Field

The present invention relates to a wireless unit for detecting impacts or impact events on a component, optionally an aircraft component. Such impact detection has applications in, for example, structural health monitoring. The present invention also relates to a method of detecting impacts or impact events using said wireless unit.

Background

It is often desirable to monitor the structural integrity of a component. One such approach is‘structural health monitoring’ (SHM), which can provide a remote assessment of the structural integrity of a component based on recorded sensor data and appropriate data processing and/or interpretation. Impact detection systems can be used as part of this structural health monitoring of a component, and typically consist of a network of sensors/actuators (which can remain fixed to the component to be monitored) and a controller unit for acquiring and/ or processing the data from the sensors, located locally or remotely to the sensors, depending on the application.

Whilst impact detection can be performed on both metallic and composite components, such detection is of particular importance for composite components. Composites absorb the energy of impact events, which can result in delamination, and ultimately failure, of the composite. For example, composites can experience various forms of damage (including indentation, de-lamination and fibre cracking) from impacts such as bird strikes, tool drops during manufacturing or maintenance, or stones striking the composite during taking-off or landing of an aircraft (even if the impacts are very low velocity). As such, significant degradation of the material properties of the composite can occur even when there is little or no visible damage from the impact.

Previous impact detection techniques generally fall into two categories: active sensing, which requires both sensors and actuators; and passive sensing, which requires only sensors (i.e. accelerometers or piezoelectric sensors) to monitor the status of the component to which they are attached and record acoustic or stress waves induced by impacts. Continuous passive (or active) sensing can provide a user with a prediction of the remaining lifetime of the component, for example, or help determine maintenance schedules. For example, continual monitoring of impacts can prevent unnecessary maintenance, since maintenance can be based on the present condition of the composite rather than being pre-emptive. In particular, continual monitoring can facilitate precise identification of the location and magnitude of impacts, making it possible to determine exactly what maintenance may be required. In turn, this can reduce the frequency of maintenance, and can also reduce both the amount of materials used and the time taken to carry out the maintenance. Such continual monitoring therefore has particular application in fields such as aviation (where maintenance can only be performed whilst the aircraft is grounded, during which time the aircraft is not in operation and therefore not generating revenue). However, whilst continuous monitoring allows impact events to be detected as they happen, and thus allows for real-time structural health monitoring, this approach unavoidably uses significant amounts of energy. In addition, since the shape, material, input force direction and velocity of the impactor are usually unknown, the subsequent analysis required to localise impacts can be computationally expensive, meaning the controllers or processors associated with such systems are typically large, heavy and power-consuming. Therefore, in order for an impact detection approach to structural health monitoring to be feasible for applications such as aviation, it is desirable to provide low power and low weight controllers or data loggers must be provided in order that said controller can be located on-board an aircraft without adversely affecting the performance of the aircraft (since any additional weight or power consumption affects fuel efficiency).

Many signal processing methodologies have been developed in an attempt to address these issues, including using artificial neural networks to localise impacts. Other approaches which attempt to solve issues of weight and power consumption, such as that in US 8,401,804 (Zhang), try to improve energy efficiency by utilising a trigger circuit within a structural health monitoring system. However, such approaches do not fully address the issues of power consumption and/or the computational resources used during impact detection. It is therefore desirable to provide a low power and low weight unit for data acquisition, which can be used in conjunction with existing sensing and data analysis systems.

Such a system also has utility in other applications, such as the monitoring of physical characteristics or structural integrity of components such as pipes. In particular, such a system would be desirable in any application where remote monitoring of a component structure is required. Summary

A first aspect is provided in accordance with appended independent system claim 1. A second aspect is provided in accordance with appended independent method claim 15. Optional features are provided in the appended dependent claims.

In the following description, a wireless unit for detecting impact events on a component is described. The wireless unit is optionally a wireless sensing unit for passively sensing or detecting signals from sensors. The wireless unit comprises a power source; a comparator module (or means for comparing) coupled to the power source and arranged to compare a characteristic of a sensor output signal received at the wireless unit to a predetermined threshold, the comparator module being further arranged to selectively output a trigger signal based on the comparison; a processor module (or means for processing) coupled to the power source and arranged, in response to receiving the trigger signal, to process the sensor output signal to generate a record of an event; and a wireless transmission module (or means for transmitting), wherein the processor module is arranged to provide at least part of the generated record to the wireless transmission module for transmission to a device external to the wireless unit (or external device). Optionally, the processor module is arranged to operate in a power-saving mode until receiving the trigger signal from the comparator module. The wireless unit implements an event-triggered mechanism, where a trigger signal‘wakes up’ the processor module only when processing of a signal is required. For example, processing of a signal can be required only when a sufficiently large event occurs on the monitored component (i.e. large enough to be detrimental to the component). At all other times, the processor module operates in a low power, or power saving mode. This can reduce power consumption since the processor module operates in a power-saving mode most of the time, entering a full power (or‘awake’) mode only when a large enough impact event occurs. Such a trigger-based approach is more energy efficient than impact detection methods which use‘always on’ devices, and is particularly suited to impact detection due to the rare, random and transitory nature of large impacts. The threshold can be predetermined based on the nature of expected events, or a specific size of event deemed to be detrimental to the component, for example. The wireless unit is provided with its own power source, e.g. a battery, a fuel cell or a power generation module such as an energy harvester, which allows the wireless module to be deployed in locations where no external power source is provided. Since the processor module (and optionally the wireless transmission module) are only triggered by (sufficiently large) impact events, the wireless unit has relatively low power consumption, so it is feasible for it to be powered by an internal power source (for example, to be battery powered). Moreover, the low power consumption means the power source can power the unit for a much longer period of time than always-on devices, which can allow the wireless unit to be deployed in remote locations (since the battery would not need to be frequently changed, or an energy harvester could generate electricity from ambient energy sources, such as vibration, airflow or light).

