Multi-source Energy Harvesting Powered Acoustic Emission Sensing

System for Rotating Machinery Condition Monitoring Applications

Wensi Wang

, Anderson Machado Ortiz

, Ningning Wang

, Michael Hayes

, Brendan O’Flynn

and Cian O’Mathuna

Tyndall National Institute, Dyke Parade, Cork, Ireland

Department of Energy, Fundaci

o CTM Centre Tecnol

ogic, Manresa, Barcelona, Spain

Keywords:

Acoustic Emission Sensor, Wireless Sensor System, Preventive Maintenance, Energy Harvesting.

Abstract:

This study concerns with the acoustic emissions (AE) monitoring for the applications of rotating machine fault

detections. The ﬁrst prototype wireless AE sensor system with vibrational, thermal and light energy harvest-

ing power supply is presented in this paper. This prototype regularly records the AE signal at 150-250KHz

frequency bandwidth and compares with known gear/bearing fault patterns. The data are compressed and

transmitted via Zigbee based wireless transceiver to nearby central control unit. The multiple-source energy

harvester proposed in this work generates 1.56mW from 0.048g vibration energy and 3.37mW thermoelectric

energy when deployed on the 62

C metal surface of a gas compressor. This battery-less wireless AE system

achieves power autonomy from environmental energy, realizing a self-powered easy-to-deploy wireless data

transmitting based health monitoring solution for wide range of rotating machinery.

1 INTRODUCTION

Rotating machinery such as electric motor, turbine

and air compressor are widely equipped in great many

industries, such as general manufacturing, power gen-

eration, refrigeration and many others. In rotating

machinery, gear and bearing are critically important

components. They are required to operate with high

reliability for extended period of time in harsh envi-

ronmental conditions. Unexpected Faults (UFs) of

the gear and bearing may lead to damages of entire

machine. Consequently, these UFs will cause consid-

erable machine repair/replacement cost, and also the

associated labour and downtime costs. Therefore, pe-

riodic non-destructive testing (NDT) and on-line con-

dition monitoring (OCM) are often used to conduct

preventive maintenance (PM) in order to effectively

diagnose and prevent further development of faults

(Bastianini et al., 2013).

Since early 2000s, the method of Acoustic Emis-

sion Monitoring (AEM) has been proposed for PM

applications (De Silva, 2010) and (Bohse, 2013).

Acoustic Emissions (AE) in rotating machinery are

transient elastic waves produced by the interactions

of two media of gears/bearings in relative motions.

AEM methods “listen” to and process the elastic wave

signals and used the interpreted information to di-

agnose the potential faults in the early stage of sur-

face/subsurface crack formation.

The evolution of wireless sensor networks (WSN)

technologies in the last decade provides an unique op-

portunity for the further development of NDT systems

(Grosse and Kr

uger, 2006). The micro-controller

(MCU) and low-power wireless transceiver based

WSN module (mote) is substantially less expensive

and less power-hungry than the conventional pro-

grammable logic controller (PLC) and Modbus se-

rial communication based monitoring system. In ad-

dition, the battery powered WSN can be deployed

with minimal installation cost. WSN based AEM

system has been proposed for infrastructure (bridges

and buildings) structural health monitoring applica-

tions (Ledeczi et al., 2009) and (Ayg

un and Gungor,

2011).

However, a bottleneck of WSN development is the

limited battery energy. The 0.5 milliwatts ultra-low

power WSN system can only achieve 6-month bat-

tery lifetime when powered from 1000mAh battery in

optimal condition. The task to regularly replace bat-

tery can become expensive when the number of WSN

motes is large and even impossible when the mote is

placed in difﬁcult-to-access locations.

Energy harvesting technologies provide a feasible

solution for WSN mote power supply. This method

492

Wang W., Machado Ortiz A., Wang N., Hayes M., O’Flynn B. and O’Mathuna C..

Multi-source Energy Harvesting Powered Acoustic Emission Sensing System for Rotating Machinery Condition Monitoring Applications.

