New Initial development of wireless acoustic emission sensor Motes … · 2018. 12. 21. · Initial...

Smart Structures and Systems, Vol. 6, No. 3 (2010) 197-209 197

Initial development of wireless acoustic emission sensor

Motes for civil infrastructure state monitoring

Christian U. Grosse1*, Steven D. Glaser

2

and Markus Krüger3

Department of Non-destructive Testing, cbm, Technische Universität München, Baumbachstr. 7,

D-81245 München, Germany

Department of Civil and Environmental Engineering, University of California, Berkeley, 455 Davis Hall, CA

94720-1710, USA

Department of Non-Destructive Testing and Monitoring Techniques, Materialprüfungsanstalt Universität

Stuttgart, Pfaffenwaldring 4, D-70550 Stuttgart, Germany

(Received July 15, 2008, Accepted July 1, 2009)

Abstract. The structural state of a bridge is currently examined by visual inspection or by wired sensor

techniques, which are relatively expensive, vulnerable to inclement conditions, and time consuming to undertake.

In contrast, wireless sensor networks are easy to deploy and flexible in application so that the network can adjust

to the individual structure. Different sensing techniques have been used with such networks, but the acoustic

emission technique has rarely been utilized. With the use of acoustic emission (AE) techniques it is possible to

detect internal structural damage, from cracks propagating during the routine use of a structure, e.g. breakage of

prestressing wires. To date, AE data analysis techniques are not appropriate for the requirements of a wireless

network due to the very exact time synchronization needed between multiple sensors, and power consumption

issues. To unleash the power of the acoustic emission technique on large, extended structures, recording and local

analysis techniques need better algorithms to handle and reduce the immense amount of data generated.

Preliminary results from utilizing a new concept called Acoustic Emission Array Processing to locally reduce

data to information are presented. Results show that the azimuthal location of a seismic source can be

successfully identified, using an array of six to eight poor-quality AE sensors arranged in a circular array

approximately 200 mm in diameter. AE beamforming only requires very fine time synchronization of the sensors

within a single array, relative timing between sensors of 1 µs can easily be performed by a single Mote servicing

the array. The method concentrates the essence of six to eight extended waveforms into a single value to be sent

through the wireless network, resulting in power savings by avoiding extended radio transmission.

Keywords: wireless; sensor network; acoustic emission; structural health monitoring.

1. Introduction

Continuous monitoring of structural behavior and health can provide data that allow better

understanding of its structural performance, and in turn allows prediction of its durability and

remaining life time. In Europe, many structures originate from the middle of the twentieth century, built

to replace structures destroyed during the Second World War. Concrete structures, the most common,

are typically designed for a 50- to 100-year life span, so that many current structures are rapidly

*Corresponding Author, Professor, E-mail: [email protected]

198 Christian U. Grosse, Steven D. Glaser and Markus Krüger

approaching the end of their design life. This problem is especially serious for railway bridges which

are confronted with increasing axle loads and higher train speeds that very often exceed the structural

design loads. In this context, a European Research Project was approved in the Sixth Framework Program

(Sustainable Bridges 2007). One objective of the project was to provide monitoring techniques that

could help the bridge owners to quantify the in situ structural behavior of their bridge stock.

To function properly, a monitoring and inspection procedure must be reliable, inexpensive and simple

to implement. The techniques used should be easy to adapt to different types of structures and structural

elements – a large variety exist and adaptation is time consuming. Given these facts, the development

and application of wireless sensor pods (often referred to as Motes), incorporating micro-electromechanical

systems (MEMS) - based microsensors, is a powerful solution to our problems (Glaser et al. 2005,

2006, 2008a). A thorough overview of wireless sensor platforms is presented by Lynch (2007). From

past work, wireless monitoring system equipped with accurate yet low cost sensors can reduce

structural monitoring costs dramatically.

One objective of monitoring civil engineering structures is to detect damage to structural parts which

can reduce the load bearing capacity and remaining useful lifetime. The detection and localization of

steel tendon failure or concrete cracking in bridge structures are examples of such a monitoring task.

Another example is the determination of steel cable forces by dynamic measurements (Feltrin et al.

2006, Meyer et al. 2006).

The rubric of structural health monitoring can be extended to include the construction process itself.

