This paper presents dynamic characteristics and damage inspection of beams with actual fatigue cracks by using a
boundary-effect evaluation method (BEEM) to perform space-wavenumber analysis of operational deflection shapes
(ODSs) and a conjugate-pair decomposition (CPD) method to perform time-frequency analysis of dynamic responses of
some points to a harmonic excitation. BEEM is for locating and estimating small structural damage by processing ODSs
measured by a full-field measurement system (e.g., a scanning laser vibrometer or a camera-based motion measurement
system). BEEM is a nondestructive spatial-domain method based on area-by-area processing of ODSs and it works
without using any structural model or historical data for comparison. CPD uses adaptive windowed regular harmonics
and function orthogonality to perform time-frequency analysis of time traces by extracting time-localized regular and/or
distorted harmonics. Both BEEM and CPD are methods for local spectral analysis based on local, adaptive curve fitting.
Numerical simulations and experimental results show that dynamic characteristics of beams with actual fatigue cracks
are different from those of beams with open artificial cracks. Moreover, results show that the combination of BEEM and
CPD for space-wavenumber and time-frequency analysis provides an accurate tool for damage inspection of thin-walled
structures with actual cracks.
Here we present methods for modeling, analysis, and design of metamaterial beams for broadband vibration
absorption/isolation. The proposed metamaterial beam consists of a uniform isotropic beam with many small spring-mass-
damper subsystems integrated at separated locations along the beam to act as vibration absorbers. For a unit cell of
an infinite metamaterial beam, governing equations are derived using the extended Hamilton principle. The existence of
stopband is demonstrated using a model based on averaging material properties over a cell length and a model based on
finite element modeling and the Bloch-Floquet theory for periodic structures. However, these two idealized models
cannot be used for finite beams and/or elastic waves having short wavelengths. For finite metamaterial beams, a linear
finite element method is used for detailed modeling and analysis. Both translational and rotational absorbers are
considered. Because results show that rotational absorbers are not efficient, only translational absorbers are
recommended for practical designs. The concepts of negative effective mass and stiffness and how the spring-mass-damper
subsystems create a stopband are explained in detail. Numerical simulations reveal that the actual working
mechanism of the proposed metamaterial beam is based on the concept of conventional mechanical vibration absorbers.
It uses the incoming elastic wave in the beam to resonate the integrated spring-mass-damper absorbers to vibrate in their
optical mode at frequencies close to but above their local resonance frequencies to create shear forces and bending
moments to straighten the beam and stop the wave propagation. This concept can be easily extended to design a
broadband absorber that works for elastic waves of short and long wavelengths. Numerical examples validate the
concept and show that, for high-frequency waves, the structure’s boundary conditions do not have significant influence
on the absorbers’ function. However, for absorption of low-frequency waves, the boundary conditions and resonant
modes of the structure need to be considered in the design. With appropriate design calculations, finite discrete spring-mass-
damper absorbers can be used, and hence expensive micro- or nano-manufacturing techniques are not needed for
design and manufacturing of such metamaterial beams for broadband vibration absorption/isolation.
An experimental and numerical investigation was carried out for the feasibility of determining damage development in
carbon/epoxy composite laminates with a circular hole. The study involved flash thermography to map the damaged
region in these specimens. First, thermographic response of implanted delaminations in composite laminates was
examined using both numerical and experimental approaches. Next, damage development around a circular hole in a
composite specimen was examined using thermography. The specimen was cyclically loaded until fracture to generate
different levels of damage near the center of the specimen. Thermographic images of damage development at different
fractions of fatigue life were recorded. Numerical models of thermal response of the region with evolving damage were also examined to assess the effectiveness of flash thermography under different conditions. The carbon/epoxy composite laminate was pulsed and examined using laser vibrometry in order to characterize the modes of stress wave propagation. Numerical simulations of a similarly excited carbon/epoxy panel were carried out as well.
Composite materials are profuse emitters of acoustic emissions (AE) during the occurrence of damage. In some cases, the signals measured are due to growth of noncritical damage. To have a more accurate SHM technique, distinguishing between critical and non-critical damage is necessary. This work focuses on monitoring damage growth and relating AE signals of critical growth to the reduction in structural integrity. Specifically, AE signals were experimentally measured in carbon fiber reinforced polymers (CFRP) specimens. Once the specimens were prepared, they were placed under quasi-static and fatigue loading until failure. Attenuation of AE signals were studied and mode-I delamination tests were conducted as well.