The provision of a wireless transmission module allows (at least part of) the generated record to be transmitted to another, external, device for further processing, such as a central (host) station for data repository and impact evaluation (localization or other analysis). The wireless transmission module can, for example, be arranged to transmit processed impact data (in the form of a generated record of the event) to a remote device located elsewhere on an aircraft on which the wireless unit is deployed for impact evaluation.

Since the processor module (or local processor) of the wireless unit only performs processing of the sensor output signal when triggered by an event on the monitored component (and even then only performs some processing of said signal, i.e. does not perform full analysis on the sensor output signal), power consumption by the processor is reduced. Moreover, by providing at least part of the generated record for

transmission, power consumption by the wireless transmission module can also be reduced (since the amount of data in the generated record can be much less than in the raw data from the sensors, which means the total amount of data needed to be transmitted by the wireless transmission module can be reduced). These features can thus contribute to the low power consumption performance of this wireless unit. The combination of a wireless unit which employs wireless transmission and is self- powered, or powered by its own internal power source, can also allow deployment of the wireless unit in remote or hard to reach locations, since no existing infrastructure is required when using the wireless unit. The wireless unit may optionally further comprise a filtering module (or means for hltering) arranged to filter the sensor output signal, wherein the comparator module is arranged to receive the sensor output signal from the filtering module (in order to compare the filtered sensor output signal to the predetermined threshold). The filtering module can increase the robustness of the wireless unit by eliminating operational noises (i.e. by attenuating ambient vibrations which manifest as noise within the sensor output signal, for example, low-frequency vibrations). In particular, by filtering out such (low-frequency) vibrations, a lower threshold can be used for generation of the trigger signal. As such, lower energy impact events can be detected. This can lead to an improved wireless unit for detecting impacts, since more impact events may meet the predetermined threshold criteria and thus be detected and processed by the processor module. The filtering module may additionally or alternatively be arranged to filter out other (for example, high-frequency) noises, depending on the application of the wireless unit and the ambient operating environment. Optionally, the processor module is arranged to process the sensor output signal for a predetermined period of time after receiving the trigger signal. The predetermined period of time may be fixed, or may be determined based on the sensor output signal. For example, a large impact with a high energy may lead to the processor processing the sensor output signal for a longer period of time than a lower energy impact. This can improve the robustness of the wireless unit, and the accuracy of any, subsequent, impact localisation analysis since sufficient information about every impact should be captured by the processor.

In some arrangements, the processor module is arranged to return to the power saving mode after the predetermined period of time. In other arrangements, the processor module is arranged to return to the power-saving mode after providing the at least part of the generated record to the transmission module. Both implementations reduce the power consumption of the processor module, and can thereby improve the energy efficiency of the wireless unit.

Optionally, the processor module is arranged to provide a signal to the transmission module after generating the record of an event, and, optionally, the transmission module is arranged to transition out of a power-saving mode in response to receiving the signal from the processor module in order to transmit the at least part of the generated record to the external device. Optionally, the transmission module is arranged to return to the power saving mode after transmitting the at least part of the generated record to the external device. These optional implementations can help to further reduce the power consumption of the wireless unit, since the transmission module operates in a low-power or power saving mode when not in use (in the same way that the processor does).

Optionally, the processor module is arranged to both record the sensor output signal and process the sensor output signal. The processing can be in real-time with respect to the recording, or subsequent to the recording. Optionally, the wireless unit further comprises a storage module (or means for storing), wherein the processor module is further arranged to cause the storage module to store at least part of the generated record (and/or the recorded sensor output signal, where appropriate). Such an arrangement may be useful when the external device is out of communication range of the wireless unit (i.e., the wireless transmission module is not able to transmit the at least part of the generated record to the external device). Instead, the wireless unit can store at least part of the generated record until such time as the record can be provided to an external device for analysis. This arrangement can therefore ensure that information is not lost simply because the wireless transmission module is not capable of transmitting to an external device at any given time, and thus can help provide a more robust unit.

Optionally, the generated record of an event comprises one or more parameters extracted by the processor module from the sensor output signal. These parameters may comprise at least one of: a time of arrival of the sensor output signal, an energy of the sensor output signal, a voltage of the sensor output signal, an amplitude of the sensor output signal, an energy of a first peak of the sensor output signal, a voltage of the first peak of the sensor output signal, and an amplitude of the first peak of the sensor output signal. One or more of these parameters may be used to detect impact event(s). For example, by analysing the time of arrival of signals from each of a plurality of sensors, the times of arrival can be correlated or otherwise analysed to localise the impact event. Furthermore, by analysing an amplitude or magnitude of signals from each of a plurality of sensors, both a location and energy of the impact event can be determined.

Optionally, the processor module is further arranged to calculate an estimate of the sensor output signal between a time the event occurred and a time the processor began to process the output sensor signals. In other words, sensor output signals which were not processed by the processor because it was in a low-power or power saving mode can be recreated. This arrangement can thus provide data from the initial stages of the impact event, which can facilitate improved localisation of impact events. Optionally, calculating an estimate comprises applying a data recovery algorithm, e.g. a moving average method to the sensor output signals processed by the processor (in other words, the processor is arranged to apply a moving average). Utilising a moving average can also smooth the signals processed by the processor, which can improve extraction of parameters such as time of arrival or the first peak. Optionally, the wireless unit further comprises a housing. The housing is arranged to enclose the power source, the comparator module and the processor module. In other words, the wireless unit is a wireless, self-contained, unit for impact detection. In some arrangements, the housing may also enclose at least a part of the wireless transmission module, although in other arrangements the wireless transmission module may be provided external to the housing in order to improve transmission to the external device (i.e. to avoid any shielding effect from the housing). The housing may further at least partially enclose the filtering module and/or a rectifier module, as appropriate.

