Sensor system for acoustic monitoring

Abstract

A sensor system (1) acquires an acoustic signal from an asset. The system monitors the acoustic signal and generates a trigger signal in response to an event in the acoustic signal. The system also includes means for performing digital signal processing (16) of the acoustic signal when in an awake state. The signal processing means (16) is configured to transition from a sleep state to the awake state in response at least in part to the trigger signal.

Description

[0001] SENSOR SYSTEM FOR ACOUSTIC MONITORING

[0002] The present disclosure relates to the field of sensor systems, particularly those designed for acquiring acoustic signals from an asset. In embodiments, a sensor system is configured to operate in a low power state, transitioning to a high-power state for digital signal processing when an event is detected in the acoustic signal.

[0003] Ultrasonic sensors and their applications have seen significant advancements in recent years. The growing need for remote, real-time monitoring of assets has led to the development of various sensor systems. These sensor systems, such as the one disclosed in US 10,466,209 B2, typically include components for acquiring acoustic signals from an asset and processing these signals in some meaningful way.

[0004] US 10,466,209 B2 describes a sensor system which utilizes a transducer for converting an analogue transmit signal to an ultrasonic transmit signal and vice versa. The transducer is housed in a unit that includes numerous other components like a processor, wireless data transmitter, transmit and receive circuitry along with an A / D converter that aids in digital conversion processes. This reported system follows a mechanism where the processor triggers the transmit / receive circuit and A / D converter repeatedly to obtain a digitized composite signal. Thereafter, the digitized composite reflected signal is processed to generate an A-scan signal which is then wirelessly transmitted as data for transmission to a discrete collection device.

[0005] Although this technology presents certain advancements in ultrasonic sensor technology by incorporating elements such as wireless data transmission and digital conversion processes, it still has its limitations. For instance, it is not well suited to use cases where sensing at regular intervals does not perform satisfactorily.

[0006] SUMMARY

[0007] The present disclosure generally relates to a sensor system that can monitor high- frequency ultrasonic signals from an asset on a continuous or quasi-continuous basis, notwithstanding the power constraints associated with wireless sensor systems and other low-power systems. A first aspect of the disclosure provides a sensor system comprising: means for acquiring an acoustic signal from an asset; means for monitoring the acoustic signal and generating a trigger signal in response to an event in the acoustic signal; and means for performing digital signal processing of the acoustic signal when in an awake state. The means for acquiring an acoustic signal (referred to interchangeably herein as the “signal acquisition means”) is configured for continuous or quasi-continuous monitoring of the asset based on high frequency acoustic signals. The means for acquiring an acoustic signal (referred to interchangeably herein as the “signal processing means”) is configured to transition from a sleep state to the awake state in response at least in part to the trigger signal.

[0008] The term “asset” refers to an apparatus that is to be monitored using the sensor system described herein. In general, an asset that can be monitored using the sensor system will generate acoustic signals (in other words, sound and ultrasound) that are indicative of the status of the asset. For example, the asset may generate one or more acoustic signals when operating normally. The asset may cease to generate some or all of those acoustic signals when some or all of its functionality is inoperative. The asset may generate one or more different acoustic signals when it is operating abnormally (e.g., when there is a currently a fault with the asset, or when a fault or failure is developing). The acoustic signals generated by the asset when operating normally and / or when operating abnormally may be characterised by particular frequency domain components, time domain components, amplitudes or any combination thereof (e.g., a particular pattern of frequencies and / or amplitudes over time). As such, an asset which generates acoustic signals that are suitable for monitoring using the sensor system may be a mechanical, electrical or electromechanical apparatus. An asset may be a system comprising any combination of mechanical, electrical and / or electromechanical apparatuses. An asset may, or may not, comprise moving parts.

[0009] The term “event” refers to a change in an acoustic signal generated by an asset. An event may occur as a result of either normal operation of the asset (e.g., the asset starting or ceasing to operate in accordance with a predetermined schedule or in response to predetermined environmental factors) or abnormal operation of the asset (e.g., in response to a fault with the asset, or when a fault or failure is developing). Example events include but are not limited to sounds associated with bearing wear or failure, rolling contact fatigue, component misalignment, gear mesh issues, cavitation, looseness of parts, leaks, pressure relief valve activation, metal fatigue, corrosion, cracking and / or crack growth, electrical arcing, general impacts, fluid flow (or the absence or fluid flow), changes in phase of a fluid flow, and changes in the rate of fluid flow.

[0010] The term “quasi-continuous monitoring” refers to monitoring that is substantially continuous, but not perfectly continuous. For example, the monitoring may not be perfectly continuous if one or more components of the sensor system (and, in particular, one or more components of the signal acquisition means) intermittently or periodically enter a power-saving state in which their power consumption is reduced. However, when the asset is being monitored quasi-continuously, the time spent in such a power-saving state is negligible in comparison to the periodicity and duration of the acoustic signals generated by the asset. Hence, no events occurring at the asset are missed by quasi- continuous monitoring.