Relative motion between surfaces of mechanical parts causes surface wear and damage. The degradation to the surfaces has been monitored using vibration characteristics as well as acoustic emissions generated during the relative surface movements. In particular, acoustic emission signals were found to be sensitive to some of the microscopic processes occurring at the frictional interface. In this study, friction between two surfaces was monitored experimentally under controlled conditions. Relative velocity, contact pressure, and surface roughness values were varied in the experiments. Friction related acoustic emission signals were recorded and analyzed to understand the relationship between the signals generated and the physical processes giving rise to these signals. Information related to the stick-slip movements during cyclic motion, in the experiments, was observed from the signals. Features of the waveforms were found to reveal the conditions existing at the friction interface. In particular, the changes in the surface roughness and contact pressure were readily observed from the acoustic emission signals.
This paper shows that time-frequency analysis is most appropriate for nonlinearity identification, and presents advanced
signal processing techniques that combine time-frequency decomposition and perturbation methods for parametric and
non-parametric identification of thin-walled structures and other dynamical systems. Hilbert-Huang transform (HHT) is
a recent data-driven adaptive time-frequency analysis technique that combines the use of empirical mode decomposition
(EMD) and Hilbert transform (HT). Because EMD does not use predetermined basis functions and function
orthogonality for component extraction, HHT provides more concise component decomposition and more accurate timefrequency
analysis than the short-time Fourier transform and wavelet transform for extraction of system characteristics
and nonlinearities. However, HHT's accuracy seriously suffers from the end effect caused by the discontinuity-induced
Gibbs' phenomenon. Moreover, because HHT requires a long set of data obtained by high-frequency sampling, it is not
appropriate for online frequency tracking. This paper presents a conjugate-pair decomposition (CPD) method that
requires only a few recent data points sampled at a low frequency for sliding-window point-by-point adaptive timefrequency
analysis and can be used for online frequency tracking. To improve adaptive time-frequency analysis, a
methodology is developed by combining HHT and CPD for noise filtering in the time domain, reducing the end effect,
and dissolving other mathematical and numerical problems in time-frequency analysis. For parametric identification of a
nonlinear system, the methodology processes one steady-state response and/or one free damped transient response and
uses amplitude-dependent dynamic characteristics derived from perturbation analysis to determine the type and order of
nonlinearity and system parameters. For non-parametric identification, the methodology uses the maximum
displacement states to determine the displacement-stiffness curve and the maximum velocity states to determine the
velocity-damping curve. Numerical simulations and experimental verifications of several nonlinear discrete and
continuous systems show that the proposed methodology can provide accurate parametric and non-parametric
identifications of different nonlinear dynamical systems.
Machine parts often contain components which experience relative motion during service. Relative motion between
surfaces causes fatigue crack, wear and eventual surface deterioration. Acoustic emission based machinery condition
monitoring is a method which can potentially be used to monitor the state of damage present on surfaces during service.
This research deals with changes that occur in the characteristics of acoustic emission signals due to altering surface
roughness and texture caused by friction loading. A test fixture was used to simulate friction between surfaces of
comparable surface finish and obtain acoustic emission signals generated in the process. The different characteristics of
signal waveforms obtained at different instances during the test were examined. It was shown that some features like
amplitude and duration of the waveforms are sensitive to surface wear.
The classification of acoustic emission source mechanisms based on features related to the physics of acoustic emission
signal generation is considered in this paper. Numerically generated acoustic emission waveforms are used for this
purpose. Conventional acoustic emission parameters such as rise-time, duration, and frequency content do not
effectively characterize acoustic emission waveforms for the purpose of identifying the source mechanisms. Features
unique to the different source mechanisms and relative positions of the sensor with respect to the source were identified
and extracted from numerically obtained acoustic emission waveforms. This feature selection appears to be successful in
capturing the differences related to the source mechanisms considered here. Correlation coefficients of the 45 features
with different waveforms were first obtained, and their principal components determined. The dominant principle
components were found to adequately characterize the waveforms and relate them to their source mechanisms. Better
than 90 percent success was seen when only the first two principle components were employed, even in noisy signals
This paper presents a dynamics-based methodology for accurate damage inspection of thin-walled structures by
combining a boundary-effect evaluation method (BEEM) for space-wavenumber analysis of measured operational
deflection shapes (ODSs) and a conjugate-pair decomposition (CPD) method for time-frequency analysis of time traces
of measured points. BEEM is for locating and estimating small structural damage by processing ODSs measured by a
full-field measurement system (e.g., a scanning laser vibrometer or a camera-based motion measurement system).