Optionally, a system comprising the wireless unit and a sensor network is provided. Sensors of the sensor network are arranged to detect vibrations of a component with which they are associated (for example, to which they are coupled) and to output the sensor output signal to the wireless unit (optionally wirelessly), the sensor output signal representing the vibrations of the component. Any suitable sensors maybe employed, for example passive sensors such as accelerometers or piezoelectric sensors may be provided to continually monitor the status of the component and record acoustic or stress waves induced by impact events. Passive sensors can reduce the power consumption of the sensor network, since no activation power needs to be applied to the sensors. Alternatively, an active sensor network of sensors and actuators can be used.

A method for detecting impact events on a component using the above described wireless unit of the first aspect is also described. The method comprises, receiving, at a wireless unit comprising a power source, a sensor output signal from a sensor network; comparing, at the wireless unit, a characteristic of the received sensor output signal to a predetermined threshold; selectively outputting a trigger signal based on the comparison; causing a processor of the wireless unit, in response to receiving the trigger signal, to transition out of a power saving mode; processing, at the processor, the sensor output signal to generate a record of an event; and providing, to a wireless transmission module of the wireless unit, at least part of the generated record for transmission to a device external to the wireless unit (i.e. to an external device).

Optionally, receiving the sensor output signal comprises wirelessly receiving the signal. In other words, the sensor network comprises a network of sensors (and optionally actuators) and a transmission module for wirelessly transmitting signals from the sensor network to the wireless unit. Alternatively, the wireless unit is directly connected or coupled to the sensor network through a wired connection. A network of multiple wireless units and sensor networks can be provided which can map or monitor multiple components; the sensor output signals are locally processed by the associated wireless unit and then transmitted to one (or more) external device(s). Optionally, before comparing a characteristic of the received sensor output signal to a predetermined threshold, the method further comprises filtering, at the wireless unit, the sensor output signal. Such filtering can improve the robustness of the wireless unit, and of the method, by allowing lower energy impact events to be detected (and therefore providing a fuller picture of impact events on the component). When the sensor output signals are filtered, the step of comparing a characteristic of the sensor output signal instead comprises comparing a characteristic of the filtered sensor output signal to the predetermined threshold.

Optionally, the method further comprises transitioning the processor back into the power-saving mode after providing the at least part of the generated record to the transmission module. In combination with the processor being in the power saving mode before receiving the trigger signal, this feature can reduce the power

consumption of the device. This feature can thus provide a low-power method for detecting impact events, which can have application in fields such as aviation.

Optionally, generating the record of an event comprises extracting one or more parameters from the sensor output signal. Optionally, the one or more extracted parameters comprise at least one of: a time of arrival of the sensor output signal, an energy of the sensor output signal, a voltage of the sensor output signal, an amplitude of the sensor output signal, an energy of a first peak of the sensor output signal, a voltage of the first peak of the sensor output signal, and an amplitude of the first peak of the sensor output signal. Optionally, the method further comprises calculating, at the wireless unit, an estimate of the sensor output signal between a time the event occurred and a time of causing the processor to process the sensor output signal. Optionally, the method further comprises storing at least part of the generated record at a storage module of the wireless unit. Additionally or alternatively, the method optionally further comprises wirelessly transmitting, from the wireless unit, at least part of the generated record to the external device. In some arrangements, the transmitting occurs in real time. This can enable real-time localisation of impacts, which may be critical to performance of an aircraft or other vehicle comprising the wireless unit, for example. In other arrangements, the transmission occurs at a later time than the impact and it is the stored record (or at least part of the generated record stored in the storage module) which is transmitted. For example, impact events maybe monitored whilst the aircraft is in the air but the generated record, or part of the generated record, may not be transmitted until the aircraft has landed).

Once the transmission to an external device has occurred, the method may further comprise localising and/or analysing the impact event based on at least part of the generated record. Such localisation could occur either on board an aircraft (for example, localisation can occur in real-time after transmission of the record to an external data processor already installed on the aircraft), or once the aircraft had landed (where, for example, the wireless unit can be interrogated by an external device in order to extract the at least part of the generated record). Analysing the impact event can comprise determining at least one of: an energy, an input force and/or velocity, and a shape of the impactor.

It will be understood that one or more of the comparator module (or means for comparing), the processor module (or the means for processing), the transmission module (or the means of transmitting), and the filtering module (or the means for filtering) may be integrated or otherwise form part of a single processor or data logger component coupled to the power source and at least partially enclosed within the housing. Alternatively, one or more of said modules may be provided independently of the other modules in separate processing component(s) coupled to the power source and at least partially enclosed within the housing. Moreover, any of the above described optional features may be combined with any of the other features described herein, in any suitable combination, provide there are no contradictions within such a combination. It will be understood in particular that features described with reference to the first aspect may be applicable to the second aspect and may be combined with any features described with reference to the second aspect, and vice versa.