[0011] The term “high frequency acoustic signals” preferably refers to acoustic waves with a frequency of the order of at least tens of kilohertz (kHz). More particularly, the high frequency acoustic signals acoustic waves may have a frequency of the order of hundreds of kHz. The high frequency acoustic signs may have a frequency of the order of megahertz (MHz). In general, a high frequency acoustic signal in accordance with the present disclosure have a frequency in the range of 10kHz to 500kHz, and preferably in the range of 100kHz to 500kHz.

[0012] In some implementations of the present disclosure, the signal processing means is a dedicated digital signal processing chip. In other implementations of the present disclosure, the signal processing means is a general-purpose microprocessor or microcontroller. The term “means for performing digital signal processing” should be construed accordingly.

[0013] The signal processing means has an awake state and a sleep state. In the awake state, substantially all the functions of the signal processing means are enabled. In the sleep state, some functions of the signal processing means are disabled. Thus, the signal processing means consumes less power in the sleep state than in the awake state. The signal processing means may have more than one sleep state, where each sleep state is characterised by different functions of the signal processing being disabled to achieve a particular balance between performance / functionality and power consumption. The signal processing means is the component of the sensor system with the highest power consumption. The average power consumption of the sensor system as a whole can, therefore, be significantly reduced by maintaining the signal processing means in a sleep state for as long as possible.

[0014] The average power consumption of the sensor system may be at least five times lower in the sleep state than in the awake state. When the average power consumption in the sleep state is significantly lower (e.g., at least five times lower) than in the awake state, the sensor system can be powered by batteries without the need for the batteries to be recharged or replaced on a frequent basis. For example, a battery-powered sensor system may be capable of monitoring an asset for many weeks, or even months or years, until its batteries require recharging or replacement. This can allow the sensor system to monitor assets in locations that cannot easily be served by mains power, without requiring an onerous maintenance schedule for the sensor system itself.

[0015] The difference between the power consumption of the sensor system in the sleep state and in the awake state is largely determined by the power efficiency of the signal processing means when in the sleep state.

[0016] Embodiments provide a sensor system designed for capturing and analysing acoustic signals from an asset. Embodiments comprise components for acquiring, monitoring and digitally processing such signals.

[0017] In embodiments, an acquisition component is configured for continuous or quasi- continuous capture of an acoustic signal from an asset. The acoustic signal may be captured passively by a ‘listening’ transducer.

[0018] In embodiments, a state transition mechanism is configured for comparing an amplitude of such acoustic signals with a predetermined threshold and generating a trigger signal when this threshold is reached.

[0019] In embodiments, a digital signal processing component is designed to transition from a low-power sleep state to an awake state in response, at least partially, to a trigger signal generated due to an event in such an acoustic signal. It is noted that digital signal processing tends to consume considerably more power than signal acquisition.

[0020] In embodiments, the state transition mechanism can compare an amplitude of the acoustic signal with multiple different thresholds and generate respective trigger signals indicative of different thresholds being reached. This multi-threshold comparison approach allows for more nuanced control over activation of digital signal processing.

[0021] In embodiments, a digital signal processing component receives these trigger signals along with respective timestamps indicating when they were received. This timestamped data provides valuable insights into sequence and timing of events within the monitored acoustic signals. Moreover, pattern matching can be performed on these timestamps to enable ‘intelligent’ activation of digital signal processing based on inferred events in the acoustic signal.

[0022] In embodiments, the state transition mechanism comprises a plurality of comparators. Each comparator may be configured to compare the acoustic signal with a respective one of the multiple different thresholds. The digital signal processing component may be configured to set a different threshold for each of the plurality of comparators.

[0023] In other embodiments, the digital signal processing component is configured to set a first threshold of the multiple different thresholds used by the state transition mechanism. The digital signal processing component may be configured to vary the power consumption of the sensor system by increasing or decreasing the first threshold. The signal processing means can increase or decrease the threshold to achieve a desired balance between power consumption and sensitivity.

[0024] In embodiments, a second threshold of the plurality of thresholds used by the state transition mechanism is a periodic signal. The state transition mechanism may be configured to generate a pulse width modulated signal by comparing the acoustic signal with the periodic signal. The system may further comprise a clock oscillator coupled to an input of a counter via a gate. The pulse width modulated signal may be provided as a control input to the gate. The digital signal processing component may be configured to read the value of the counter after transitioning from the sleep state to the awake state. In embodiments, the system includes components for demodulating acoustic (e.g., ultrasonic) signals upstream of comparison to the threshold(s). The amplitude of the demodulated signal may be compared with the threshold(s). Demodulation of high- frequency signals prior to comparison enables the comparison to be performed by relatively simple circuitry. The components for demodulating acoustic signals may include an envelope detector.

[0025] In embodiments, the system includes one or more components for conditioning acoustic signals upstream of demodulation.