BEEM is a nondestructive spatial-domain method based on area-by-area processing of ODSs and it works without using
any structural model or historical data for comparison. Similar to the short-time Fourier transform and wavelet
transform, CPD uses adaptive windowed regular harmonics and function orthogonality to perform time-frequency
analysis of time traces by extracting time-localized regular and/or distorted harmonics. Both BEEM and CPD are local
spectral analysis based on local, adaptive curve fitting. The first estimation of the wavenumber for BEEM and the
frequency for CPD is obtained by using a four-point Teager-Kaiser algorithm based on the use of finite difference.
Numerical simulations and experimental results show that the combination of BEEM and CPD for space-wavenumber
and time-frequency analysis provides an accurate tool for damage inspection of thin-walled structures.
Mating parts often experience repetitive relative motion termed fretting which results in friction, wear, as well as
acoustic emission signals. Acoustic emission signals have the potential for monitoring the condition of the surfaces
participating in the frictional process. In structural health monitoring studies, where the focus is on quantifying crack
growth related acoustic emission signals, the signals generated by other mechanisms give rise to undesirable false
positives. A major source of such false positives are fretting related signals. The present paper describes an experimental
approach for characterizing the friction related acoustic emission signals. A test fixture is developed to obtain fretting
related signals under controlled conditions. The waveforms are analyzed to extract features common to these signals. A
comparison of acoustic emission signals related to fretting and crack growth is provided.
In this research, the acoustic emissions from simulated crack growth and incremental crack growth in a cyclically loaded
aluminum panel were detected by acoustic emission sensors. One of these sensors was comprised of an array of thin
strips of piezoelectric material bonded to the specimen and electrically connected. The geometry of these sensor strip
arrays and their orientation to the fracture site enabled the sensors to capture the shear component of the acoustic
emission waveform. Cyclical loading was used to grow the crack, allowing sensor performance to be assessed in
comparison to bonded and resonant sensors. The detection of the shear wave is of particular interest as the shear
component of fretting events is often small, providing a possible means of discriminating between critical events (crack
propagation) and sources of minimal concern (fretting). Shear modes were detected in the acoustic emissions from both
the simulated crack growth and the crack growth due to cyclical loading.
Structural buckling can lead to failure, but beyond the initial onset of buckling there exists a stable region within which
there may be no material damage, the structure retains the load carrying capacity, and can recover fully elastically when
unloaded. Reclaiming this region as valid design space requires the full understanding of changes in the structure that
take place in this post buckled region that precede failure. In addition, there is also a critical need for a monitoring
technique that can measure the extent of excursion of a given structural element into the post buckled region and ensure
that there is sufficient margin of safety. Such a monitoring technique based on vibration characteristics of buckled
structures is proposed in this paper. In this research, to determine the feasibility of such an approach, the natural
frequencies and mode shapes of an aluminum shear panel were monitored while the panel was undergoing different
levels of buckling under uniform edge shear. The changes seen in frequencies and mode shapes were found to a measure
indicative of the level of buckling deformation.
The acoustic emission (AE) technique is a promising tool for monitoring the integrity of structural members while they
are in service. The major obstacle in deploying this technique is the presence of noise from extraneous sources that
generate false positives. Identification and separation of noise from crack related signals are of interest. Friction induced
AE is a prominent source of noise in structural members. When a structural member having riveted or bolted joints is
subjected to cyclic loading, the mating parts of the surfaces experience very small relative motion that results in a
localized rubbing process usually termed as "fretting". The fretting process is a prolific source of AE signals. As signals
from crack growth as well as fretting emanates from the same region in riveted joints, it is difficult to discriminate crack
related AE from fretting related AE, unless the distinct characteristics of the two signals are well understood. However,
fretting related AE signals are also of noteworthy interest in tribological applications as they contain significant
information about the surface conditions. To understand friction induced AE signals, numerically simulated fretting
signals are analyzed. Greenwood and Williamson's multiple asperity contact model is used to generate fretting signals
numerically. The results are also compared with experimentally obtained fretting signals.
Identification of the source mechanism and measurement of source strength are important requirements for wider field
application of the acoustic emission technique. It is difficult to relate a given source event to resulting acoustic emission
waveforms in experimental results. However, it is practical to simulate such source events using numerical simulations
and examine the resulting waveforms. The present paper uses such an approach to identify the patterns embedded in the
waveforms and their variation with relative positions of the source and sensor. Important elements in the waveforms are
shown to have strong variation with respect to the relative positions of the source and sensor. The resulting amplitude
variations should be taken into account in the measurement of acoustic emission source strength. In addition, it is shown
that the shear horizontal wave has a prominent component in the normal stresses in the radial direction. Acoustic
emission waveforms obtained from the numerical simulations were also used to demonstrate pattern classification of
these waveforms and identify the source mechanisms. The three elements of the waveforms, So, Ao, and Shear, were
considered as the basic elements of the waveform. These elements have different frequency bandwidths that are directly related to the impulse duration of the incremental crack growth. Correlation coefficients between these elements and the acoustic emission waveforms were used as a means for identifying the source type.