Brief description of the drawings

The below description is reference to the following Figures:

Figure l is a schematic illustration of a system comprising a wireless unit and a sensor network, and a device external to the system;

Figure 2 is a schematic illustration of the system of Figure l;

Figure 3 illustrates an example comparison process performed by the comparator of the wireless unit;

Figure 4 provides experimental results illustrating detection of an impact, based on a trigger threshold, and the subsequent processing of sensor output signals by the wireless unit;

Figure 5 provides experimental results illustrating the power consumption of a wireless unit as described herein, as compared to an‘always-on’ device; and

Figure 6 illustrates an example method for detecting impact events on a component.

Detailed description

With reference to Figure 1, a system 100 comprising a wireless unit 104 and a sensor network 102 is provided. Here, wireless unit 104 is a wireless sensing unit for passively sensing sensor network 102. System 100 is a local system for impact detection which may be located on-board an aircraft, for example. Local system 100 can be combined with an external device 108 to form an overall impact detection system, for example, when external device 108 is brought into communication range of the wireless unit 104 of system 100.

The wireless unit 104 is arranged to receive a sensor output signal 116 from the sensor network 102 and process said sensor output signal 116. The sensor network 102 is directly coupled by wired connection to wireless unit 104, but in other examples the receiving by the wireless unit 104 can be by any suitable wired or wireless means. The sensor output signal 116 provides information indicative of vibrations and/or impacts on a component to which the sensor network 102 is attached. The sensor output signal 116 is processed by the wireless unit 104 to generate a record 122 of an impact event. The generated record 122 is provided to a wireless transmission module 124 of the wireless unit 104 for transmission to the external device 108 (i.e. a device external to the wireless unit) for post-processing, such as impact localization and estimation of the impact magnitude.

The external device 108 may be located near wireless unit 104 at all times to form the impact detection system, or may be brought into wireless communication range of wireless unit 104 only when the generated record 122 is to be collected and/or analysed. For example, when system 100 is implemented in an aircraft, external device 108 may be brought into wireless communication range with transmission module 124 of wireless unit 104 only after the aircraft has landed (during a routine maintenance schedule, for example). Alternatively, external device 108 maybe permanently located on the aircraft, but remote from wireless unit 104. External device 108 may be, for example, a central host or central data acquisition system.

Sensor network 102 can comprise any suitable set or array of sensors (and optionally actuators) affixed to a component in any manner which allows for evaluation of the component. The sensors, or sensing elements, are capable of detecting vibrations of a component and can either passively monitor a structure for stress waves resulting from an impact (analysis of such stress waveforms can enable information about any corresponding damage to be determined), or monitor the structure for stress waves actively transmitted through the structure by actuators (comparison of the resulting waveforms to the original stress waves can indicate damage).

Sensor network 102 as described herein is a passive sensing array. However, in other examples, the sensor network may by an active sensing array and comprise multiple actuating and sensing elements (e.g. actuators and sensors) placed at discrete locations on the component for transmitting stress waves through the component and detecting resulting waveforms, respectively. One such suitable sensor/actuator is a lead zirconate titanate piezoelectric transducer, which can act as both a sensor and an actuator. Each piezoelectric transducer converts electrical signals to stress waves in order to actively query a component, and converts resulting detected stress waves to electrical signals, which can be sent to the wireless unit for analysis. Such a lead zirconate titanate piezoelectric transducer can also be used within passive sensor network 102, where the sensor output signals 116 received by wireless unit 104 are the electrical signals produced by the transducer, but it will be understood that any other suitable sensor maybe used within sensor network 102. With reference to Figure 2, system 100 is described in more detail. Wireless unit 104 comprises a housing (not shown), a processor module, or processor, 106 and a comparator module, or comparator, 114 in addition to wireless transmission module 124. Wireless unit 104 also comprises a power source (not shown) coupled, or connected, to the comparator module and the processor module (either directly or indirectly) and arranged to supply power to said modules. The modules and the power source are at least partially enclosed within the housing of the wireless unit 104. For example, the processor, comparator and power source may be fully enclosed within the housing, and the wireless module may be at least partially enclosed within the housing, or may also be fully enclosed within the housing.

The processor 106 can be any suitable processing device (e.g., a processor core, a microprocessor, an ASIC, a FPGA, a controller, a microcontroller, etc.). Processor 106 can be one processor or a plurality of processors that are operatively connected. The different modules described herein maybe implemented as part of processor 106, or may be provided separately, but operatively connected to, processor 106.

Wireless unit 104 is arranged to receive sensor output signal 116 from sensor network 102, here a network of sensors mounted directly on component 110. Sensor network 102 comprises N sensors (102# 1 to 102#N) in this example, but as discussed above, sensor network 102 can comprise any suitable number of sensors or sensing elements (i.e. one or more), as required. Sensor output signal 116 may comprise signals from one or all of the sensors of the sensor module 102. In some examples, sensor output signal 116 can comprise two sets of signals, signals 116a and signal 116b; comparator 114 may receive signals 116a and processor 106 may receive signals 116b. Alternatively, processor 106 may receive signals 116a via the comparator 114 and no signals 116b are provided. Signals 116a and 116b may be the same, for example the signals from the sensor module 102 may simply be split and provided to two different components of the wireless unit 104. Alternatively, the signals 116a, 116b maybe provided via different signal paths and represent different, or differently processed, information about the component 110, as described with reference to Figure 2. In some examples, wireless unit 104 may further comprise a filtering module 112.