DOI: 10.5220/0004593804920499

In Proceedings of the 10th International Conference on Informatics in Control, Automation and Robotics (ICINCO-2013), pages 492-499

ISBN: 978-989-8565-70-9

 2013 SCITEPRESS (Science and Technology Publications, Lda.)

“harvests” various forms of environmental energy

such as vibration (Zhu et al., 2012a), thermoelectric

(Im et al., 2012) and solar/light (Wang et al., 2012)

into electricity to power WSN mote. The power man-

agement circuity of energy harvester stores the energy

in electric double-layer capacitor (supercapacitor) or

thin ﬁlm solid state battery and utilizes the energy

when needed. The utilization of the small but “inﬁ-

nite” ambient energy provides a potentially indeﬁnite

battery lifetime for WSN systems. Recently the con-

cept of multiple sources energy harvesting has been

proposed to harvest energy from more than one type

of energy (Weddell et al., 2013).

Figure 1: Energy Harvesting Powered Wireless AEM Mod-

ule for Gas Compressor Monitoring.

By integrating and optimizing the three key tech-

nologies: 1) AE sensing, 2) Wireless sensor system,

and 3) Energy harvesting, a new type of machinery

preventive maintenance system is proposed. This pa-

per introduces a vibration/thermoelectric/light pow-

ered wireless sensor module based acoustic emission

monitoring system for rotating machinery fault detec-

tion application. Fig.1 illustrates the application of

this proposed system in gas compressor monitoring.

The remainder of this paper presents the working

principal, prototype implementation and the prelimi-

nary results of the proposed system. Section 2 sum-

marizes related work in the area of machinery moni-

toring and energy harvesting powered WSN systems.

Section 3 introduces the system architecture and sub-

systems. The ﬁrst part of Section 4 presents the AE

sensor and related signal processing. The rest part of

Section 4 shows the multiple-source energy harvest-

ing system and power management circuits. Section

5 demonstrates the prototype implementation and the

preliminary test results as of April 2013. Final section

concludes the main ﬁndings and planned future work.

2 RELATED WORK

The concept of rotating machinery monitoring has

been addressed in many publications and and im-

plemented in many commercial products. The most

widely used method of NDT is vibration monitor-

ing (VM) in the past 50 years (McFadden and Smith,

1984), (De Silva, 2010). More recently, the method

of Acoustic Emission Monitoring (AEM) has been

employed for PM applications. Compared with con-

ventional vibration monitoring methods, AEM shows

several advantages:

a) A substantial drawback of VM is the vibration en-

ergy from other parts of the machine (e.g. shaft

and cooling fan) is often several orders of mag-

nitude higher than the vibration energy of the de-

fect gear and bearing, i.e. low signal to noise ratio

(SNR). Acoustic signal has been proved to have

signiﬁcantly improved SNR during fault detection

(Bohse, 2013).

b) Since AE sensing is a non-directional/non-contact

technique, AEM system has greater insensitivity

in positioning sensors, which reduces the installa-

tion costs. In addition, fewer sensors are needed

in AEM systems than VM systems (Loutas et al.,

2011).

c) With VM system, when a signiﬁcant change in vi-

bration can be observed, the remaining lifetime of

the gear/bearing is very short. However, for AEM

system the acoustic emission generated during the

formation of cracks can be detected in the rela-

tively early stage (Mba and Rao, 2006). The ad-

vantage of early detection increases the chance of

preventive maintenance when AEM is employed.

Due to these advantages, AEM technique has been in-

creasingly adopted in recent years for machine and

structure health monitoring applications.

In (L

edeczi et al., 2011), Ledeczi et.al. demon-

strates a bridge structural monitoring system with AE

sensors. The DSP unit in this implementation is

based on a FPGA instead of low cost MCU due to

the data processing speed concern in MCU. However,

recent development of MCU towards higher speed

shows that the data process capability can be met by

advanced MCUs such as ARM Cortex-M4 (Rutzig,

2013). Weddell et.al. presented a tunable frequency

electromagnetic transducer powered WSN system for

vehicle ferry engine monitoring application (Weddell

et al., 2012). Since the diesel engine features different

vibration frequency from electric motors, the energy

harvesting device is also different.

Lubieniecki and Uhl presented an work in the area

of harvesting thermal energy from high speed rotating

Multi-sourceEnergyHarvestingPoweredAcousticEmissionSensingSystemforRotatingMachineryCondition

MonitoringApplications

493

bearings(Lubieniecki and Uhl, 2012). It investigated

the correlation between the rotating speed and the har-

vested thermoelectric energy based on their conﬁgu-

ration. It has concluded that with a rotating speed of

6000 revolutions per minute, more than 35mW aver-

age power can be harvested from the thermoelectric

generator (TEG).