Among different failure modes observed in structures, loss of stability due to buckling is a major concern. Buckling may
be induced because of overload or as a consequence of other types of failures in the structure. This paper examines two
techniques, namely, vibration based analysis, and stress wave propagation analysis for detecting this onset of instability.
The responses of a bar and a plate are used to illustrate the effectiveness of the two approaches. These analyses were
performed through finite element simulations and limited experiments. Changes in vibration frequencies and mode shapes are found to provide good indications of the impending failure as well as its progress. Changes in the wave propagation characteristics showed some limited success in detecting the incipient buckling.
The development of high sensitivity sensors capable of accurately reproducing propagating Lamb waves is crucial for
the success of AE based structural health monitoring applications. Plate like members are the most common elements
encountered in structural health monitoring. The stress waves propagate as guided waves or Lamb waves in these
members. While the traditional acoustic emission sensors are sensitive to displacements normal to the surface of the
structural member, the bonded sensors are sensitive to the surface strains. A calibration procedure specifically for the
Lamb wave modes is devised using a Laser Vibrometer. The calibration was performed by observing the stress waves
propagating in aluminum plates. Based on this calibration, it is established that the bonded PZT sensors reproduce the
stress waveforms in these structures reasonably well. This ability was probably responsible for the success of these
sensors in distinguishing different source mechanisms and correlation with crack growth rates seen in past studies. In
addition to this calibration, the two simulated AE sources were also modeled using finite element technique. The results
of the numerical simulation were found to correlate well with the experimental results.
A large number of sensors are required to perform real-time structural health monitoring (SHM) to detect acoustic emissions (AE) produced by damage growth on large complicated structures. This requires a large number of high sampling rate data acquisition channels to analyze high frequency signals. To overcome the cost and complexity of having such a large data acquisition system, a structural neural system (SNS) was developed. The SNS reduces the required number of data acquisition channels and predicts the location of damage within a sensor grid. The sensor grid uses interconnected sensor nodes to form continuous sensors. The combination of continuous sensors and the biomimetic parallel processing of the SNS tremendously reduce the complexity of SHM. A wave simulation algorithm (WSA) was developed to understand the flexural wave propagation in composite structures and to utilize the code for developing the SNS. Simulation of AE responses in a plate and comparison with experimental results are shown in the paper. The SNS was recently tested by a team of researchers from University of Cincinnati and North Carolina A&T State University during a quasi-static proof test of a 9 meter long wind turbine blade at the National Renewable Energy Laboratory (NREL) test facility in Golden, Colorado. Twelve piezoelectric sensor nodes were used to form four continuous sensors to monitor the condition of the blade during the test. The four continuous sensors are used as inputs to the SNS. There are only two analog output channels of the SNS, and these signals are digitized and analyzed in a computer to detect damage. In the test of the wind turbine blade, multiple damages were identified and later verified by sectioning of the blade. The results of damage identification using the SNS during this proof test will be shown in this paper. Overall, the SNS is very sensitive and can detect damage on complex structures with ribs, joints, and different materials, and the system relatively inexpensive and simple to implement on large structures.
Fatigue crack growth in a lap joint specimen extracted from a retired aircraft fuselage was monitored using bonded
continuous acoustic emission sensors. The specimen lasted nearly 350,000 cycles of tension-tension cyclic loading.
During this period a large number of acoustic emission signals were collected. Two distinct classes of events were
observed during this test. The first group of events consist of low amplitude, long rise time and long duration events
which could be attributed to fretting between various surfaces. The second group of events had short rise time and short
duration and is thought to be from fatigue cracks. This interpretation is based on the waveform characteristics observed
during this test and patterns seen in acoustic emission signals from known fatigue cracks in previous studies. Based on
this assumption the crack growth process appear to have initiated after 200,000 cycles of fatigue load and accelerated
during the final 20,000 cycles. The final fracture of the specimen occurred in the grip area and indications of this
impending failure were evident in the acoustic emission data. In addition, acoustic emission data also suggest fatigue
crack growth in an area inaccessible for visual examination.