[0026] In embodiments, the system is further configured for periodic transitioning into awake state by its digital signal processing component with periods ranging from minutes up to hours.

[0027] Such periodic activations enable regular data acquisition during prolonged sleep states thereby maintaining sufficiently consistent monitoring while still achieving substantial power savings.

[0028] The sensor system disclosed herein may advantageously be a wireless sensor system. That is to say, the sensor system may be powered by one or more batteries and, optionally, may be configured to report information to an external entity (such as a server) via a wireless communication link.

[0029] The skilled person will appreciate that except where mutually exclusive, a feature described in relation to any one of the aspects, examples or embodiments described herein may be applied to any other aspect, example, embodiment or feature. Further, the description of any aspect, example or feature may form part of or the entirety of an embodiment of the invention as defined by the claims. Any of the examples described herein may be an example which embodies the invention defined by the claims and thus an embodiment of the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Embodiments will be described, by way of example only, with reference to the accompanying drawings in which:

[0031] Figure 1 is a schematic representation of a sensor system according to a first embodiment of the present disclosure;

[0032] Figure 2 illustrates the output signals produced by comparators of the sensor system of Figure 1 ;

[0033] Figure 3 is a schematic representation of a sensor system according to a second embodiment of the present disclosure;

[0034] Figure 4 illustrates an input signal and the output of the PWM comparator of the sensor system of Figure 3; and

[0035] Figure 5 illustrates an input signal, the output of the PWM comparator and the output of the gate of the sensor system of Figure 3.

[0036] DESCRIPTION OF EMBODIMENTS

[0037] Various embodiments provide a system for continuously monitoring acoustic signals received through a medium such as a gas, liquid or a solid and identifying an event though profiling of the time and frequency domain.

[0038] Example features of such systems may include one or more of the following:

[0039] 1 . Continuous, or quasi-continuous, monitoring of acoustic signals in the 10KHz to 500KHz band for the purpose of detecting acoustic events.

[0040] 2. Very low current consumption for battery powered operation typically using a lithium D-cell.

[0041] 3. Acoustic events are identified by amplitude profiling and / or frequency domain profiling of the acoustic signal.

[0042] 4. Automatic adaptation to determine correct gain settings, threshold settings and frequency domain profiling settings.

[0043] 5. Event processor for event profile detection. With reference to Figure 1 , a sensor system 1 according to a first embodiment comprises an acoustic transducer 10, a variable gain amplifier 11 , an amplitude demodulator 13, a threshold detector 14 and a processing unit 16. The sensor system 1 optionally further comprises an analogue to digital converter (ADC) 15 and / or a communication module 17. The variable gain amplifier 11 , amplitude demodulator 13 and threshold detector 14 may all be implemented using discrete electronic circuits. For example, the threshold detector 14 may be implemented using operational amplifiers or comparators or, to reduce power consumption, the threshold detector 14 may be implemented using discrete transistors. The variable gain amplifier 1 1 , amplitude demodulator 13 and threshold detector 14 collectively form a low power continuous monitoring portion 2 of the sensor system 1. The ADC 15, processing unit 16 and communication module 17 may all be implemented on a single chip 4. The processing unit 16 is able to operate in either an awake state or a sleep state. The processing unit 16 consumes less power in the sleep state than in the awake state.

[0044] The acoustic transducer 10 receives acoustic signals coupled via a medium and produces a signal which is fed to the variable gain amplifier 11 . Internal to the amplifier 1 1 is a programmable filter 12 for selecting one or more specific frequencies. The acoustic transducer is selected based at least in part on its bandwidth, as it should be suitable for passively detecting acoustic signals of the order of at least tens of kHz, preferably hundreds of kHz, and in some cases MHz. Similarly, the programmable filter may be configured to select frequencies of up to 500kHz, up to 1 MHz, or higher. The programmable filter 12 may implement a plurality of bandpass filters, e.g. 50kHz-100kHz, 100kHz-150kHz, 150kHz-200kHz. The acoustic transducer 10 may be a piezoelectric transducer or any other type of transducer capable of converting an acoustic wave in the selected frequency range to an electrical signal.

[0045] The amplitude demodulator 13 demodulates the acoustic signal and feeds it to the threshold detector 14. It will be appreciated that the amplitude demodulator 13 separates relatively low-frequency information from high-frequency component of the signal. Various demodulation techniques would be suitable, such as envelope detector, and off- the-shelf componentry may be used. In embodiments where programmable filter 12 implements a plurality of bandpass filters, the amplitude demodulator 13 may receive and demodulate a subset of the bandpass filter outputs, for example just the output of the highest frequency bandpass filter. For the avoidance of doubt, the acoustic signal does not comprise a low-frequency information signal that has been amplitude modulated onto a higher frequency carrier signal. Rather, the acoustic signal inherently comprises components at different frequencies, ranging from relatively low frequencies (of the order of 1 kHz to 10kHz) to relatively high frequencies (of the order of 10kHz to 100kHz). It has been found that an amplitude demodulator 13 is well-suited to separating the low-frequency components of the acoustic signal from the high-frequency components. The amplitude demodulator 13 may be implemented using an envelope detector circuit. The signal output by the amplitude demodulator 13 is indicative of the status of the asset, but is more easily processed by downstream components than the original acoustic signal.