3 SENSOR SYSTEM

ARCHITECTURE

The prototype system block diagram is illustrated in

Fig. 2. The system consists of four main building

blocks: 1) Sensor Layer; 2) DSP & RF Module; 3)

Energy Harvesting Module; 4) Expert System.

Figure 2: Energy Harvesting Powered Wireless AEM Sys-

tem Architecture.

The sensor layer for AE signal sensing includes

AE sensors, ampliﬁer, ﬁlter and ADC. It also has a

C Interface for optional temperature sensors.

The DSP and RF module consists of ARM Cortex-

4M MCU as the DSP unit, NXP JN5148 Zigbee

wireless module for regular RF communication, ﬂash

memory and USB interface. The data processing is

mainly based on fractional Fourier transform (FRFT)

signal process. The results are compared with fault

patterns to “ﬂag” the possible fault. When the fault

pattern is repeatedly detected, an alarm signal is sent

to expert system for further investigation. The pro-

cessed data is compressed and sent to expert system

via the Zigbee module. Flash memory is used to tem-

porarily store the data in the case of unsuccessful RF

data transmission. USB interface is only used during

system maintenance and possible upgrade.

The energy harvesting (EH) module includes 1)

rectiﬁer and DC/DC converter for electromagnetic

(vibration) energy harvester, 2) ultra-low voltage

(UL) DC/DC converter for thermoelectric generator

(TEG), 3) maximum power point tracking (MPPT) for

indoor photovoltaic cells (PV), 4) supercapacitor and

solid state battery (thin ﬁlm battery) as energy storage

unit (ESU) and 5) charge/discharge control circuity to

conditioning the input/output power from ESU. The

ESU state of charge (SoC) is sent to MCU to monitor

the condition of energy harvester.

The proposed prototype consists of four sub-

system layers as shown in Fig.3 : Energy har-

vester power management layer, Sensor layer (ampli-

ﬁer/ﬁlter/ADC), MCU layer and Zigbee communica-

tion layer. Three AE sensors can be connected to this

system.

Figure 3: Energy Harvesting Powered Wireless AEM Sys-

tem Prototype 3D Illustration (the protection case is not

shown in this illustration).

Vibration energy harvester (VEH), thermoelec-

tric energy generator (TEG) and photovoltaic (PV)

cells can be connected to the energy harvester layer.

The main energy storage unit in this implementa-

tion consists of two 5F supercapacitors. The pro-

totype (with IP45 protection case) is measured at

150mm×120mm×40mm. The ingress protection rat-

ing of the case is IP-54. All subsystems have been

prototyped and manufactured as of April 2013.

4 ENERGY HARVESTING

POWERED WIRELESS AEM

SYSTEM DESIGN

The proposed prototype is the ﬁrst version of the tech-

nology demonstrator. Before the integrating the sub-

systems into the ﬁnal prototype, each subsystem is de-

ICINCO2013-10thInternationalConferenceonInformaticsinControl,AutomationandRobotics

494

sign and their performance is investigated. This sec-

tion introduces the design of each subsystem.

4.1 AE Sensor System and Signal

Processing

The condition monitoring system is essentially based

on the feature extraction of AE signals. substantial ef-

forts was concentrated on the signal processing of AE

waveforms. Since the “fault pattern” of gear/bearing

is the main diagnostic parameter, long term gear test-

ing was conducted to identify the fault patterns in var-

ious frequency bandwidths. Fig.4 shows the test set-

up of the gear surface/subsurface crack formation ex-

periment.

Figure 4: Gear Surface Crack/Wear Acoustic Test Set-up.

The area highlighted in Fig.4 is the root circle of

a gear where the crack most likely to form. The AE

fault pattern tests were conducted with several differ-

ent types and stages of crack formations. A typical

formed crack (late-mid stage) is illustrated and high-

lighted in Fig.5. The gear acoustic emissions test re-

sults are presented when the crack is excited by the

contacts with other gear in Fig.5 (1-5).

Two types of data analysis techniques have been

considered in this work: fractional Fourier transform

(FRFT) and Daubechies wavelet (Grosse et al., 2002).