This paper focuses on actuating mode shapes of cellulose-based electro-active paper (EAPap) in order to investigate its suitability as actuators. Firstly, actuating mechanism of EAPap is addressed based on intrinsic characteristics of cellulose structures under electric fields. EAPap actuator is then fabricated by embedding gold as electrodes into both sides of cellophane sheets. Actuating mode shapes under electric fields are phenomenological measured via laser scanning vibrometer at different exciting frequencies as well as relative humidity. Various actuating performances with large deformations are obtained by applying low electric fields, which can produce a suitable deformation capability with light weight, low power consumption and simple fabrication. Experimental results provide that EAPap can be used as a potential actuating candidate for shape control of smart structures, along with bio-inspired actuator materials.
A small size prototype of a Structural Neural System (SNS) was tested in real time for damage detection in a
laboratory setting and the results are presented in this paper. The SNS is a passive online structural health
monitoring (SHM) system that can detect small propagating damages in real time before the overall failure of the
structure is realized. The passive SHM method is based on the concept of detecting acoustic emissions (AE) due to
damage propagating. Propagating cracks were identified near the vicinity of a sensor in a composite specimen
during fatigue testing. In the composite specimen, in additions to a propagating crack, fretting occurred because of
slipping contact between the load points and the composite specimen. The SNS was able to predict the location of
damage due to crack propagation and also detect signals from fretting simultaneously in real time.
A multiagent framework for data acquisition, analysis, and diagnosis in health management is proposed. It uses the contract net protocol, a protocol for high-level distributed problem solving that provides adaptive and flexible solutions where task decomposition and assignment of subtasks is natural. Java is used to wrap implementations of existing techniques for individual tasks, such as neural networks or fuzzy rule bases for fault classification. The Java wrapping supplies an agent interface that allows an implementation to participate in the contract net protocol. This framework is demonstrated with a simple Java prototype that monitors a laboratory specimen that generates acoustic emission signals due to fracture-induced failure. A multiagent system that conforms to our framework can focus resources as well as select important data and extract important information. Such a system is extensible and decentralized, and redundancy in it provides fault tolerance and graceful degradation. Finally, the flexibility inherent in such a system allows new strategies to develop on the fly. The behavior of a non-trivial concurrent system (such as multiagent systems) is too complex and uncontrollable to be thoroughly tested, so methods have been developed to check the design of a concurrent system against formal specifications of the system’s behavior. We review one such method-model checking with SPIN-and discuss how it can be used to verify control aspects of multiagent systems that conform to our framework.
Recently a new structural health monitoring system that employs a "continuous acoustic emission sensor" and an embeddable local processor has been proposed. The development of a processor that integrates the functions of signal conditioning, feature extraction, data storage, and digital communication is currently in progress. A prototype of this local processor chip has been developed. The integration of a continuous sensor with an embeddable local processor can potentially enable an inexpensive method of monitoring large and complex structures using acoustic emission signals. Such a system can reduce the cost, complexity, and weight of the required instrumentation. It is potentially scalable to large and complex structures and could be integrated into the structural material.
The success of the acoustic emission based structural health monitoring technique depends on its ability to discriminate between valid acoustic emission signals and ambient noise. In addition, the technique should be able to identify the damage mode from the acoustic emission waveforms. This paper focuses on the use of acoustic emission technique for the identification of failure modes in composite materials. Three types of failure modes in glass fabric epoxy composite laminates are considered. These are two types of delamination growth and transverse crack growth. Wavelet analysis is used to extract time frequency information from the acoustic emission signals. Different features of the waveform including the frequency components, Symmetric and Antisymmetric components, and amplitudes are used to classify the signals and identify the failure modes. The laboratory tests indicate that it is possible to distinguish the individual failure modes under consideration. It was also possible to filter out spurious AE signals that originate from extraneous sources using an appropriate choice of sensors and frequency components. An attempt is made to relate the rate of damage growth with the detected acoustic emission signal parameters.
Most structural health monitoring analyses to date have focused on the determination of damage in the form of crack growth in metallic materials or delamination or other types of damage growth in composite materials. However, in many applications, local instability in the form of buckling can be the precursor to more extensive damage and unstable failure of the structure. If buckling could be detected in the very early stages, there is a possibility of taking preventive measures to stabilize and save the structure. Relatively few investigations have addressed this type of damage initiation in structures. Recently, during the structural health monitoring of a wind turbine blade, local buckling was identified as the cause of premature failure. A stress wave propagation technique was used in this test to detect the precursor to the buckling failure in the form of early changes in the local curvature of the blade. These conditions have also been replicated in the laboratory and results are reported in this paper. A composite column was subjected to axial compression to induce various levels of buckling deformation. Two different techniques were used to detect the precursors to buckling in this column. The first identifier is the change in the vibration shapes and natural frequencies of the column. The second is the change in the characteristics of diagnostic Lamb waves during the buckling deformation. Experiments indicate that very small changes in curvature during the initial stages of buckling are detectable using the structural health monitoring techniques. The experimental vibration characteristics of the column with slight initial curvatures compared qualitatively with finite element results. The finite element analysis is used to identify the frequencies that are most sensitive to buckling deformation, and to select suitable locations for the placement of sensors that can detect even small changes in the local curvature.