[0046] The threshold detector 14 comprises one or more comparators 6. In the example of Figure 1 , the threshold detector 14 comprises a plurality of comparators 6. The plurality of comparators 6 are arranged in an array. That is, the comparators 6 are connected in parallel with one another. Each comparator has a noninverting input, an inverting input and an output. The noninverting input of each comparator 6 receives the demodulated acoustic signal produced by the amplitude demodulator 13. The inverting input of each comparator 6 receives a respective threshold voltage. In one implementation, the respective thresholds of the comparators 6 are configurable by the processing unit 16 and are set in such a way that several discrete trigger levels are set. In more detail, the processing unit 16 may supply a different analogue DC voltage to the inverting input of each comparator 6. The processing unit 16 can output such analogue DC voltages even when it is the sleep state. In an alternative implementation (not illustrated in Figure 1 ), the respective threshold voltages of the comparators are fixed, rather than being configurable by the processing unit 16. In this implementation, the threshold voltages may be set by a potential divider network, which supplies a different analogue DC voltage to the inverting input of each comparator 6.

[0047] The output of each comparator 6 is provided as an input to the processing unit 16. When a comparator 6 trips (i.e., when the output of the comparator 6 changes due to the demodulated acoustic signal exceeding the threshold), it wakes up the processing unit 16 which captures a timestamp indicative of when the trip occurred. A plurality of comparators 6 may trip in a sequence, resulting in a corresponding series of timestamps at the processing unit 16. The processing unit additionally captures respective timestamps indicative of when the trips end, enabling calculation of the time for which a given comparator output remained high.

[0048] In one example, the processing unit 16 may be configured to detect a rising edge and / or a falling edge in an output of each comparator 6 when in the sleep state. Upon detecting a rising and / or falling edge, the processing unit 16 may generate an interrupt, which causes the processing unit 16 to transition from the sleep state to the awake state. Once in the awake state, the processing unit 16 may capture a timestamp by reading the current value on a counter (not shown in Figure 1 ), and saving that value to memory. In the case of a rising edge, the value of the counter is representative of the time at which the comparator 6 tripped. In the case of a falling edge, the value of the counter is representative of the time at which the comparator 6 ceased to be tripped. In both cases, the value of the counter is a timestamp. Other ways of waking the processing unit 16 and capturing a timestamp are possible, depending on the capabilities of the processing unit.

[0049] Figure 2 shows several graphs of voltage against time to illustrate how the comparators 6 of the sensor system 1 generate trigger signals 51 -58 in response to an acoustic signal 50. In this example, the trigger generator 14 of the sensor system 1 has eight comparators 6. The inverting input of each comparator 6 is supplied with a different threshold voltage by the processing unit 16. More specifically, a first comparator is supplied with a threshold voltage of 0.1 V, a second comparator is supplied with a threshold voltage of 0.2V, a third comparator is supplied with a threshold voltage of 0.3V and so on up to the eighth comparator, which is supplied with a threshold voltage of 0.8V. The envelope of the acoustic signal, which is produced by the amplitude demodulator 13 and applied to the noninverting input of each comparator 6, is indicated by reference sign 50. The envelope signal 50 is a single cycle of a 100Hz sine wave, with an amplitude between 0.0V and 1.0V. Initially, the amplitude of the envelope is 0.0 V and the outputs of all eight comparators are low (i.e., at logic zero). At 1 ms, the amplitude of the envelope signal 50 exceeds 0.1 V, which causes the first comparator to trip and generate a trigger signal 51 . That is, the output of the first comparator transitions from low to high (i.e. from logic zero to logic one). At 1 ,4ms, the amplitude of the envelope signal 50 exceeds 0.2V, which causes the second comparator to trip and generate a trigger signal 52. The remaining six comparators progressively trip as the amplitude of the envelope signal 50 increases, and generate respective trigger signals 53-58. At 4ms, the amplitude of the envelope signal 50 exceeds 0.8V, which causes the eighth comparator to trip and generate a trigger signal 58. All eight comparators generate a respective trigger signal 51 -58 until 6.1 ms. At 6.1 ms, the amplitude of the envelope signal 50 falls below 0.8V, which causes the eighth comparator to cease to generate trigger signal 58. That is, the output of the eighth comparator transitions from high to low. At 6.7ms, the amplitude of the envelope signal 50 falls below 0.7V, which causes the seventh comparator to cease to generate trigger signal 57. The remaining six comparators progressively cease to generate trigger signals 56-51 as the amplitude of the envelope signal 50 decreases. At 9ms, the amplitude of the envelope signal 50 falls below 0.1 V, and all comparators cease to generate a trigger signal. The trigger signals 51 -58 generated by the comparators are input to the processing unit 16. The processing unit 16 transitions from a sleep state to the awake state when it receives a rising edge and / or a falling edge of a trigger signal 51 -58. The processing unit 16 captures a respective timestamp indicating the time at which the rising edge and / or falling edge of each trigger signal 51 -58 occurred. A timestamp may be captured at the time at which a trigger signal 51 -58 was generated and / or at the time at which a trigger signal 51 -58 ceased to be generated.