Filtering module 112 may act on some, or all, of the signals received at the wireless unit 104. For example, the filtering module 112 may act only on signals 116a which are to be used by the comparator 114, and signals 116b to be subsequently processed by processor 106 may not be filtered by the filtering module 112. Alternatively, all signals 116 may be filtered by the filtering module 112, as illustrated in Figure 2. In some arrangements, the wireless unit 104 does not comprise a filtering module, but filtering maybe performed at the sensor network 102, for example using in-built electronics of the sensor(s) of the network. When the filtering is performed at the sensor network 102, signals 116a (and optionally signals 116b) arrive at the wireless unit 104 already filtered.

Where appropriate, signals 116a and/or signals 116b may also be rectified as well as, or instead of, being filtered in order to convert said signals from alternating to direct current. Such rectification can be combined with any features of the first and/ or second aspects. In the example described with reference to Figure 2, a rectifier module 126 may be provided within the wireless unit 104. The rectifier module 126 is arranged to receive filtered signals from filtering module 112 and rectify the signals to form signals 116b which are to be provided to the processor 106. In this example, signals 116a which are to be provided to the comparator 114 from filtering module 112 are not rectified. However, only signals 116a, or both signals 116a and 116b may be rectified by rectifier module 126, as appropriate. Alternatively, rectification may be applied at the sensor network 102 such that rectified sensor output signals are received at the wireless unit 104. Filtering is preferably used in order to minimise the influence of ambient vibrations, which register as noise in the sensor output signals. For example, in order to minimise the influence of ambient vibrations when wireless unit 104 is used in aviation application, a high pass filter with a cut-off frequency of 2 kilohertz (kHz) is applied (ambient vibration testing conditions for a turbojet are normally below 2 kHz, see DO- 160 Environmental Conditions and Test Procedures from Radio Technical Commission for Aeronautics, RTCA). In order to reduce power consumption, which would enable the filtering module to operate continuously, a passive high pass filter is preferably used. Such a filter can effectively eliminate the influence of local, low-frequency, vibrational noises without requiring any power supply. It will be understood that, depending on the application to which the wireless unit 104 described herein is applied, a different filter maybe applied by/ within the filtering module (for example, a high- pass, low-pass, or bandwidth filter with any suitable cut-off frequency). Filtering module 112 can comprise a filter circuit arranged to filter the signals.

Once sensor output signals 116a are received at the comparator 114, the comparator 114 compares one or more characteristics of the sensor output signal to a predetermined threshold 118 (when the sensor output signals received by the comparator 114 are filtered, the comparator module 114 is instead arranged to compare the characteristics of the filtered sensor output signal to the threshold 118). The characteristic(s) used in the comparison may be a voltage or an energy of the sensor output signal 116a, or any other suitable characteristic which is representative of the signal.

This comparison process is described further with reference to Figure 3, which shows an example design schematic of the comparator module 114. As with the filtering module, the comparator 114 described herein has low power consumption in order that the wireless unit 104 can be deployed on board an aircraft, for example, as a stand- alone, wireless, product (i.e., without drawing power from the aircraft itself).

Furthermore, a high responsiveness of the comparator 114 (i.e., a short latency between comparison of the characteristic of the sensor output signal to threshold 118 and the output of the trigger signal 120) is required in order to reliably detect impact events. To achieve this reliable, event-triggered, detection of impacts the comparator operates in an always on, or full power, mode, so it is important that power consumption of the comparator is minimised.

The comparison of a characteristic of the sensor output signal 116a to the threshold 118 can be implemented by one or more operational amplifier (op-amp) comparators. For example, a comparator such as the LM339A by Texas Instruments maybe used, which component has a quick response time. Each input channel (Ch 1, Ch 2, ... , Ch n) of the comparator module receives a sensor output signal from a different sensor and is connected to a comparator circuit for comparison of a characteristic of the received input signal to threshold 118. For example, a voltage (optionally, voltage data after the filtering and/ or rectifying described above has been applied by the wireless unit 104 or the sensor network 102) of each sensor output signal (the characteristic) may be compared to a reference voltage (the predetermined threshold 118). However, any other suitable characteristic of the sensor output signals (i.e. any characteristic other than the voltage) can be compared to any other suitable predetermined threshold. In one example where a characteristic of the voltages of the signals from the different sensors are compared with a reference voltage, reference voltage 118 is generated by a voltage divider circuit (as shown within the dashed box of Figure 3). The reference voltage threshold 118 can be calculated using the following equation:

Vcc * R2 / (R1+R2),

where Ri, R2 are the resistances of the resistors of the voltage divider circuit and Vcc is the voltage from a power supply arranged to supply power to the comparator module. For example, for one application of the above described wireless unit, Ri = 7.4 MW and R2 = 680 1<W. By changing the ratio of R1/R2, the reference voltage 118 can be adjusted, which means the energy of the impacts to be detected can be adjusted. The predetermined threshold 118 can be adjustable by a user, or user adjusted.

The value of resistor Rt of the comparator module 114 shown in Figure 3 also needs to be carefully selected because any components within the comparator module consume power, which can affect the energy efficiency of the wireless unit 104. For resistor Rt, low power consumption of the comparator module can be achieved with a large resistance, but lower resistance values can reduce the response time of the comparator module in generating and outputting the trigger signal. As the system responsiveness can be more important than power consumption (capturing initial details of the impact events is critical for effective impact detection and analysis), and the current used when the comparator module is active can be averaged out by the extremely short period of time for which the comparator module is active, this resistor can be selected to have a resistance Rt = 1 1<W, for example. The quiescent current (i.e. the inactive current, or the current drawn when no impact events occur) of the whole comparator module 114 was measured to be 0.81 mA (i.e.), when the above resistor values were used.