This paper discusses a proof test procedure for estimating and extending the fatigue life of composite coupons. The estimates were based on the acoustic emission data collected during the described proof test procedure. A group of coupon specimens that included both undamaged as well as damaged ones were tested to verify the ability to estimate the fatigue durability. For majority of the specimens tested the fatigue life of the coupons is inversely proportional to the cumulative AE energy collected during the proof test procedure. Based on the trend that was established, a new group of specimens AE based proof test was performed and using the acoustic emission response, the life was estimated. If one could estimate the fatigue life, it would be possible to identify those specimens, which are likely to fail prematurely. For such specimens it may be possible to extend the fatigue life by appropriate reduction in the cyclic load amplitude. This hypothesis was tested on the last group of specimens. The results obtained during the life extension phase actually show that it is possible to identify the specimens, which are likely to have short life and extend the fatigue life by subjecting them to less demanding load history.
Fatigue crack growth during the service of aging aircrafts has become an important issue and the monitoring of such cracks in hot spots is desirable. A structural health monitoring system using an acoustic emission technique under development for monitoring safety of such structures is described in this paper. A “continuous sensor” formed by connecting multiple sensor nodes in series arrangement to form a single channel sensor is proposed to monitor acoustic emission signals. This paper describes the work in progress on developing sensors, instrumentation, and measurement technique applicable to on-board monitoring of fatigue cracks in 7075-T6 aluminum lap joints. The traditional AE sensors as well as bonded nodes of continuous sensors described above were used to monitor acoustic emission signals emanating from crack growth in aluminum 7075 T6 specimens. It was possible to differentiate the signals due to crack growth from noise signals arising from fretting as well as RF pickup. The sensitivity of the bonded sensor under development was comparable to commercial high sensitivity resonant frequency AE sensors. The relationship between acoustic emission parameters and the crack growth rate in the aluminum specimens is examined.
Structural Health Monitoring ideally would check the health of the structure in real time all the time. Simplifying the sensor system and the data acquisition equipment plays a very important role in achieving this goal. This paper discusses a practical technique that uses long continuous sensors and biomimetic signal processing to simplify health monitoring. The testing of a structural neural system with an updated analog processor module is discussed in this paper. A neuron is formed by connecting sensor elements to an analog processor. The structural neural system is formed by connecting multiple neurons to mimic the signal processing architecture of the neural system of the human body. This approach reduces the required number of data acquisition channels and still predicts the location of damage within a grid of miniature neurons. Different types of sensors can also be used. A piezoelectric ribbon sensor can sense damage due to impacts or crack growth because these damages generate Lamb waves that are detected by the neural system. The neuron can also receive diagnostic waves generated to check the structure on demand and when it is not in operation. In addition, new continuous multi-wall carbon nanotube sensors are being used as strain and crack detection neurons that operate during both static and dynamic loading. In general, the Structural Neural System may provide an advantage for the continuous monitoring of most large sensor systems in which anomalous events must be detected, and where it is impractical to have a separate channel of data acquisition for each sensor. Moreover, the data reduction technique and damage detection algorithm are easy to understand, simple to implement, reliable, and many sensor types can be used.
Detecting and locating cracks in structural components and joints that have high feature densities is a challenging problem in the field of Structural Health Monitoring. There have been advances in piezoelectric sensors, actuators, wave propagation, MEMS, and optical fiber sensors. However, few sensor-signal processing techniques have been applied to the monitoring of joints and complex structural geometries. This is in part because maintaining and analyzing a large amount of data obtained from a large number of sensors that may be needed to monitor joints for cracks is difficult. Reliable low cost assessment of the health of structures is crucial to maintain operational availability and productivity, reduce maintenance cost, and prevent catastrophic failure of large structures such as wind turbines, aircraft, and civil infrastructure. Recently, there have also been advances in development of simple passive techniques for health monitoring including a technique based on mimicking the biological neural system using electronic logic circuits. This technique aids in reducing the required number of data acquisition channels by a factor of ten or more and is able to predict the location of a crack within a rectangular grid or within an arbitrarily arranged network of continuous sensors or neurons. The current paper shows results obtained by implementing this method on an aluminum plate and joint. The plates were tested using simulated acoustic emissions and also loading via an MTS machine. The testing indicates that the neural system can monitor complex joints and detect acoustic emissions due to propagating cracks. High sensitivity of the neural system is needed, and further sensor development and testing on different types of joints is required. Also indicated is that sensor geometry, sensor location, signal filtering, and logic parameters of the neural system will be specific to the particular type of joint (material, thickness, geometry) being monitored. Also, a novel piezoresistive carbon nanotube nerve crack sensor is presented that can become a neuron and respond to local crack growth.