[0050] The processing unit 16 assesses the series of timestamps to validate an event in the received acoustic signal. The processing unit 16 can perform pattern matching on the timestamps from the comparators 6 to validate events by using an algorithm such as a sliding window or a dedicated pattern matching engine. The system has one or more predefined patterns that represent valid sequence(s) of comparator triggers for the event(s) being monitored. The pattern(s) may specify, for example, an order of comparator activations and the expected time intervals between them, a number of activations of a given comparator within a period of time, the time for which a comparator output was high, or the like. The processing unit 16 can maintain a sliding window of a specific size based on the expected time interval between comparator triggers. As timestamps are generated, the processing unit 16 adds them to the window. The processing unit continually checks if the timestamps within the window match (one of) the predefined pattern(s). If the timestamps match (one of) the pattern(s) within the window, the event is considered validated. The window then slides forward to accommodate the next timestamp. If the timestamps do not match (one of) the pattern(s) within the allotted time window, the processing unit 16 can either flag the event as invalid, or reset the window and wait for a new pattern to begin. Alternatively, the processing unit 16 may implement a dedicated pattern matching engine. A simple example of an event which may be flagged as invalid is one whose duration is below a threshold. Once a valid event is identified in the received acoustic signal, the processing unit 16 causes relatively high-resolution capture of the acoustic signal in real time using an analogue to digital converter (ADC) 15.

[0051] The processing unit 16 performs more advanced digital signal processing on the digital signal generated by the ADC 15, such as frequency domain analysis, to classify the type of event. In some embodiments where programmable filter 12 implements a plurality of bandpass filters, the ADC 15 may be used to analyse a set of bandpass filter outputs which is different from a subset of the bandpass filter outputs received and demodulated by the amplitude demodulator 13. For example, the ADC 15 may be used to analyse all of the bandpass filter outputs, or all except the output of the highest frequency bandpass filter, for example.

[0052] Processed data, comprising inter alia final event results, are transmitted to receiving equipment via a communication module 17. There a several options for the communication module. Cellular networks provide expansive geographical coverage, making them ideal for geographically dispersed sensor deployments. They are suitable for real-time data delivery with minimal latency, but tend to carry ongoing subscription costs and potentially higher power consumption compared to other options. Low-Power Wide-Area Networks (LPWAN) technologies (including, but not limited to, LoRaWAN, Sigfox, NB-loT, ISA100 and WirelessHART) prioritize low power consumption, enabling extended sensor operation on lower power budgets, and tend to perform well in long- range transmission. Wi-Fi offers good bandwidth, making it well suited applications requiring high data rate transmission, but its range tends to be limited and it typically requires pre-existing network infrastructure for connectivity. Bluetooth Low Energy (BLE) can be suitable for battery-powered sensors with moderate data transmission requirements, offering a balance between power efficiency and moderate communication range. Satellite communication is an option for geographically remote deployments where other methods might be unavailable. For wired communication, options range from simple, low-power protocols like SPI and I2C for short-distance connections, to robust industrial solutions like RS-485 or 4-20mA for multi-sensor networks and Ethernet or CAN for high data rates and real-time needs. In a variation of the first embodiment of the sensor system 1 , the ADC 15 is omitted. In this variation, the sensor system 1 is unable to capture the acoustic signal at a high resolution. It has been found that the series of timestamps obtained by the processing unit 16 contain sufficient information about the acoustic signal to allow many types of event to be understood from the timestamps alone. Therefore, the ADC 15 can be omitted to reduce both the power consumption of, and the number of components in, the sensor system 1 .

[0053] A sensor system 20 according to a second embodiment will now be described with reference to Figure 3. The sensor system 20 comprises an acoustic transducer 10, a variable gain amplifier 11 , an amplitude demodulator 13, a threshold detector 34, a clock oscillator 30, a gate 31 , a counter 32, and a processing unit 16. The sensor system 20 optionally further comprises a communication module 17. The acoustic transducer 10, variable gain amplifier 1 1 , amplitude demodulator 13, processing unit 16 and communication module 17 are substantially the same as those discussed above in relation to the first embodiment of the sensor system 1 and, therefore, need not be described again. The threshold detector 34 of the sensor system 20 differs from the threshold detector 14 of the sensor system 1 .