With further reference to Figure 2, the comparator module is arranged to selectively output a trigger signal 120 based on the comparison of the characteristic of the sensor input signals 116a with the threshold 118. For example, the comparator module 114 may output the trigger signal 120 if the voltage of the sensor input signal (the characteristic) is greater than or equal to the reference voltage 118 (the threshold), or if the voltage is less than the reference voltage. Alternatively, any other suitable characteristic of signals 116a may be compared to a predetermined threshold and a trigger signal 120 selectively output based on the comparison. After generation by the comparator module 114, trigger signal 120 is output to, and then received by, processor 106. In response to receiving the trigger signal 120, processor 106 is arranged to transition from a power-saving mode (i.e. a low-power or a ‘sleep’ mode) to a full power mode (‘awake’ or‘wake-up’ mode). In other words, the processing module 106 stays in the low-power or power saving mode for the most time, but‘wakes up’ or transitions to a full power mode (almost) immediately upon receipt of the trigger signal 120. A quick transition between power modes is important if initial impact event data is to be effectively captured or recorded, in other words, the processor modules should‘wake up’ from the power saving mode as soon as possible.

Once the processor has transitioned out of the power saving mode in response to receiving the trigger signal, the processor 106 is arranged to process the sensor output signals from the sensor module 102. As discussed above, the sensor output signals processed by the processor 106 may be received directly from the sensor network 102 or from another component such as the rectifier module 126 (i.e. signals 116b), or maybe received via the comparator 114 (signals 116a). The processor 106 should preferably employ a high sampling rate in order to record the transitory impacts. Moreover, the number of available channels should be as high as possible in order to allow more sensors of sensor module 102 to be connected to processor 106 of wireless unit 104, since this allows a larger area of component 110 to be monitored by a single system 100. Optionally, the processor 106 records the voltage (or other characteristic) from each input channel on which the sensor output signals 116 are received using one or more analog-to-digital converters (ADCs), optionally, built-in ADCs. Optionally, processor 106 can be implemented as a microcontroller (or microcontroller unit,‘MCU’), which can provide a low weight and low power consumption processor.

In one example implementation, the local processing functions of processor 106 may be implemented using microcontroller STM32L476ZE from STMicroelectronics. This MCU has 24 ADC channels with a sampling rate of up to 5.33 million samples a second. It has seven different low-power operation modes, including: sleep, stop, standby shut- down mode, etc. The MCU is based on the high-performance ARM® Cortex®-M432- bit core operating at a frequency of up to 80 MHz.

As discussed above, in passive and low-power applications of the wireless unit 104, the power consumption in any power saving mode of the processor 106, and the wake-up time of the processor, are critical to overall performance. Table 1 below summarises the different low-power modes and their operational characteristics of microcontroller model STM32L476ZE: low-power modes stop 1 and stop 2 both present a good balance between current (i.e. power) consumption and wake-up time. However, it will be understood that any suitable microcontroller or processor may be used as processor 106. Preferably, any such processor has a current consumption in the low-power mode of less than 10 micro Amperes (mA), optionally less than 7 mA, optionally less than 2 pA. Preferably, any such processor also has a wake-up time (i.e. the time taken to transition from a low-power mode to a fully awake made) of less than 10 microseconds (ps), optionally less than 8 ps, optionally less than 5 ps.

Low -power mode C urrehl c ons urnpti on Wake-un lime

Fable 1: the different low power modes and their associated operational characteristics for microcontroller model SFM32L476ZE (SFMicroelectronics), with and without a

Real Fime Clock (RFC).

Fhe process of detection of an impact and generation of a trigger signal 120 by the comparator 114, and the processing of the sensor output signal by the processor 106, is described in more detail with reference to Figure 4, which provides experimental results of testing of the system 100 under vibrational conditions expected in the operational environment of an aircraft. Ambient vibrations were generated by a closed- loop vibration set-up consisting of a FMS 2110E shaker and a 2050E09-FS power amplifier. A composite plate comprising a piezoelectric sensor was mounted on the shaker. Fhe vibration frequency was varied from 10 Hz to 700 Hz. An impact was induced at a time t=o ms, as is indicated by the dotted line and label in Figure 4.

As shown in Figure 4(a), the amplitudes of the ambient vibrational noise (i.e. the signals from before the impact event) can be as high as lV. If the filtering module 112 (and here the additional rectifier module 126) is not utilised, the trigger threshold 118 has to be set to a relatively high value, otherwise local processing module 106 can be easily woken up by ambient noises even though no impact has occurred. If the trigger is not set sufficiently high, false alarms (and thus unnecessary power consumption) can arise as a consequence of the noise being detected.

By applying the sensor output signals 116 through the filtering module 112, low- frequency noises can be removed effectively, as shown in Figure 4(b), where the trigger threshold 118 was set to 0.25 V (see dotted line and label). The processor 106 therefore stays in the low-power mode, even though the amplitude of the unfiltered noise would be larger than the trigger threshold, as indicated in Figure 4(a). The applied filter (i.e. the filter applied by the filter circuit of the filtering module) therefore allows a lower threshold for triggering processor 106 to be applied by wireless unit 104, because the low-frequency vibration events are removed by the filtering process. A lower predetermined threshold 118 can be beneficial, since it can allow lower-energy impacts to be detected. A lower threshold 118 for triggering the processor also ensures that the processor 106 transitions from the power saving or low-power mode at an earlier stage of the impact, and is therefore able to record or process more of the impact event data; data from the initial stages of the impact can be of particular benefit in determining time of arrival and the first peak of the signal, for example.