This paper examines the use of continuous sensors to detect damage in composite materials. Continuous sensors contain multiple interconnected sensor nodes that can be integrated into an artificial neural system as an array of sensor nerves. The advantage of this passive health monitoring approach is that the sensor system is highly distributed and uses parallel processing allowing large structures to be monitored for damage using a small number of channels of data acquisition. In the paper, the continuous sensor system is modeled and simulated by solving the elastic response of a plate and the coupled piezoelectric constitutive equations. The model and simulation allow the sensor system to be optimized for a particular material and plate size. The simulation predicts that acoustic waves representative of damage growth can be detected anywhere in the plate using a simple artificial neural system. To improve the sensitivity of the continuous sensor, unidirectional active fiber composite sensors were built from piezoceramic ribbon preforms. Manufacturing of the active fiber composite sensors is also discussed in the paper. The continuous sensors were evaluated in a realistic test to show their ability detect acoustic emissions caused by damage to a composite material. The sensors were mounted on narrow glass fiber plates and tested to failure in a mechanical test machine. Results from the experiments are presented.
This paper discusses recent advances in modeling and simulation of an artificial neural system and simulation of wave propagation for designing structural health monitoring systems. An artificial neural system was modeled using piezoceramic nerves and electronic components. Wave propagation in a panel is modeled using classical plate theory and a closed-form solution of wave propagation and reflection is obtained. Equations representing a half sine input similar to a projectile impact or a tone burst excitation were added to the existing algorithm that predicts the response of the artificial neural system due to impulse inputs. Firing switches have been modeled in the simulation to predict the sequential firing of the neurons as the waves pass over them. Also, new active fiber sensors have been designed for use in the artificial neural system. Simulated responses of the artificial neural system are shown in this paper and indicate that large neural systems can be formed with hundreds of sensor nodes. Experiments were performed to study a small neural system on a glass fiber panel. Waves were induced in the panel due to a lead break to simulate a crack and due to an impact from an impact hammer. Testing showed the location of a crack could be determined within the unit cell of the neural system for an orthotropic panel.
This paper discusses the development of continuous Active Fiber Composite sensors to detect damage in composite materials. Continuous sensors contain multiple interconnected sensor nodes that can be integrated into an artificial neural system as an array of sensor nerves. Continuous sensors have demonstrated a possibility of damage detection in large structures when used as a part of Artificial Neural System. The advantage of this passive health monitoring approach is that the sensor system is highly distributed and uses parallel processing allowing large structures to be monitored for damage using a small number of channels of data acquisition. In the paper, the continuous sensor system is modeled and simulated by solving the elastic response of a plate and the coupled piezoelectric constitutive equations. The model and simulation allow the sensor system to be optimized for a particular material and plate size. The simulation predicts that acoustic waves representative of damage growth can be detected anywhere in the plate using a simple artificial neural system. To improve the sensitivity of the continuous sensor, unidirectional active fiber composite sensors were built from piezoceramic ribbon preforms. Manufacturing of the active fiber composite sensors is discussed in the paper. The continuous sensors were evaluated in a realistic test to show their ability detect acoustic emissions caused by damage to a composite material. The sensors were mounted on narrow glass fiber plates and tested to failure in a mechanical test machine. Results from the experiments are also presented.
Most structural health monitoring analyses to date have focused on the determination of damage in the form of crack growth in metallic materials or delamination or other types of damage growth in composite materials. However, in many applications local instability in the form of buckling can be the precursor to more extensive damage and unstable failure of the structure. If buckling could be detected in the very early stages, there is a possibility of taking preventive measures to stabilize and save the structure. Relatively few investigations have addressed this type of damage initiation in structures. Recently, during the structural health monitoring of wind turbine blades, local buckling was identified as the cause of premature failure. Results from this investigation suggested that stress waves could be used for detecting the early signs of change in the local curvature that precedes buckling type of failure in this structure. These conditions have been replicated in the laboratory and detailed investigation on the ability of low frequency vibrations to detect the buckling displacement has been carried out. The experiment was performed on a composite bar. The results clearly show that low frequency vibrations could be used to detect the onset of buckling in which the local deflection is only of the order of 0.25 inches.
