state of the drilling rig includes identifying that the drilling rig is:performing a drilling operation;performing a tripping operation;performing a core extraction operation;performing a borehole cleaning operation;performing a borehole stabilising or sealing operation;performing a surveying operation;performing a wireline logging operation;undergoing drill bit replacement;undergoing addition, removal or replacement of one or more sections of a drill string;undergoing maintenance; or inactive.

3. A method in accordance with claim 1 or claim 2, wherein the one or more metrics include at least one metric relating to a drilling operation performed by the drilling rig.

4. A method in accordance with claim 3, wherein the at least one metric relating to a drilling operation includes one or more of:rotational speed of a drill string of the drilling rig;weight on a drill bit of the drilling rig;a depth of the drill bit; and / ora rate of penetration.

5. A method in accordance with any of the preceding claims, wherein the one or more metrics include at least one metric relating to a tripping operation performed by the drilling rig.

6. A method in accordance with any of the preceding claims, wherein the one or more metrics include at least one metric relating to a core extraction operation performed by the drilling rig.

7. A method in accordance with any of the preceding claims, wherein the one or more metrics include one or more of:a starting time of the operation performed by the drilling rig;an end time of the operation performed by the drilling rig time; and / or a duration of the operation performed by the drilling rig.

8. A method in accordance with any of the preceding claims, wherein the one or more metrics include at least one metric relating to inactivity of the drilling rig.

9. A method in accordance with claim 8, wherein the at least one metric relating to inactivity of the drilling rig includes one or more of:a starting time of a period of inactivity of the drilling rig;an end time of the period of inactivity of the drilling rig; and / ora duration of the period of inactivity of the drilling rig.

10. A method in accordance with any of the preceding claims, wherein processing the audio samples and the vibration samples to identify a state of the drilling rig comprises:processing the audio samples, the vibration samples and / or features derived therefrom using a trained classifier, the trained classifier being configured to classify input data as being indicative of a particular state of the drilling rig.

11. A method in accordance with any of the preceding claims, wherein processing the audio samples and the vibration samples to determine one or more metrics of an operation performed by the drilling rig comprises:processing the audio samples, the vibration samples and / or features derived therefrom using a trained classifier, the trained classifier being configured to estimate a value of the one or more metrics.

12. A method in accordance with any of the preceding claims, further comprising: operating the drilling rig in a predetermined manner in which the state of the drilling rig and / or one or more metrics of the operation performed by the drilling rig are known;receiving audio samples and vibration samples when the drilling rig is operating in the predetermined manner; andlabelling the audio samples and vibration samples received when the drilling rig was operating in the predetermined manner with the known state of the drilling rig and / or the one or more known metrics of the operation.

13. A method in accordance with claim 12, further comprising training a classifier to identify a state of the drilling rig by using the audio samples and vibration samples received when the drilling rig was operating in the predetermined manner, and / or features derived therefrom, and the known state of the drilling rig as training data.

14. A method in accordance with claim 12 or claim 13, further comprising training a classifier to estimate a value of one or more metrics of the operation performed by the drilling rig by using the audio samples and vibration samples received when the drilling rig was operating in the predetermined manner, and / or features derived therefrom, and the one or more known metrics of the operation as training data.

15. A method in accordance with any of the preceding claims, further comprising: determining an optimised value of one or more parameters of the drilling rig and / or a drilling process.

16. A method in accordance with claim 15, further comprising:controlling the drilling rig and / or the drilling process using the optimised value of the one or more parameters of the drilling rig and / or drilling process.

17. An apparatus for monitoring a drilling rig, wherein the apparatus comprises a processor configured to perform a method in accordance with any of the preceding claims.

18. An apparatus in accordance with claim 17, wherein the apparatus further comprises one or more microphones.

19. An apparatus in accordance with claim 18, wherein the one or more microphones include a directional microphone.

20. An apparatus in accordance with claim 19, wherein the directional microphone is oriented towards a portion of the drilling rig that produces a sound that characterises a state of the drilling rig and / or a metric of an operation performed by the drilling rig.

21. An apparatus in accordance with any of claims 18 to 20, wherein the one or more microphones include an omnidirectional microphone.

22. An apparatus in accordance with any of claims 18 to 21, wherein at least one microphone of the one or more microphones comprises a magnet for detachably attaching the microphone to the drilling rig.

23. An apparatus in accordance with any of claims 17 to 33, wherein the apparatus further comprises one or more accelerometers.

24. An apparatus in accordance with claim 23, wherein at least one accelerometer of the one or more accelerometers comprises a magnet for detachably attaching the accelerometer to the drilling rig.

25. A drilling rig comprising an apparatus for monitoring a drilling rig in accordance with any of claims 17 to 24.

26. A processor-readable medium comprising instructions which, when executed by a processor, cause an apparatus comprising the processor to perform a method in accordance with any of claims 1 to 16.

27. A computer program comprising processor-executable instructions which, when executed by a processor, cause an apparatus comprising the processor to perform a method in accordance with any of claims 1 to 16.

Images