Previously the contractor simply implemented a given design ordered by the owner, but the current

trend is for clients to commission certain performance requirements to be met by the finished product -

performance-based design. The contracting process becomes the determination of the performance

criteria, and delivery becomes a long-term fulfillment of these criteria. This arrangement can only take

place if the performance states can be measured and quantified, and the measurement utilized in a

decision making process (Glaser and Tolman 2008b). The processes needed for the evaluation of the

structural state at delivery and during operation are increasingly dependent on sensor data and valid

models to turn the data into indicators of physical behavior, and decision making tools to determine

whether the performance requirements are being met.

2. Reasons for wireless monitoring

Most existing monitoring systems use traditional wired-sensor technologies, typically using a large

number of sensors (i.e., more than ten) which are connected through costly long cables and therefore

will be installed on only a few structures. A wireless monitoring system with MEMS-based sensors

(microsensors) could reduce these costs significantly (Fig. 1) (Glaser 2004, 2005).

The cost savings are in large part a function of the lack of wiring, installation, and maintenance, and

these benefits also increase the variety of field applications. The aerially dense monitoring now made

physically and financially possible provides very detailed information about structural state, in so

allowing better and more cost effective maintenance schedules (Grosse et al. 2005). Only after certain

changes in the structural behavior have been identified will physical inspection (either by means of

non-destructive testing or visual methods) be necessary, and proper repair can be made immediately

after the identification of the defect. This reduces the risk of further damage. In fact, such monitoring

systems, linked with proper system models, allows for predictive maintenance scheduling so that the

actual macroscopic failure never occurs.

Initial development of wireless acoustic emission sensor Motes for civil infrastructure state monitoring 199

A structural health monitoring system can require data to be continuously transmitted (e.g., using the

internet or SMS protocols) to the supervisor. Each sensor node, which is itself a complete, small,

measurement and communication system, has to be powered and the energy cost optimized. Further on,

the down-aggregation of data to a few meaningful values is required to prevent system overload and

loss of precious power to excessive radio transmission.

The entire monitoring system has to withstand a rough environment. For example, it has to be resistant

against oil, fuel, salt, alkali and other chemicals. The boards are developed for rough environment,

mounted in sealed enclosures following the IP64/65 standards of water protection. The sensors have to

be robust and durable so that their measured data is reproducible and reliable over the monitoring

lifetime. Furthermore, the system stability, which includes the wireless data transfer to and from the

sensor nodes, must be high.

3. Mote clustering and sensor networking

Wireless sensor networks consist of an array of many nodes (Motes), each having one or several

different sensors on board. After the recording and a preliminary analysis of the data by the Mote, the

data has to be transmitted using, for example, a networked radio transmission system from a Mote to a

base station or supervisor, for further data processing or proper generation of alarm messages (Fig. 1).

Data transfer reliability can be improved by storing, then sending a signal (with a check-bit) on

command, rather than operating in real time. There are several different network topologies commonly

used for wireless sensor arrays, including the star and the multi-hop topology (Culler et al. 2003). A

further branch up the tree would be the formation of functional clusters of nearby motes (Fig. 2). In

addition to the local signal processing taking place on a single Mote, data within clusters can be

aggregated at intermediate nodes, further processed, and forwarded in compound packets to save the

amount of radio communication, i.e., energy consumption. Adaptive cluster formation and management can

also help in deciding whether or not an event is related to a structural defect or change in structural

behavior (Cano et al. 2008, Liu et al. 2004).

Such a networked sensing system has several advantages, among them is cost efficiency; portability,

and a wide variety of sensors can be used help constrain the modelling to identify the status of the

Fig. 1 Scheme for wireless sensing of large structures using radio frequency transmission techniques and MEMS-

based sensors. Data is sent from the base station to the supervisor by using e.g., internet or SMS protocols


structure. The reliability of a structural health identification algorithm is greatly enhanced by combining

physical quantities obtained by a multitude of sensors at a multitude of locations on the structure.

Establishment of a correlation between recorded data and structural performance is difficult and

should be based on the interpretation expertise of the user, implying a natural application of Bayesian

statistics. Embedding some local processing capabilities in each Mote has the result of turning data into

information locally, which decreases the number of bits that need to be sent over radio; remember that it

costs at least ten times the energy to send a bit than compute a bit.