Whilst the Fourier transform is only localized in fre-

quency domain, wavelets are localized in both time

and frequency domain. Daubechies wavelet with a

10-level signal decomposition may require more data

processing than FRFT (Ching et al., 2004). Fig.6

shows the detected fault pattern at 200KHz when the

late-mid stage of the crack is scanned with a 150KHz-

300KHz ﬁlter.

Low power consumption of data processing and

transmission is a main challenge in the design of AE

WSN system. The MCU based device consumes

80-180mW power during “active” mode and 50µW

“sleep” mode power. The power consumption of AE

WSN mote is summarized in Table.1.

Since the harvested power from ambient environ-

ment is at 1mW level, the AE WSN system is pro-

Figure 5: Gear Crack Formation and Acoustic Emissions

Signal When Excited by the Contacts with Other Gear (1-

5).

Figure 6: Gear Fault (crack) Pattern in Frequency Domain

Analysis 150-300KHz FRFT Results.

grammed to perform duty cycling operation (periodic

active-sleep-active cycles) in order to minimize aver-

age power consumption. On average, the active mode

time is approximately 1.37 seconds followed by the

sleep mode time of 1 to 10 minutes in the tests. The

average power consumption ranges from 0.22mW to

1.76mW subject to the operational duty cycles.

Multi-sourceEnergyHarvestingPoweredAcousticEmissionSensingSystemforRotatingMachineryCondition

MonitoringApplications

495

Table 1: Power Consumption of AE WSN Mote (T

: sleep

mode time).

AE Power Power Time Energy

Consumption (mW) (Sec) (mJ)

Sleep Mode 0.047 60.00 2.820

Data ACQ 188.1 0.020 3.760

(3 AE Sensors)

Data FRFT & 76.26 0.920 70.16

Compression

Diagnosis 86.16 0.210 18.09

Algorithm

RF Transmission 61.38 0.220 13.50

Total (1min T

) 1.760 61.37 108.3

Total (3 mins T

) 0.620 181.3 113.9

Total (10 mins T

) 0.220 601.3 133.7

4.2 Multiple Source Energy Harvesting

Energy harvesting is proposed in this work to collect

environmental energy and convert the harvested en-

ergy into usable form (Power/Voltage/Current etc.).

On-site experiments had been carried out to investi-

gate the available ambient energy sources in an indus-

trial cold store facility. The mechanical vibration en-

ergy and surface temperatures on various positions of

the main air compressor units (shown in Fig.1) have

been measured to study the “harvest-able” energy. In

this characterization, 0.025g to 0.05g vibration has

been detected in addition to the 60 - 70

C surface

temperature on the rotary screw air compressor (near

the air outlet).

Based on the characterization, thermoelectric en-

ergy harvesting is proposed in this work. Thermo-

electric generator is a device that utilizes Seebeck ef-

fect which directly converts temperature difference

into electricity (Ramadass and Chandrakasan, 2011).

Thermoelectric materials high in positive/negative

Seebeck coefﬁcient such as Bi

are prepared into

P and N types of thermo-elements. One P type and

one N type thermo-elements are then connected in se-

ries via (copper) contacts as shown in Fig.7. In this

conﬁguration the P/N thermo-elements are connected

in parallel from heat transfer perspective and a pair of

thermo-elements forms a thermo-couple.

One thermo-couple can only generate small volt-

age difference (1-2mV) when 50-100

C tempera-

ture is applied on the “hot” side of TEG. An array

of thermo-couples are used to form a TEG module

which normally consists of several hundreds thermo-

couples.

Fig.8 shows a Bi

based TEG module with

P/N type thermo-element measured at approximately

1mm

. In this module, 255 thermo-couples are used

Number of Thermo

-couples Pairs N

Height of Ceramic

Substrate L

…

Substrate Cold

Side Temp T

Substrate Hot

Side Temp T

– T

= ∆T

P N

Cross Section

Area A

Length of Thermo-element L

Contact Cross-section

Area A

Length of Contact L

Thermo-couple

Cold Side Temp T

’

Thermo-couple

Hot Side Temp T

’

Temperature Difference on Thermo-couple

T’

– T’

= ∆T’

TEG Module

Thermo-couple

(Copper/gold) Contacts

Figure 7: Thermoelectric Generator (TEG): Thermo-

Couple and TEG Module.