Figure 4 (c) illustrates the signal 116b recorded by the local processing module 106. It can be seen that in this example noise due to ambient vibrations has been eliminated by passing the signals from each sensor channel through filtering module 112 and rectifier module 126 as described above with reference to Figure 2. In other words, processor 106 can process or record the filtered, rectified, sensor output signals. However, some data from the initial stages of the impact is lost due to the wake-up delay of the processor (i.e., the delay in the processor transitioning from the power saving mode to the fully awake made). Therefore, in some instances processor 106 is arranged to calculate an estimate of the signals between a time the event occurred and a time the processor began to process the output sensor signals. In some examples, calculating an estimate can comprise applying a data recovery method, e.g. a moving average method, to the output sensor signal 116 to estimate the data from immediately after the event occurred which was not originally processed. With further reference to Figure 2, processor 106 is arranged to process the sensor output signals received and generate a record of an event 122, for example an impact event on the component 110, from the sensor output signals 116. The generated record 122 may comprise one or more parameters extracted by the processor 106 from sensor output signal 116. The one or more extracted parameters may comprise at least one of: a time of arrival of the sensor output signal, an energy of the sensor output signal, a voltage of the sensor output signal, an amplitude of the sensor output signal, an energy of a first peak of the sensor output signal, a voltage of the first peak of the sensor output signal, and an amplitude of the first peak of the sensor output signal.

At least part of the generated record 122 is provided to the transmission module 124 for transmission to the external device 108. Providing at least part of the generated record 122 may therefore comprise providing at least one of these extracted parameters to the transmission module 124 for transmission to the external device 108. Wireless transmission module 124 is optionally a wireless transmission module which utilises the Zigbee communication specification (Zigbee is a low-power, low data rate, and close proximity, i.e., personal area, wireless network, and thus is suitable for low-power applications such as continual monitoring of aircraft components). However, any other suitable communication specification(s) or protocol(s) can be employed, as required (for example, Bluetooth or Bluetooth Low Energy).

The one or more extracted parameters provided to the transmission module 124 are the (at least) part of the generated record 122 necessaiy for performing the desired impact detection and localisation analysis. For example, the time of arrival (or ToA) can be extracted from each of the processed sensor output signals 116 and provided to the transmission module to enable impact localisation to be performed by the external device 108. By providing at least part of the generated record 122 of the impact event to the wireless transmission module, rather than the sensor output signals, power consumption can be minimised both at the processor 106 (in providing the data to the wireless transmission module 124 for transmission) and at the transmission module (in providing the record to the external device), since the generated record can contain less data than the raw sensor output signals.

In some arrangements, wireless unit 104 may further comprise a storage module comprising memory, which is arranged to store the generated record 122 of an event. The memory can include one or more non-transitory computer-readable storage mediums, such as RAM, ROM, EEPROM, EPROM, flash memory devices, magnetic disks, etc., and combinations thereof. The memory is connected to processor 106 and can store data and instructions, which instructions can be executed by the processor 106 to cause the processor 106 to perform operations (for example generating record 122 and providing the record 122 to the wireless transmission module 124).

Processor 106 may provide the at least part of the generated record 122 to the transmission module 124 only when transmission module 124 is capable of

transmitting the at least part of the generated record to the external device 108. For example, when the external device 108 is not local to the transmission module of wireless unit 104, the processor 106 may cause the storage module to store the generated record until such time as the external device 108 is close enough for the transmission module to transmit the at least part of the generated record to said external device, at which time the processor may provide some or all of the generated record to the wireless transmission module.

Alternatively, processor 106 may automatically transmit the at least part of the generated record 122 to the transmission module 124 in real time, i.e., without storing the record 122 in permanent memory, and transmission module 124 may also automatically transmit the received at least part of the generated record to the external device 108. It will be understood that this arrangement may only be implemented when the external device is located in the vicinity of the transmission module (i.e., it is located on-board an aircraft along with system 100, or has been brought near system 100 when the aircraft is grounded).

To further reduce power consumption, the processor module may be arranged to provide a signal to the transmission module 124 after generating the record 122, which signal causes the transmission module, upon receipt, to transition out of a power saving mode. In other words, the transmission module 124 also operates in a low-power power saving mode most of the time, only entering a fully awake mode after an impact event has been detected by the comparator 114. In this arrangement, the processor module controls the operational mode (active or power-saving) of the wireless transmission module. The processor maybe arranged to return to its power saving mode after providing the at least part of the generated record to the transmission module (or, optionally, after receiving a confirmation indicating that the at least part of the generated record has been successfully received by the external device). The transmission module may be arranged to return to its power saving mode after transmitting the at least part of the generated record to the external device (or, optionally, after said confirmation is received). In other arrangements, where the processor 106 is arranged to cause the storage module to store at least part of the generated record of an event 122, for example because the external device 108 is not within communication range of the transmission 124, the processor may be arranged to return to its power saving mode after a predetermined period of time from receipt of the trigger signal 120 from the comparator 114. The predetermined period of time may be determined based on the sensor output signal 116 received by the processor 106. For example, large or high- energy impact events may cause the processor 106 to processor signals for a longer period of time than a shorter or lower energy impact event. With reference to Figure 5, the power consumption of a wireless unit 104 as described herein is compared to a device which is always on (i.e., does not transition from a low- power mode to a full power mode). The power consumption is modelled as a function of the percentage of input events over time. According to Figure 5, when the impact activity is low the low-power wireless unit 104 has much lower current consumption than the current consumption of an‘always-on’ device. Consequently, the lifetime of the power source (i.e. a battery) can be extended significantly as compared to a device implemented without a trigger arrangement (i.e. an always on device), provided impact events are transient or rare (it can be seen in Figure 5 that when impact events are very frequent the current consumption of the wireless unit 104 increases, since the processor 106 is in a full power mode more often).