Finally, two other advantages of wireless sensor networks should be stressed. Scalability is an issue if

the stakeholder wants to extend the monitoring area or needs more data over space or time. Existing

wireless sensor network (WSN) techniques enables self-organization so that sensor nodes can be added

or removed at any time without external reorganization of the network in the future. Implemented pre-

processing algorithms might need updating to adjust to new requirements, or for more efficient data

reduction. Most of the developed sensor nodes are remotely reprogrammable, i.e., that the user can

change the algorithms implemented in each sensor.

4. Development of a wireless acoustic emission sensing network

One method to monitor the change of structural state is by so-called acoustic emission monitoring

(Grosse and Ohtsu 2008). Acoustic emissions are the seismic signal from a sudden change of strain

within or on a sample. Common examples are fracture growth and sliding. The signals have a bandwidth of,

say, 5 kHz to several MHz. McLaskey and Glaser (2009) report measuring displacements as small as

1 pm over this bandwidth. All AE sensors are piezoelectric, usually made from PZT-5A. The propagating

AE signal caries information about the source mechanism as well as the material through which it

travelled, so AE is a good indication of structural damage mechanisms. There have been several

problems which have so far precluded the use of AE monitoring with Motes. Among them are the issue

of absolute time synchronization between Motes, and the large amount of data that has to be transferred

between Motes. Our methodology overcomes many of the serious limitations inhibiting the use of AE

in wireless sensor networks.

We have developed a multi-sensor acoustic emission (AE) Motes suitable for monitoring civil

engineering structures. Each Mote is composed of one or more sensors, a data acquisition and processing

unit, a wireless transceiver, and a battery power supply (Fig. 3) (Krüger et al. 2005). As much as

possible, the device uses commercial-of-the-shelf hardware. The acquisition and processing unit is

equipped with a low-power microcontroller offering an integrated analogue to digital converter (ADC)

Fig. 2 Scheme of a sensor network using clustered sensor nodes


and sufficient data memory (RAM) to store the measurements and allow for calculations. The Mote

also incorporates signal conditioning circuitry interfacing the sensors to the ADC. In the following

sections, some of these components are described in greater detail; also see Krüger et al. (2006).

Each Mote has to be powered and energy use optimized – each bit transmitted is wasteful. Using

multi-hop techniques, the data from each Mote can propagate through the network by hops from Mote

to Mote, each some tens to hundreds of meters. If the data is a waveform vector rather than pseudo-

static scalars, the number of possible hops becomes limited because the volume of data, increasing at

each hop, overwhelms bandwidth. Therefore, the concentration of the large amount of data into focused

information is necessary.

4.1 Array processing

Array processing of AE data is a powerful method of concentrating many tens of thousands of data

points recorded from one AE event into a single value – the azimuthal direction of the source relative to

the known orientation of the sensor. If this is the information needed by the operator, the method is an

obvious boon. In fact, a bridge is extremely seismically noisy, so that any AE signal recorded will have

a low S/N ratio. Since a cow’s ear cannot be made into a silk purse, there is actually little information

present in a recorded AE signal beyond its directional relation to a sensor array.

Similar to phased-array signal processing techniques developed for other nondestructive evaluation

methods, this technique adapts beamforming tools developed for passive sonar and seismological

applications for use in AE source localization and signal discrimination analyses. It has been used

extensively in radar (Haykin 1985), sonar (Carter 1981), and exploratory seismology (Justice 1985,

Kelly 1967), and has been utilized as a tool for non-invasive testing techniques for spacecraft (Holland

et al. 2006), pipelines and pressure vessels (Luo and Rose 2007, Santoni et al. 2007), and medical

applications (Kim et al. 2006). It has also been used for active damage detection in civil engineering

materials (Sundararaman et al. 2005, Azar and Wooh 1999), but it has not been applied to the method

of acoustic emission (McLaskey and Glaser 2009).

For beamforming applications the user must assume that the wave field is relatively constant normal

to the direction of propagation of waves incident upon the array (Dudgeon 1977). AE sources are

usually considered point sources, so the “delayed replica” assumption is only valid if the distance

between the source and receivers is large compared to the distance between neighbouring sensors (easy

Fig. 3 Device developed by Smartmote and University of Stuttgart. Left: Mote including sensor and data processing

board, radio transmission unit, antenna and container, Right: Concept of the sensor and data processing board


to insure). The relatively delayed signals can be combined (or stacked) to form an array output with

improved S/N.