Figure 8: TEG Module.

to form this TEG, connected via copper contacts and

supported with ceramic substrates. This type of TEG

is tested with hot side temperature ranging from 50

to 80

C. Passive heat sink (similar to typical CPU

heat sink) is mounted on the cold side of TEG with

thermal compound applied on the interface between

TEG and heat sink. The measured results of thermo-

electric generator are summarized in Table.2.

Table 2: TEG Characterizations Results at Matched Load.

Heat Source 50 60 70 80

Temp (

Module Temp 2.5 4.0 5.5 7.5

Diff. ∆T (

Measured 0.212 0.336 0.464 0.632

Voltage (V)

Measured Max. 1.384 3.544 6.704 12.46

Power (mW)

The generated voltage on the load is between

0.212V to 0.632V in this test. This voltage is lower

ICINCO2013-10thInternationalConferenceonInformaticsinControl,AutomationandRobotics

496

than the minimal start-up voltage of most boost con-

verters. This work adopts a power management de-

sign based on Texas Instruments BQ25504 energy

harvesting chipset. BQ25504 features a start-up volt-

age of 40mV. BQ25504 requires a start-up current of

several mA in order to obtain a 1.8V power supply

voltage on the storage capacitor. A maximum power

point tracking (MPPT) function adjusts the duty cy-

cle of boost converter in order to match the input

impedance of the boost converter to the TEG internal

resistance, thus, a matched impedance. In this work,

an additional load switch and an output voltage regu-

lator (buck/boost converter) are used to supply a reg-

ulated 3.3V voltage output. The load switch and the

enable pin (SHDN) of the buck/boost converter are

controlled by “VBAT OK” pin (Digital battery good

indicator) of BQ25504. In this way, the buck/boost

converter only starts up when BQ25504 is fully oper-

ational. Therefore, the cold start issue of buck/boost

converter can be avoided.

Figure 9: Schematics of Energy Harvester Power Manage-

ment Circuit.

In addition to the thermoelectric energy harvest-

ing, vibrational energy is widely available in the tar-

geted deployment scenarios. In electrical motor sys-

tems, the vibration frequency is highly dependent on

the mains frequency (power line frequency). In this

measurement, as shown in Fig.10, the resonant fre-

quency peaks around 50Hz with 48mg acceleration.

The electromagnetic vibration energy harvester

adopted in this design is Perpetuum FSH module (Zhu

et al., 2012b). The AC power generated from the

energy harvester is rectiﬁed by the Perpetuum FSH

module internal full bridge rectiﬁer. The DC/DC

power conditioning of the vibration energy harvester

is also BQ25504. The main difference is the MPPT

circuit is by-passed in this design.

A low voltage indoor photovoltaic power manage-

ment circuit is also built in this design with BQ25504.

Figure 10: Vibration Energy Harvester Power Management

and Measured Vibration of Air Compressor.

Different from impedance match in thermal electric

energy harvesting, the maximum power point voltage

of PV cell is around 76% of its open circuit voltage

(O’Donnell and Wang, 2009). The MPPT control sig-

nal of BQ25504 is adjusted accordingly for the PV

energy harvesting.

The multi-source energy harvester prototype is

implemented and tested on the air compressor unit to

verify the feasibility and functionality of the proposed

design. The energy harvester prototype and its test re-

sults are presented in next section.

5 PROTOTYPE DEPLOYMENT

AND PRELIMINARY RESULTS

The energy harvester for powering AE system pro-

totype was tested on an industrial air compressor in

a large scale cold store facility. Accelerometer and

thermo-couples are used to measure the temperature

and vibration at various part of the air compressor.

The data is recorded using a portable Picolog-1000

data acquisition system with labview interface. The

purpose of this deployment study is to determine the

suitable deployment position of energy harvester for

AE systems. The optimal position where energy har-

vesters can be deployed to is on the air-oil separator

of the air compressor.

The temperature measured on the outlet is mea-

sured at 69

C and 62

C on the surface of the air-oil

separator. The vibration energy is measured at be-

tween 0.025g and 0.048g at 49.3Hz to 49.7Hz fre-

quency during the experiment (when compressor is

operating). The deployment characterization of en-

ergy harvester is illustrated in Fig.11.