In the example described with reference to Figure 5, power source lifetime is estimated based on the capability of two AA batteries. For example, when the impact activity is 0.1%, the average current for the low-power wireless unit (0.99 mA) is 12 times lower than the average current of an always-on device (11.7 mA), and the lifetime is extended from 341 hours to 4034 hours. Also, when the impact activity is low, which is the typical situation for impacts on aircraft, the average current consumption (0.99 mA @ 0.1% impact activity) for the low-power wireless unit 104 is close to the current consumption of the wireless unit when in the power saving or low-power mode (0.98 mA) - in other words, the additional power consumed in order to operate the wireless unit is only marginally more than the consumption in the low-power mode. As such, use of an event-triggered mechanism to cause transition of a processor module out of a power saving mode in response to a trigger signal from the comparator can provide a wireless unit which can have significantly lower power consumption than traditional, always on, devices, and consequently which can have significantly extended battery lifetime (when the number of impact events over time is relatively low, as is the case for most applications in which such a wireless, low power, system would be beneficial). However, as indicated in Figure 5, the power saving advantages of a trigger based mechanism are not realised when the percentage of impact events over time is high.

A method 600 of detecting impacts using the wireless unit of the first aspect is described with reference to Figure 6. At step S610, sensor output signal(s) are received at the wireless unit from a sensor network. The received sensor output signals are indicative of vibrations of a component to which the sensor network is attached, or with which the sensor network is associated. Before step S610, when wireless unit 104 is part of system 100 comprising sensor network 102, method 600 may further comprise sensing, at the sensor network, vibrations of the component.

At step S620, a characteristic of the received sensor output signal is compared to a predetermined threshold. For example, a voltage of the received signals is compared to a predetermined reference voltage. The reference voltage, or threshold, is optionally adjustable. In other words, the level of impacts on the component which are detected can be adjusted according to the specific application. More than one characteristic may also be compared, if appropriate. Before step S620, method 600 may optionally comprise, filtering, at the wireless unit, the sensor output signal. Alternatively, any filtering may be performed at the sensor network. When method 600 comprises filtering, step 620 comprises comparing characteristic(s) of the filtered signals, rather than of the received signals, to a predetermined threshold.

At step S630, if the characteristic satisfies the threshold criteria during the comparison (for example, if the characteristic exceeds, or is less than, or otherwise meets the comparison criteria to the predetermined threshold 118), a trigger signal is output to a processor to cause the processor to transition from a power saving mode to a fully awake mode. In other words, the trigger signal is selectively output based on the comparison. The processor of the wireless unit transitions out of the power saving mode in response to receiving the trigger signal. At step S640, the received sensor output signals are processed by the processor to generate a record of an event, or an event record, which characterises the impact event experienced by the composite. Optionally, method 600 comprises calculating, at the wireless unit, an estimate of the sensor output signal between a time the event occurred and a time of causing the processor to process the sensor output signal. Optionally, this calculating can comprise smoothing, averaging, applying a moving average, etc., as determined by the specific application and properties/characteristics of the sensor output signals 116 received by the processor.

Optionally, method 600 further comprises extracting one or more parameters from the sensor output signals to form the generated record. In other words, the generated record comprises at least one parameter extracted by the processor, which parameter represents one or more aspects of the sensor output signal. For example, the extracted parameter may represent the time of arrival of the signal at the processor, or the time of arrival of the first peak. Optionally, the one or more extracted parameters comprise at least one of: a time of arrival of the sensor output signal, an energy of the sensor output signal, a voltage of the sensor output signal, an amplitude of the sensor output signal, an energy of a first peak of the sensor output signal, a voltage of the first peak of the sensor output signal, and an amplitude of the first peak of the sensor output signal. At step S650, at least part of the generated record of the event is provided to a wireless transmission module of the wireless unit for transmission to an external device (a device external to the wireless unit). At step S660, the processor transitions back to the power saving mode. This transition back to the power saving mode can occur at any suitable time in method 600; for example, the transition back to the power saving mode may occur after providing the at least part of the generated record to the transmission module for transmission to the external device.

At step S670, the at least part of the generated record is wirelessly transmitted to the external device from the wireless unit. Optionally, the transmitting occurs in real time, i.e. as the sensor output signals are processed by the processor. Optionally, method 600 further comprises storing at least part of the generated record at a storage module of the wireless unit until step S670 can be performed, or permanently (i.e. regardless of whether or not step S670 has been performed).

Steps S610 to S670 occur at the wireless unit 104. At step S680, the at least part of the generated record is received at the external device 108 from the wireless unit 104, and at step S690, the at least part of the generated record is analysed to localise the impact event on the composite. These receiving and analysing steps occur at the external device. The localised impact event can be used for evaluating or determining maintenance schedules associated with the component. Other properties of the impact can optionally also be analysed. For example, an energy of the impact may be determined from the generated record, or an input force/direction of the impactor which caused the impact assessed.

The above steps of method 600 can be performed in any suitable order. Moreover, any other features described above with reference to the first aspect may be implemented in the method of the second aspect. For example, method 600 may further comprise rectifying one or more sensor output signals, for example, rectifying the received sensor output signals before they are processed by the processor of the wireless unit. It is noted herein that while the above describes various examples, these descriptions should not be viewed in a limiting sense. Rather, there are several variations and modifications which may be made without departing from the scope of the appended claims.