Instead of using P-wave picking algorithms, this method uses the energy rich Rayleigh wave (Kelly

1967) and a small array of four to eight sensors. Instead of using a distributed array, the beamforming

AE method relies on a small array of sensors spaced closely enough that, in the frequency range of

interest (less than 50 kHz), all sensors will detect AE waves which have propagated along similar paths,

have been affected by similar attenuation and scattering. In the beamforming AE method, the direction

of arrival of the AE waves can be determined simply from the relative time delays of individual

acoustic emission signals.

For the reported tests, arrays of six to eight inexpensive, low frequency (50 kHz) resonant-type AE

sensors were set into an approximately 250 mm diameter circle. This size easily allows the array to be

serviced by a single AE Mote, and only the backazimuth needs to be sent back through the network.

Relative timing between each array sensor can easily be kept to 1 µs within a single Mote. The assumption

that the structure is plate-like, so the depth of source is a higher order term, is valid for the bridge-deck

structures monitored. In general beamforming can accurately locate the AE source to within five to ten

degrees of the actual azimuth (McLaskey et al. 2008). As expected, better quality sensors and denser

arrays gave more accurate results, but not meaningfully better for most field applications.

4.2 Sensors

There are different alternatives to obtain data related to structural state. Passive sensors do not require

electric power since they obtain their energy directly from the change of physical quantities. Piezoelectric

materials are an example of such materials. But active sensors, while drawing current, have many

enhanced abilities as to sensitivity, linearity, range of sensing, and many MEMS-based active sensors

incorporate signal conditioning circuitry and/or A/D-converters which greatly simplifies the signal

processing chain. MEMS-based sensors are available for many but not all applications, so Motes must

be able to communicate with conventional sensors as well. We will call a Mote addressing both micro

and macro-sensors a hybrid sensor Mote. Although the relevant macro-sensors operate in the Mote at

low-power, most will be replaced by microsensors as soon as they are available.

Hybrid Motes must be designed to optimize the data acquisition and to best match the in-situ

requirements. A hybrid Mote is best designed with independent sub-boards, for example signal

conditioning of strain, and piezoelectric data (from acoustic emission sensors), and several have been

developed by the University of Stuttgart, with help from EMPA (Eidgenössische Materialprüfungs-

und Forschungsanstalt, Switzerland). The developed sub-board (Fig. 4), has two parallel strain

measurement circuits and a full front-end for resistive sensors with temperature compensation using

dummy strain gages, as well as calibration and zero compensation by software (Fig. 4, right).

Implementation and development of the electric components, layout, and manufacturing of prototypes is

always in progress.

A piezoelectric sensor (i.e., AE sensor) signal conditioning board (Fig. 5) was developed, consisting

of two channels per board, with the opportunity to at present implement two boards in one Mote. Each

channel can be filtered and amplified individually. Amplification can be chosen between 100 x and

1000 x, as can user-specific anti-aliasing filters. The A/D conversion takes place within the TI MSP430

microcontroller, yielding 12-bit amplitude resolution at sample rates of up to 100 kHz, depending on

the number of active channels. The number of samples that are recorded after the detection of an event

is configurable, as well as pretrigger length and the trigger threshold. For performance reasons, the


MSP430’s DMA capabilities are used for transferring the sampled data to internal memory. In active

mode, samples are stored to a circular buffer until the trigger interrupt is set off. The remaining samples

of interest are then recorded and afterwards the sampling stops for the time of data transmission.

4.3 Sensor fusion within and among Motes

Correlation of AE data with the other data obtained by each Mote (temperature, humidity, strain, etc.)

will lead to further understand local structural behaviour. For example, a cross-check of AE activity

with increasing strain or with a sudden or abnormal increase of the ambient or intra-structural temperature

can give further insight into possible damage mechanisms at work. Such sensor data correlations will

Fig. 4 Signal conditioning board for strain and developed graphical user interface for calibration

Fig. 5 Mote consisting of a processor board with 4-channel signal conditioning for acoustic emission analysis (left)

and industrialized sensor node prototype attached to a reinforced concrete structure for crack monitoring

(right)

Fig. 6 Analogue signal conditioning with interrupt generation (right: prototype circuit board)


also decrease the amount of data transmitted after implementing intelligent data processing and interpretation

algorithms.