A new approach for the Health Monitoring of structural systems is described in this paper. This technique is based on detecting the acoustic emission signals from damage progression in structures using an array of sensory nodes. Two different sensor configurations that could be used for monitoring wide areas on a structure are discussed. An reliable and cost effective health monitoring system can be an enabling technology for the widespread use of newly discovered high performance materials and design concepts in structural applications. Without a reliable health monitoring system, the lack of service experience and the susceptibility of new classes of materials to unexpected and unknown failure modes will likely delay their acceptance into actual structures. The proposed sensory system mimics biological neurons in its architecture and such an architecture can reduces the cost and complexity of the monitoring system. It is potentially scalable to large and complex structures and could be integrated into the structural materials. The paper summarizes recent work related to this sensory system and provides some new results.
Recent structural health monitoring techniques have focused on developing global sensor systems that can detect damage on large structures. The approach presented here uses a piezoelectric sensor array system that mimics the biological nervous system architecture to measure acoustic emissions and dynamic strains in structures. The advantage of this approach is that the number of channels of data acquisition used for an N-by-N sensor array can be reduced from N<SUP>2</SUP> to 2N. For large arrays the number of data acquisition channels is tremendously reduced. When transient damage events occur on the structure, the array output time histories can be recorded and the location of the excitation can be accurately determined using combinatorial logic. A trade-off is the difficulty of extracting individual sensor time histories from the array outputs without a neural network or a regressive technique. Only the sums of the sensor strains of each row and column can be exactly calculated using the voltage outputs of the array. The array approach allows efficient use of data acquisition instrumentation for structural health monitoring. Applications for the sensor array include crack and delamination detection, dynamic strain measurement, impact detection, and localization of damage on large complex structures.
This paper discusses the potential for using Piezoceramic and Nanotube materials to develop an artificial neural system for structural health monitoring. An artificial neural system array was modeled using piezoceramic nerves and electronic components. The neural system was simulated using one hundred dual-output sensor nodes on a four-foot square composite panel. The nodal outputs were combined into twenty neuron firing signals, one row time signal, and one column time signal. This system was able to detect and locate acoustic waves and large strains in the panel. Also discussed, is the potential for using nanotubes for building the artificial neural system. In carbon nanotubes, an electrochemical process can be used to achieve low voltage actuation at high strain, but the process velocity is slow and a structural polymer electrolyte must be used for ion exchange. Carbon and boron nitride nanotubes can be piezoelectric, and piezonanotechnolgy may be useful for building high bandwidth neural systems. The operating temperature of boron nitride is high and the amount of material needed to build artificial nerves is small, but the piezoelectric coefficients appear to be small. Nanotube molecular electronics and the change in conductance of nanotubes might also be used to develop artificial nerves.
This is an overview paper that discusses the concept of an embeddable structural health monitoring system for use in composite and heterogeneous material systems. The sensor system is formed by integrating groups of autonomous unit cells into a structure, much like neurons in biological systems. Each unit cell consists of an embedded processor and a group of distributed sensors that gives the structure the ability to sense damage. In addition, each unit cell periodically updates a central processor on the status of health in its neighborhood. This micro-architectured synthetic nervous system has an advanced sensing capability based on new continuous sensor technology. This technology uses a plurality of serially connected piezoceramic nodes to form a distributed sensor capable of measuring waves generated in structures by damage events, including impact and crack propagation. Simulations show that the neural system can detect faint acoustic waves in large plates. An experiment demonstrates the use of a simple neural system that was able to measure simulated acoustic emissions that were not clearly recognizable by a single conventional piezoceramic sensor.
An embedded active fiber composite tape was investigated for use as a sensor for structural health monitoring. The material has unidirectional piezoceramic fibers with interdigital electrodes on the top and bottom surfaces and is poled in the fiber direction. A long active fiber composite tape with segmented electrodes was modeled as a sensor to measure longitudinal stress waves in a uniform cantilever bar. Only plane longitudinal standing and traveling waves in the bar are modeled. The sensor was connected to an electrical tuning circuit to filter out undesirable noise due to ambient vibrations. The elastic response of the bar was compute in closed form at small time steps, and the coupled piezoceramic constitutive equations and the electrical circuit equations were solved by numerical integration using the Newmark-Beta method. Strain vibration and wave propagation responses were computed in the simulations. The simulations indicate that such a sensor will be capable of detecting damage to the bar from the change sin the wave propagation responses. Further, the sensor has adequate sensitivity to detect fiber breaks in the composite bar. An active fiber composite patch was also tested to measure vibration and simulated acoustic emissions.