[0054] The threshold detector 34 comprises a digital to analogue converter (DAC) 22, a trigger comparator 24, a waveform generator 26 and a pulse width modulation (PWM) comparator 28. The trigger comparator 24 and the PWM comparator 28 each have a noninverting input, an inverting input and an output. The demodulated acoustic signal produced by the amplitude demodulator 13 is applied to the noninverting inputs of both the trigger comparator 24 and the PWM comparator 28. The processing unit 16 applies a digital value representing a threshold to the input of the DAC 22. The DAC 22 produces an analogue voltage corresponding to the digital value, and applies that analogue voltage to the noninverting input of the trigger comparator 24. The waveform generator 26 is configured to output a periodic signal at a specific frequency. For example, the frequency of the periodic signal output by the waveform generator 26 may be of the order of 10-100kHz. The periodic signal output by the waveform generator 26 may be a symmetrical sawtooth wave (e.g., a triangle wave), an asymmetrical sawtooth wave, a sinusoidal wave or any other suitable waveform. Operation of the waveform generator 26 can be enabled or disabled by a control signal provided by the processing unit 16. The output of the waveform generator 26 is applied to the noninverting input of the PWM comparator 26. The outputs of the trigger comparator 24 and the PWM comparator 28 are supplied as inputs to the processing unit. The output of the PWM comparator 28 is also supplied as a control input to the gate 31 .

[0055] The clock oscillator 30 is a free running oscillator which is configured to output a square wave at a specific frequency. For example, the frequency of the clock oscillator 30 may be of the order of 10-100kHz. The output of the clock oscillator 30 is supplied, via the gate 31 , to the counter 32. The counter 32 is a digital counter which increments a value whenever it receives a rising edge from the clock oscillator 30. The current value stored in the counter 32 is supplied as an input to the processing unit 16. The processing unit 16 may optionally supply a reset signal to the counter 32, which causes the value stored in the counter to be set to zero. The gate 31 is connected between the output of the clock oscillator 30 and the input of the counter 32. The gate 31 is configured to isolate the output of the clock oscillator 30 from the input of the counter 32 when it receives a signal from the PWM comparator 28.

[0056] The operation of the sensor system will now be described with reference to Figures 4 and 5. In this example, waveform generator 26 generates a sawtooth wave 62. The envelope of the acoustic signal, which is produced by the amplitude demodulator 13 and applied to the noninverting inputs of the trigger comparator 24 and PWM comparator 28, is indicated by reference sign 60. The PWM comparator 28 compares the sawtooth wave 62 with the envelope signal 60. The output of the PWM comparator 28 is high when the amplitude of the envelope signal 60 is greater than the amplitude of the sawtooth wave 62. Conversely, the output of the PWM comparator 28 is low when the amplitude of the envelope signal 60 is less than the amplitude of the sawtooth wave 62. The PWM comparator 28 thereby generates a PWM trigger signal 64 having a mark-space ratio related to (e.g., proportional to) the amplitude of the envelope signal 60. The PWM trigger signal 64 is supplied as an input to the processing unit 16. The processing unit 16 transitions from a sleep state to the awake state when it receives a falling edge of the PWM trigger signal 64.

[0057] The PWM trigger signal 64 is also used to control the gate 31. More specifically, the gate 31 is open when the PWM trigger signal 64 is low, and closed when the PWM trigger signal 64 is high. The output of the gate 31 is indicated by reference sign 66 in Figure 5. The clock signal generated by the clock oscillator 30 is allowed to propagate to the counter 32 only during the ‘mark’ period of the PWM trigger signal 64 (i.e., only when the PWM trigger signal 64 is high). Hence, the counter 32 increments only during the ‘mark’ period. The processing unit 16 reads the value of the counter 32 when it transitions to the awake state in response to receiving a falling edge of the PWM trigger signal 64.

[0058] The processing unit 16 stores the value read from the counter 32. The processing unit 16 may analyse one or more of the stored values (e.g., using pattern matching) to identify an event. The processing unit 16 optionally stores a timestamp indicating the time at which the counter 32 was read. If a timestamp is stored, the timestamps of several values may be analysed as discussed above in relation to the first embodiment of the sensor system 1 .

[0059] At times when there are no events in the acoustic signal (e.g., when the asset is not generating an acoustic signal, or when the asset is generating a steady state acoustic signal), the amplitude of the envelope signal 60 is less than the analogue voltage applied by the DAC 22 to the noninverting input of the trigger comparator 24. At such times, the processing unit 16 is in a sleep state. Furthermore, the waveform generator 26 and / or the clock oscillator 30 may be disabled by the processing unit 16 to reduce the power consumption of the sensor system 20. When an event occurs, the amplitude of the envelope signal 60 exceeds the analogue voltage applied by the DAC 22 to the noninverting input of the trigger comparator 24, and the trigger comparator 24 generates a trigger signal. The processing unit 26 receives the trigger signal from the trigger comparator 24 and, in response, the processing unit 26 enables the waveform generator 26 and the clock oscillator 30. The processing unit 16 may also reset the counter 32 in response to receiving the trigger signal from the trigger comparator 24.