In addition to the local signal processing taking place on a single mote, information within clusters

can be aggregated in intermediate nodes, further processed, and forwarded as needed in compound

packets to save energy. First storing a set of data in a given sensor Mote and then sending it consecutively

through the radio module at specific time intervals, or events on request, will also improve the reliability of

data transfer because the transfer can be specifically controlled and the transmission error corrected.

The need for cluster formation and management is motivated by power concerns as well as the

necessity of deciding whether or not an event is related to a significant structural defect, or change in

structural behaviour. These clusters can organize themselves around the damage source, perform local

analysis, and send a succinct message back through the network.

5. Field tests and applications

An AE system must be able to discriminate between noise and significant damage signals from structure

deterioration. For civil structural health monitoring, the working environment (e.g., railway bridges) will

always be very noisy. A noise analysis of the working environment must first be conducted using

conventional hardware and broadband sensors to characterize the frequency band of noise at a given

bridge. As a first test of our wireless AE system, the equipment was installed for measurements of strain

and AE during static loading of a large pre-stressed reinforced-concrete bridge deck model (Fig. 7, left) at

the Technical University of Braunschweig, Germany, and at a smaller steel reinforced concrete structure at

the University of Stuttgart (Fig. 7, right). Since both structures are subjected to little ambient noise, the

influence of larger traffic noise was studied at the smaller structure (Grosse et al. 2007a).

The maximum detection radius of source to sensor array was investigated using standard ASTM E

976-99 test sources (break of a pencil lead). At a maximum, signals could be measured with usable

signal-to-noise ratio at a radius of 10 m. In practice, noisy AE signals at a radius of 4.10 m (small

source) and 6.90 m for a strong source produced by a forklift truck on top of the structure or a car (see

Fig. 7, right) were usable. This relatively long sensor distance indicates that our beamforming AE

source location method can monitor a useful area of a bridge deck, for instance.

These techniques were tested on the larger prestressed structure as well. The model bridge deck was

loaded downward, with some small eccentricity to the right of center (Fig. 8). The recorded AE

waveforms from simulated damage were subdivided according to their signal-to-noise ratio into

“category-1” (good), “category-2” (moderate) and “category-N” (possibly noise). The classification criteria

Fig. 7 “Concerto Bridge” in Braunschweig (left) and ramp like structure of the University of Stuttgart (right)


to assign the different signals into its proper group were obtained by training the Strintzis K-means

algorithm (1999), one of the most widely used (Charalampidis 2005, Ruspini 1969). By comparing the

incoming energy levels of the signals from the different sets by time it is evident that most of the energy

in the Category-1 type signals will arrive earlier than in the other two categories. Incoming power can

be compared to an assumed constant power influx that would result in the same energy for a given

observation length of N samples. Mathematically this is stated as

where si(n) denotes the n

th

signal sample at sensor i and N denotes the number of samples observed. The

function u(n) is thus a measure of the incoming power in relation with a constant power influx. Because

of the large variations in energy of the recorded signals, all waveforms were normalized according to

energy in both property spaces – the sampling interval of all waveforms was 1 !s. Effectively this made

the term to the right of the minus sign in the equation given above redundant. Fig. 9 shows some examples of

the S/N categorized signals, s(n), and their location on the deck relative to the AE array (indicated by arrow)

as well as typical u(n) functions, normalized according to energy content (smoothed). More information

is given in Grosse et al. (2007b).

ui

n( ) si

2

l 1=

l n=

!l( )

n

N

---- si

2

l( )

l 1=

l N=

!–=

Fig. 8 Side view of “Concerto Bridge (upper) and result of AE beamforming localization (lower right). Most of the

AE signals have been expected having an incidence angle of 190o

since in this direction the maximum

tensile load was applied to the bridge


In the lower right of Fig. 8, the results of beam steering localization are presented. In the beamformer

which was used here the delays are computed for an assumed direction of arrival for all apparent velocities

of the incoming wave and the corresponding signals are delayed according to the computations

performed. In the case when the true direction of arrival (backazimuth) of the incoming wave matches

the assumed one, the signals add coherently and a maximum in energy is obtained. If the computed

delays are denoted by !!ic the output of the delay-and-sum beamformer can be stated mathematically in

continuous time as

at which again si(n) denotes the n

th

signal sample at sensor i and Ns denotes the number of sensors in the

array and yc(t) is the beam formed according to a reference point c. Fig. 8 shows that a better S/N results

in a more accurate localization of the AE events. The backazimuth of Category-1 beams vary in a range

of 30o

and Category-2 in a range of 45o

while Category-N signals are more or less uncorrelated. More

information about the results of these experiments was published by Grosse et al. (2007b).