Multi-sourceEnergyHarvestingPoweredAcousticEmissionSensingSystemforRotatingMachineryCondition

MonitoringApplications

497

Figure 11: Energy Harvester Powered AE System Deploy-

ment Characterizations.

The energy harvester prototype is then deployed

on the top of air-oil separator. The result veriﬁca-

tion is based on the storage capacitor charging char-

acterization. The thermoelectric and vibrational en-

ergy harvesters are connected to the power manage-

ment circuit in order to charge the 0.47F supercapac-

itor. When charging the capacitive load, the average

charging power P

avg

can be calculated as,

avg

·V

target

2 · T

chrg

(1)

where C

is the super-capacitor capacitance, T

chrg

the total charging time.

The supercapacitor charging experiments were

conducted on TEG and vibration energy harvester

(VEH). The results are shown in Fig.12.

Figure 12: VEH and TEG Energy Harvesters Supercapaci-

tor Charging Experiments.

The room temperature in the experiment is 15

The hot side temperature of air-oil separator is 62

The supercapacitor is charged from 1.22V to 1.48V

within 325 seconds. The average harvested power of

TEG is calculated at 3.37mW.

When the VEH is excited with 48mg acceleration

at 49.5Hz, the harvested power charges the superca-

pacitor from 1.34V to 1.52V in 325 seconds. The

harvested vibrational power is 1.56mW on average

during the charging process. The combined harvested

power is calculated at 4.93mW when the air compres-

sor is operational.

By revisiting the power consumption proﬁle of

AE WSN module in Table.1, the harvested power

4.93mW is sufﬁcient to power AE WSN module

to operate with 1 minute measurement intervals

(1.76mW). The minimal measurement interval of AE

WSN module is calculated at 20 seconds, i.e. the

multi-source energy harvester enables the AE WSN

module to perform fault detection every 20 seconds.

6 CONCLUSIONS

Acoustic emissions monitoring system has demon-

strated several advantages over the conventional vi-

bration monitoring system for the application of

gear/bearing fault detections. Micro-controller based

wireless sensor networks (WSN) technologies signif-

icantly reduce the material and installation cost of in-

dustrial monitoring systems. Therefore, an approach

to conduct AE monitoring with WSN modules is pro-

posed in this work.

A main bottleneck for this type of system is

the mote power consumption can deplete the battery

within several months of deployment. An energy har-

vesting subsystem, which can harvest thermal, vibra-

tional and light energy, is then presented in this paper

to power the AE WSN mote with ambient energy.

The feasibility of powering AE WSN mote en-

tirely from energy harvesting is investigated in this

work. When deployed on an air compressor, the pro-

posed power management circuit shows that it can

harvest 3.37mW from wasted heat and 1.56mW from

machine vibration, then store the energy in superca-

pacitor type energy storage unit. The hybrid energy

harvesting subsystem generates 4.93mW when the air

compressor is operational. Based on the AE system

power consumption characterizations, the harvested

power is sufﬁcient to perform AE fault detection ev-

ery 20 seconds and achieves power autonomy in the

air compressor experiments.

All subsystems of the AE WSN system have been

built. The current system is under tests to verify

the reliability in real-world condition. Being a ﬁrst-

ICINCO2013-10thInternationalConferenceonInformaticsinControl,AutomationandRobotics

498

generation prototype, the prototype device is under-

going an optimization process from power consump-

tion/management, data processing and diagnose algo-

rithm perspectives.

ACKNOWLEDGEMENTS

This work has been funded by European seventh

framework programme (FP7) for small medium en-

terprise under research project Mosycousis (Project

reference number: 285848).

REFERENCES

Ayg

un, B. and Gungor, V. C. (2011). Wireless sensor

networks for structure health monitoring: recent ad-

vances and future research directions. Sensor Review,

31(3):261–276.

Bastianini, F., Sedigh, S., Pascale, G., and Perri, G. (2013).

Cost-effective dynamic structural health monitoring

with a compact and autonomous wireless sensor sys-

tem. In Nondestructive Testing of Materials and Struc-

tures, pages 1065–1070. Springer.

Bohse, J. (2013). Acoustic emission. In Handbook of Tech-

nical Diagnostics, pages 137–160. Springer.

Ching, J., To, A., and Glaser, S. (2004). Acoustic emission

source deconvolution: Bayes vs. minimax, fourier vs.

wavelets, and linear vs. nonlinear. Journal of the

Acoustical Society of America, 115(6):3048–3058.