[0060] The processing unit 16 can adjust the threshold of the trigger comparator 24 by changing the digital value supplied to the input of the DAC 22. In particular, the processing unit 16 can increase the threshold of the trigger comparator 24 by increasing the digital value supplied to the input of the DAC 22. This increases the likelihood of the waveform generator 26 and / or the clock oscillator 30 being disabled by the processing unit 16 and thereby reduces power consumption of the sensor system 20. Conversely, the processing unit 16 can decrease the threshold of the trigger comparator 24 by decreasing the digital value supplied to the input of the DAC 22. This reduces the likelihood of the waveform generator 26 and / or the clock oscillator 30 being disabled by the processing unit 16, and thereby increases the sensitivity of the sensor system 20 at the expense of increased power consumption. The processing unit 16 can thus achieve a desired balance between power consumption and sensitivity by adjusting the threshold of the trigger comparator 24.

[0061] The processing unit 16 can be configured to enter a sleep state when the asset is generating a continuous acoustic signal (i.e., a signal whose amplitude and frequency spectrum are substantially time invariant) over a long period of time. The presence of a continuous acoustic signal implies that the asset is operating in a steady state and there are no events of interest to detect. Thus, at times when the asset is generating a continuous acoustic signal, the processing unit 16 can enter a sleep state and transition to the awake state upon receiving a falling edge of the trigger signal generated by the trigger comparator 24. The presence of a falling edge implies that the asset has ceased to generate a continuous acoustic signal, which may be indicative of an event such as a fault. Upon waking, the processing unit 16 reads the value of the counter 32 as previously described. Depending on the type of continuous acoustic signal, the processing unit 16 can reduce power consumption even further by disabling the waveform generator 26, disabling the clock oscillator 30, entering a deeper sleep state and / or setting a lower threshold of the trigger comparator 24. The threshold of the trigger comparator 24 is maintained even when the processing unit 16 is in a sleep state. For example, the digital value applied to the input of the DAC 22 may be latched, such that the operation of the trigger comparator 24 is unaffected by the processing unit 16 entering a sleep state.

[0062] As mentioned above, the periodic signal output by the waveform generator 26 can be symmetrical or asymmetrical. The asymmetrical sawtooth wave 60 shown in Figures 4 and 5 has a linear rising edge and a non-linear falling edge. More particularly, the falling edge of the asymmetrical sawtooth wave 60 shown in Figures 4 and 5 has an exponential decay. The presence of a non-linear portion in the periodic signal output by the waveform generator 26 can increase the effective resolution of the sensor system 20 when the envelope signals 60 has a relatively low amplitude. Either or both the rising or falling edge of the periodic signal output by the waveform generator 26 may comprise a nonlinear portion.

[0063] The sensor system 20 has fewer comparators than the sensor system 1 and, therefore, has lower power consumption and a smaller number of components. However, the sensor system 1 operates faster than the sensor system 20 and, therefore, is better suited to capturing events of shorter duration.

[0064] It will be appreciated, when reading the description above, that the types of acoustic event may vary both in amplitude and frequency content. As such, the systems have the ability of adaptation by going through a training process the results of which are used to adjusts the amplifier gain, the filter parameters and frequency domain settings accordingly.

[0065] It will be appreciated that sensor systems in accordance with embodiments of the present disclosure have a far lower average power consumption in the sleep state than in the awake state, typically at least five times lower, in some cases an order of magnitude lower, in some cases two or more orders of magnitude lower. No specific method is required for determining average power consumption in a given state. By way of nonlimiting example, the average power consumption of a sensor system in sleep and awake states can be determined through current measurement: a multimeter capable of measuring current in the milliamp (mA) or microamp (pA) range can be connected in series with a power supply providing the system's required voltage. The system is then configured to one of its operating modes (e.g. awake state) and then later to another of its operating modes (sleep state), with current readings logged at regular intervals for, say, several minutes in each state. Data logging software can be employed for extended measurements, for example. Subsequent analysis calculates the average current consumption for each mode, which is then converted to average power using the formula: Pavg= lavg* Vsuppiy, where Pavgis the average power consumption (in mW), lavgis the average current draw (in mA) and Vsuppiy is the supply voltage (in V). Other suitable methodologies will be readily apparent to the skilled person.

[0066] It will be understood that the invention is not limited to the examples and embodiments above-described and various modifications and improvements can be made without departing from the concepts described herein. Except where mutually exclusive, any of the features may be employed separately or in combination with any other features and the disclosure extends to and includes all combinations and sub-combinations of one or more features described herein. Further examples of the sensor system are described in the following paragraphs.