6. Conclusions

The inspection of building structures is currently a visual process. Therefore, the condition of the structure

is examined only at the surface, and the interpretation and assessment is based on the level of experience

of the engineers. An approach to continuous structural health monitoring techniques based on wireless

acoustic emission sensor arrays was presented, which provide data from damage occurring either inside

or on the surface of a structure, allowing better estimation of structural performance and integrity.

Based on the experience of the constructor, owner, or inspector, the structurally important zones where

monitoring is needed can be greatly restrictive. In many cases it is necessary to just detect a deviation of

the “usual” behavior of the structure, i.e., an outlier in a time-series. The reliability of the estimate of damage

yc

t( ) si

t !Tic

–( )

i 1=

N

!=

Fig. 9 Typical examples of recorded bridge deck signals s(n) (high frequency) that are classified as Category-1,

Category-2 - and Category-3 signals. The dotted, smooth waveforms are the same signals normalized by

energy content, u(n)


is enhanced by combining information from different measurands, resulting in a drastic improvement

of reliability and lowering of the detection threshold of deterioration. Establishment of a correlation

between data and structural performance is difficult and should be based on the expertise of the user,

implying a natural application of Bayesian statistics. This can be done by pre-processing data at the

Mote or cluster of Motes, a tremendous advantage to transmitting all recorded data. Intelligent data

processing in the Mote or Mote clusters utilizes pattern recognition algorithms which can greatly reduce

power consumption since only meaningful data are transmitted to the sink.

A wireless sensor network system based on hybrid sensors was developed by a team of scientists

from different institutions (MPA, UC Berkeley, Smartmote, EMPA Switzerland). The network is Mote-

based and will be low cost. Since prototypes are already available, the system is now undergoing an

optimization process regarding power consumption, data acquisition and aggregation, signal analysis,

and data reduction (Krüger et al. 2006).

Acoustic emission techniques can play a significant role in the monitoring of civil engineering

structures since the method is able to detect seismic waves from damage inside and on the surface of the

structure long before a failure occurs. However, most of the existing AE data analysis techniques might

not be appropriate for the requirements of a wireless network. In order to overcome issues such as the

need of many sensors to surround the damage area for source location, and precise relative timing and

identification of the first P-wave arrival at each sensor, utilization of beamforming array processing is

shown to yield favorable azimuthal location of AE sources. The beamforming solution requires small

arrays of AE sensors (four to eight) located in a tight circle, for which all array sensors can be monitored by a

single Mote, and only one piece of information needs to be propagated back through the network. The

method can utilize any identifiable section of recorded waveform, so that poor quality sensors, and the

much stronger Rayleigh wave mode, can be used for identification. First tests showed promising results,

which will be published in detail in the near future.

Acknowledgements

A part of the presented work and developments was supported by the European Community in the frame

of the project “Sustainable Bridges” (Monitoring). Grosse is also indebted to the Deutsche Forschungs-

gemeinschaft (DFG) for the scholarship granted under GR-1664/2-1. A part of the experimental work and

especially the algorithms and data processing concerning AE array techniques was done in collaboration

with G.C. McLaskey from the University of California Berkeley, USA, and P. Chatzichrisafis from the

University of Stuttgart. Glaser and McLaskey are supported by NSF-GRF, and NSF grant CMS-0624985.

The authors are also grateful for the help of S. Bachmaier, G. Bahr and Anne Jüngert from the University of

Stuttgart as well as for the support by their colleagues from IPVS Stuttgart (O. Saukh, P. J. Marrón, K.

Rothermel) and EMPA Zürich (J. Meyer, R. Bischoff, G. Feltrin). The collaboration with the Institut für

Baustoffe, Massivbau und Brandschutz (IBMB) of the Universität Braunschweig (Brunswick, Germany) is

gratefully acknowledged.