De Silva, C. W. (2010). Vibration monitoring, testing, and

instrumentation. CRC Press.

Grosse, C. U. and Kr

uger, M. (2006). Wireless acoustic

emission sensor networks for structural health moni-

toring in civil engineering. In Proc. European Conf.

on Non-Destructive Testing (ECNDT), DGZfP BB-

103-CD. Citeseer.

Grosse, C. U., Reinhardt, H. W., Motz, M., and Kroplin,

B. (2002). Signal conditioning in acoustic emission

analysis using wavelets. NDT. net, 7(9):1–9.

Im, J.-P., Wang, S.-W., Ryu, S.-T., and Cho, G.-H. (2012).

A 40 mv transformer-reuse self-startup boost con-

verter with mppt control for thermoelectric energy

harvesting. Solid-State Circuits, IEEE Journal of,

47(12):3055–3067.

Ledeczi, A., Hay, T., Volgyesi, P., Hay, D. R., N

adas, A.,

and Jayaraman, S. (2009). Wireless acoustic emis-

sion sensor network for structural monitoring. Sensors

Journal, IEEE, 9(11):1370–1377.

edeczi,

A., V

olgyesi, P., Barth, E., N

adas, A., Pedchenko,

A., Hay, T., and Jayaraman, S. (2011). Self-sustaining

wireless acoustic emission sensor system for bridge

monitoring. In New Developments in Sensing Tech-

nology for Structural Health Monitoring, pages 15–

39. Springer.

Loutas, T., Kalaitzoglou, J., Sotiriades, G., and Kostopou-

los, V. (2011). The combined use of vibration, acous-

tic emission and oil debris sensor monitored data com-

ing from rotating machinery for the development of a

robust health monitoring system.

Lubieniecki, M. and Uhl, T. (2012). Thermoelectric en-

ergy harvester: Design considerations for a bearing

node. Journal of Intelligent Material Systems and

Structures, 23(16):1813–1825.

Mba, D. and Rao, R. B. (2006). Development of acoustic

emission technology for condition monitoring and di-

agnosis of rotating machines; bearings, pumps, gear-

boxes, engines and rotating structures.

McFadden, P. and Smith, J. (1984). Vibration monitoring

of rolling element bearings by the high-frequency res-

onance techniquea review. Tribology international,

17(1):3–10.

O’Donnell, T. and Wang, W. (2009). Power management,

energy conversion and energy scavenging for smart

systems. Ambient Intelligence with Microsystems,

pages 241–266.

Ramadass, Y. K. and Chandrakasan, A. P. (2011). A battery-

less thermoelectric energy harvesting interface circuit

with 35 mv startup voltage. Solid-State Circuits, IEEE

Journal of, 46(1):333–341.

Rutzig, M. B. (2013). Multicore platforms: Processors,

communication and memories. In Adaptable Embed-

ded Systems, pages 243–277. Springer.

Wang, W., Wang, N., Hayes, M., O’Flynn, B., and

O’Mathuna, C. (2012). Power management for sub-

mw energy harvester with adaptive hybrid energy stor-

age. Journal of Intelligent Material Systems and

Structures.

Weddell, A. S., Magno, M., Merrett, G. V., Brunelli, D.,

Al-Hashimi, B., and Benini, L. (2013). A survey of

multi-source energy harvesting systems. In Design,

Automation and Test in Europe (DATE).

Weddell, A. S., Zhu, D., Merrett, G. V., Beeby, S., and Al-

Hashimi, B. (2012). A practical self-powered sensor

system with a tunable vibration energy harvester. In

PowerMEMS 2012.

Zhu, D., Beeby, S., Tudor, M., and Harris, N. (2012a). Elec-

tromagnetic vibration energy harvesting using an im-

proved halbach array.

Zhu, D., Roberts, S., Mouille, T., Tudor, M. J., and Beeby,

S. P. (2012b). General model with experimental vali-

dation of electrical resonant frequency tuning of elec-

tromagnetic vibration energy harvesters. Smart Mate-

rials and Structures, 21(10):105039.

Multi-sourceEnergyHarvestingPoweredAcousticEmissionSensingSystemforRotatingMachineryCondition

MonitoringApplications

499