[0067] Example 1 . A sensor system comprising: means for acquiring an acoustic signal from an asset; means for monitoring the acoustic signal and generating a trigger signal in response to an event in the acoustic signal; and means for performing digital signal processing of the acoustic signal when in an awake state, wherein the signal acquisition means is configured for continuous or quasi- continuous monitoring of the asset based on high frequency acoustic signals, wherein the signal processing means is configured to transition from a sleep state to the awake state in response at least in part to the trigger signal, and wherein the average power consumption of the sensor system is at least five times lower in the sleep state than in the awake state.

[0068] Example 2. The system of Example 1 , wherein the state transition means (a) comprises means for comparing an amplitude of the acoustic signal with a threshold and (b) is configured to generate the trigger signal to indicate that the threshold has been reached.

[0069] Example 3. The system of Example 2, wherein the state transition means (a) comprises means for comparing an amplitude of the acoustic signal with a plurality of different thresholds and (b) is configured to generate respective trigger signals indicative of the different thresholds being reached.

[0070] Example 4. The system of Example 2, wherein the signal processing means is configured to receive the trigger signal and to obtain a timestamp indicative of when it was received.

[0071] Example 5. The system of Example 3, wherein the signal processing means is configured to receive the trigger signals and to obtain respective timestamps indicative of when they were received. Example 6. The system of Example 5, wherein the signal processing means is configured to perform pattern matching on the timestamps and to transition into the awake state in response to a match. Example 7. The system of Example 6, wherein the pattern matching is performed on a sequence of timestamps from one comparator.

[0072] Example 8. The system of any of the preceding Examples, further comprising means for demodulating the ultrasonic signal, wherein the comparison means is configured to compare the amplitude of the demodulated signal with the thresholds.

[0073] Example 9. The system of Example 8, further comprising means for conditioning the ultrasonic signal upstream of the demodulation means. Example 10. The system of any of the preceding Examples, wherein the signal processing means is further configured to transition into the awake state periodically, the period being of the order of minutes or hours.

Claims

CLAIMS1. A sensor system comprising: means for acquiring an acoustic signal from an asset; means for monitoring the acoustic signal and generating a trigger signal in response to an event in the acoustic signal; and means for performing digital signal processing of the acoustic signal when in an awake state, wherein the signal acquisition means is configured for continuous or quasi- continuous monitoring of the asset based on high frequency acoustic signals, and wherein the signal processing means is configured to transition from a sleep state to the awake state in response at least in part to the trigger signal.

2. The system of claim 1 , wherein the average power consumption of the sensor system is at least five times lower in the sleep state than in the awake state.

3. The system of claim 1 or claim 2, wherein the means for monitoring the acoustic signal and generating a trigger signal (a) comprises means for comparing an amplitude of the acoustic signal with a threshold and (b) is configured to generate the trigger signal to indicate that the threshold has been reached.

4. The system of any preceding claim, wherein the means for monitoring the acoustic signal and generating a trigger signal (a) comprises means for comparing an amplitude of the acoustic signal with a plurality of different thresholds and (b) is configured to generate respective trigger signals indicative of the different thresholds being reached.

5. The system of claim 4, wherein the means for comparing an amplitude of the acoustic signal with a plurality of different thresholds comprises a plurality of comparators.

6. The system of claim 5, wherein the signal processing means is configured to set a different threshold for each of the plurality of comparators.

7. The system of claim 3, wherein the signal processing means is configured to receive the trigger signal and to obtain a timestamp indicative of when it was received.

8. The system of any of claims 4 to 6, wherein the signal processing means is configured to receive the trigger signals and to obtain respective timestamps indicative of when they were received.

9. The system of claim 8, wherein the signal processing means is configured to perform pattern matching on the timestamps and to transition into the awake state in response to a match.

10. The system of claim 9, wherein the pattern matching is performed on a sequence of timestamps from one comparator.1 1 . The system of claim 4, wherein the signal processing means is configured to set a first threshold of the plurality of thresholds.

12. The system of claim 1 1 , wherein the signal processing means is configured to vary the power consumption of the sensor system by increasing or decreasing the first threshold.

13. The system of any of claims 4, 1 1 or 12, wherein a second threshold of the plurality of thresholds is a periodic signal.

14. The system of claim 13, wherein the means for comparing an amplitude of the acoustic signal with a plurality of different thresholds is configured to generate a pulse width modulated signal by comparing the acoustic signal with the periodic signal.

15. The system of claim 14, further comprising a clock oscillator coupled to an input of a counter via a gate, wherein the pulse width modulated signal is provided as a control input to the gate.

16. The system of claim 15, wherein the signal processing means is configured to read the value of the counter after transitioning from the sleep state to the awake state.

17. The system of any preceding claim, further comprising means for demodulating the acoustic signal.

18. The system of claim 17, further comprising means for conditioning the acoustic signal upstream of the means for demodulating the acoustic signal.

19. The system of claim 17 or claim 18, wherein the means for demodulating the acoustic signal comprises an envelope detector.

20. The system of any preceding claim, wherein the signal processing means is further configured to transition into the awake state periodically, the period being of the order of minutes or hours.

Images