References

Azar, L. and Wooh, S. (1999), “Experimental characterization of ultrasonic phased arrays for nondestructive

evaluation of concrete structures”, Mater. Eval., 57(2), 134-140.


Cano, C., Bellalta, B., Villalonga, P. and Perelló, J. (2008), “Multihop cluster-based architecture for sparse

wireless sensor networks”, Electronic Proceedings of the 14th European Wireless Conference 2008, Prague,

Czech Republic.

Carter, C. (1981), “Time delay estimation for passive sonar signal processing”, IEEE Trans. Acoust., Speech,

Signal Processing, ASSP-29(6), 463-470.

Charalampidis, D. (2005), “A modified k-means algorithm for circular invariant clustering”, IEEE Trans. Pattern

Anal. Machine Intell., 27(12), 1856-1865.

Culler, D., Woo, A. and Tong, T. (2003), “Taming the underlying challenges of reliable multihop routing in sensor

networks”, Proceedings of the 1st International Conference on Embedded Networked Sensor Systems, Los

Angeles, California, USA.

Dudgeon, D. (1977), “Fundamentals of digital array processing”, Proc. IEEE, 65(6), 898-904.

Feltrin, G., Meyer, J. and Bischoff, R. (2006), “A wireless sensor network for force monitoring of cable stays”,

Proceedings of the IABMAS'06 - 3rd international conference on bridge maintenance, Safety and Management,

Porto (Portugal), July.

Glaser, S.D. (2005), “Advanced sensors for monitoring our environment”, Proceedings of the 1st International

Symposium on Advanced Technology of Vibration and Sound, Miyajima, Japan.

Glaser, S.D. (2004), “Some real-world applications of wireless sensor nodes”, Proceedings of the SPIE Symposium

on Smart Structures and Materials/NDE, San Diego, California.

Glaser, S.D., Shoureshi, R. and Pescovitz, D. (2005), “Future sensing systems”, Smart. Struct. Syst., 1(1), 103-120.

Glaser, S.D., Min Chen, and Oberheim, T.E. (2006), “Terra-Scope - a MEMS-based vertical seismic array”,

Smart Struct. Syst., 2(2), 115-126.

Glaser, S.D., Ni, S.H. and Ko, C.C. (2008a), “System identification of soil behavior from vertical seismic arrays”,

Smart. Struct. Syst., 4(6), 727-740.

Glaser, S.D. and Tolman, A. (2008b), “Sense of sensing”, J. Infrastruct. Syst., 14(1), 4-14.

Grosse, C.U., Krüger, M. and Chatzichrisafis, P. (2007b), “Acoustic emission techniques using wireless sensor

networks”, Proceedings of the sustainable bridges – Assessment for future traffic demands and longer lives

(Eds. J. Bien et al.), Publ. Dolnoslaskie Wydawnictwo Edukacyjne Wroc aw, Poland.

Grosse, C.U., Krüger, M., Glaser, S.D. and McLaskey, G. (2007a), “Structural health monitoring using acoustic

emission array techniques”, Proceedings of the International Workshop on Structural Health Monitoring

(IWSHM) (Ed. Fu-Kuo Chang), Stanford University, Stanford, CA, Lancaster PA: DEStech Publications Inc.

Grosse, C.U. and Ohtsu, M. (2008), Acoustic Emission Testing: Basics for Research - Applications in Civil

Engineering, Springer publ., Heidelberg.

Grosse, C.U., Glaser, S.D. and Krüger, M. (2006), “Condition monitoring of concrete structures using wireless

sensor networks and MEMS”, Proceedings of the SPIE Vol. 6174, Smart Structures and Materials 2006:

Sensors and Smart Structures Technologies for Civil, Mechanical, and Aerospace Systems (Eds. Masayoshi

Tomizuka, Chung-Bang Yun, Victor Giurgiutiu).

Grosse, C.U., Kurz, J., Beutel, R., Reinhardt, H.W., Krüger, M., Marrón, P.J., Saukh, O., Rothermel, K., Meyer,

J. and Feltrin, G. (2005), “Combined inspection and monitoring techniques for SHM of bridges”, Proceedings

of the International Workshop on Structural Health Monitoring (IWSHM) (Ed. Fu-Kuo Chang), Stanford

University, Stanford, CA, Lancaster PA: DEStech Publications Inc.

Haykin, S. (1985), Radar processing for angle of arrival estimation, Array Signal Processing (Ed. S. Haykin),

Prentice Hall, New Jersey.

Holland, S., Chimenti, D., Roberts, R. and Strei, M. (2006), “Locating air leads in manned spacecraft using

structure-borne noise”, J. Acoust. Soc. Am., 121(6), 3484-3492.

Justice, J.H. (1985), Array processing in exploratory seismology, Array Signal Processing (Ed. S. Haykin),

Prentice Hall, New Jersey.

Kelly, E. (1967), Response of seismic arrays to wideband signals, Lincoln Laboratory, Technical Note.

Kim, K., Liu, J. and Insana, M. (2006), “Efficient array beam forming by spatial filtering for ultrasound B-mode

imaging”, J. Acoust. Soc. Am., 120(2), 852-861.

Krüger, M., Grosse, C.U. and Saukh, O. (2006), “Bridge monitoring using multihop wireless sensor networks”,

Proceedings of the Conference on Operation, Maintenance and Rehabilitation of Large Infrastructure Projects,

Bridges and Tunnels (Eds. Vincentsen and Larssen), IABSE Report No. 3, Copenhagen (on CD-ROM).

l


Krüger, M., Grosse, C.U. and Marrón, P.J. (2005), “Wireless structural health monitoring using MEMS”,

Proceedings of the International Symposium Damage Assessment of Structures (Eds. W. M. Ostachowicz et

al.), Gdansk, Poland, Zürich: Trans Tech.

Liu, J., Chen, Y. and Liestman, A. (2004), Clustering algorithms for ad hoc wireless networks (Eds. Y. Xiao and

Y. Pan), Ad Hoc and Sensor Networks, Nova Science Publisher.

Luo, W. and Rose, J. (2007), “Phased array focusing and guided waves in a viscoelastic coated hollow cylinder”,

J. Acoust. Soc. America, 121(4), 1945-1955.

Lynch, J.P. (2007), “An overview of wireless structural health monitoring for civil structures”, Phil. Trans. R.

Soc. Lon. A., Mathematical and Physical Sciences, Mathematical and Physical Sciences, The Royal Society,

London, 365(1851), 345-372.

McLaskey, G.C., Glaser, S.D. and Grosse, C.U. (2008), “Acoustic emission beamforming for enhanced damage

detection”, Proceedings of the SPIE Vol. 6932, Smart Structures and Materials 2008: Sensors and Smart

Structures Technologies for Civil, Mechanical, and Aerospace Systems (Ed. Masayoshi Tomizuka).

McLaskey, G.C. and Glaser, S.D. (2009), “High-fidelity conical piezoelectric transducers and finite element

models utilized to quantify elastic waves generated from ball collisions”, Proceedings of the SPIE Smart Structures

and Materials 2009: Sensors and Smart Structures Technologies for Civil, Mechanical, and Aerospace Systems

(Ed. Masayoshi Tomizuka).

Meyer, J., Bischoff, R., Feltrin, G. and Saukh, O. (2006), “A structural health monitoring system based on a

wireless sensor network”, Proceedings of the 3rd International Workshop on Structural Health Monitoring,

Granada, Spain.

Ruspini, E. (1969), “A new approach to clustering”, Information Control., 15(1), 22-32.

Santoni, G., Yu, L., Xu, B. and Giurgiutiu, V. (2007), “Lamb wave-mode tuning of piezoelectric wafer active

sensors for structural health monitoring”, J. Vib. Acoust., Transact. ASME, 129(6), 752-762.

Strintzis, M. (1999), Pattern Recognition (in Greek), Kyriakidis Brothers’ Publishing.

Sundararaman, S., Adams, D. and Rigas, E. (2005), “Structural damage identification in homogeneous and

heterogeneous structures using beamforming”, Struct. Health Monit., 4(2), 171-190.

Sustainable Bridges (2007), Sustainable Bridges – Assessment for future traffic demands and longer lives.

Integrated Project in the Sixth Framework Programme on Research, Technological Development and Demonstration

of the European Union, FP6-PLT-001653, http://www.sustainablebridges.net.

New Initial development of wireless acoustic emission sensor Motes … · 2018. 12. 21. · Initial...

Documents

Transcript of New Initial development of wireless acoustic emission sensor Motes … · 2018. 12. 21. · Initial...