Genda Chen

Direct identification of structural parameters from dynamic responses with neural networks

A novel neural network-based strategy is proposed and developed for the direct identification of structural parameters (stiffness and damping coefficients) from the time-domain dynamic responses of an object structure without any eigenvalue analysis and extraction and optimization process that is required in many identification algorithms for inverse problems. Two back-propagation neural networks are constructed to facilitate the process of parameter identifications. The first one, called emulator neural network, is to model the behavior of a reference structure that has the same overall dimension and topology as the object structure to be identified. After having been properly trained with the dynamic responses of the reference structure under a given dynamic excitation, the emulator neural network can be used as a nonparametric model of the reference structure to forecast its dynamic response with sufficient accuracy. However, when the parameters of the reference structure are modified to form a so-called associated structure, the dynamic responses forecast by the network will differ from the simulated responses of the associated structure. Their difference can be assessed with a proposed root mean square (RMS) difference vector for both velocity and displacement responses. With the associated structural parameters and their corresponding RMS difference vectors, another network, called parametric evaluation neural network, can be trained. In this study, several 5-story frames are considered as example object structures with simulated displacement and velocity time histories that mimic the measured dynamic responses in practice. The performance of the proposed strategy has been demonstrated quite satisfactorily; the error for the estimation of each stiffness or damping coefficient is less than 10% even in the presence of 7% noise. Numerical simulations show that the accuracy of the identified parameters can be significantly improved by injecting noise in the training patterns for the parametric evaluation neural network. The proposed strategy is extremely efficient in computation and thus has potential of becoming a practical tool for near real time monitoring of civil infrastructures.

Damage Detection of Reinforced Concrete Beams with Novel Distributed Crack/Strain Sensors

Coaxial cables are used as distributed sensors to detect cracks or measure strains in reinforced concrete (RC) structures with the electrical time domain reflectometry (ETDR). An emphasis was placed on the development and validation of a type of novel cable sensors. The new sensor was designed based on the change in topology of its outer conductor under strain conditions instead of the change in geometry of a conventional cable sensor. Finite difference time domain (FDTM) models of four types of cables were established to understand the inter-relationship among various design parameters. They were calibrated with a standard displacement transducer in nine tension tests and then mounted near the surface of six RC beams of 0.91-m long to validate their sensitivity and performance. The bending test results of RC beams are in general agreement with those from the calibration test of cables. Both indicate that the proposed sensors are over 15-80 times more sensitive than sensors based on commercial coaxial cables. The test results of the beams also show that the measured reflection coefficient correlates well with the measured crack width. Overall, the new cables have a sensitivity of 160-750 milli rho per unit crack width in mm and a spatial resolution of 50 mm.

A Temperature Self-Compensated LPFG Sensor for Large Strain Measurements at High Temperature

In this paper, a CO2 laser-induced long-period fiber-grating (LPFG) optic sensor was packaged with a hybrid mechanism of elastic attachment and gauge length change for large strain measurements in a high-temperature environment. An emphasis was placed on the use of two cladding modes (LP06 and LP07) of a single LPFG sensor for simultaneous strain and temperature evaluations so that exact temperature was used to compensate strain measurements. Both strain and temperature sensitivities of the LPFG sensor, as well as the strain transfer ratio due to a combined effect of elastic attachment and gauge length change, were analytically derived and validated with tension tests at elevated temperatures. The strain sensitivity of the LPFG sensor switched sign from negative for LP06 or lower modes to positive for LP07 or higher modes, whereas its temperature sensitivity remained positive. The sign switch for the strain sensitivity resulted from two competing changes of grating period and effective refractive index as the gratings are subjected to an axial strain. The LPFG sensor was demonstrated to be operational up to 700°C for a strain measurement of up to 1.5%.

Fringe Visibility Enhanced Extrinsic Fabry–Perot Interferometer Using a Graded Index Fiber Collimator

We report a fringe visibility-enhanced extrinsic Fabry-Perot interferometer (EFPI) by fusion splicing a quarter-pitch length of a graded-index fiber (GIF) to the lead-in single mode fiber (SMF). The performance of the GIF collimator is theoretically analyzed using a ray matrix model and experimentally verified through beam divergence angle measurements. The fringe visibility of the GIF-collimated EFPI is measured as a function of the cavity length and compared with that of a regular SMF-EFPI. At the cavity length of 500 m, the fringe visibility of the GIF-EFPI is 0.8, while that of the SMF-EFPI is only 0.2. The visibility-enhanced GIF-EFPI may provide a better signal-to-noise ratio (SNR) for applications where a large dynamic range is desired.

Crack detection of a full-scale reinforced concrete girder with a distributed cable sensor

A new concept of designing cable sensors for health monitoring of large-scale civil infrastructure has recently been proposed by the present authors. The concept was developed based on the change in topology of the outer conductor of a coaxial cable sensor. One such sensor was fabricated with its outer conductor tightly wrapped with a commercial tin-plated steel spiral that was covered with solder. It was mounted near the surface of a 15 m long reinforced concrete (RC) girder with a 762 mm square hollow cross section and 152 mm thick walls. The girder was tested under a progressively increasing cyclic torsion creating 45° inclined cracks around and along the girder. The main objectives of this study were to implement the distributed cable sensor technology in large-scale reinforced concrete structures, to understand the performance of a sensor under cyclic loading for detecting and locating cracks, and, finally, to address implementation issues such as signal loss, non-uniformity in sensor construction, and recoverability.

Seismic effectiveness of tuned mass dampers considering soil–structure interaction

This paper presents how soil–structure interaction affects the seismic performance of Tuned Mass Dampers (TMD) when installed on flexibly based structures. Previous studies on this subject have led to inconsistent conclusions since the soil and structure models employed considerably differ from each other. A generic frequency-independent model is used in this paper to represent a general soil–structure system, whose parameters cover a wide spectrum of soil and structural characteristics. The model structure is subjected to a stationary random excitation and the root-mean-square responses of engineering interest are used to measure the TMDs performance. Extensive parametric studies have shown that strong soil–structure interaction significantly defeats the seismic effectiveness of TMD systems. As the soil shear wave velocity decreases, TMD systems become less effective in reducing the maximum response of structures. For a structure resting on soft soil, the TMD system can hardly reduce the structural seismic response due to the high damping characteristics of soil–structure systems. The model structure is further subjected to the NS component of the 1940 El Centro, California earthquake to confirm the TMDs performance in a more realistic environment. Copyright

Semiempirical electromagnetic modeling of crack detection and sizing in cement-based materials using near-field microwave methods

Detection and characterization of cracks in cement-based materials is an integral part of damage evaluation for health monitoring of civil structures. Microwave signals are able to penetrate inside of dielectric materials (e.g., cement-based materials) and are sensitive to local, physical, geometrical, and dielectric variations in a structure. This makes microwave nondestructive testing and evaluation (NDT&E) techniques suitable for inspection and health monitoring of civil structures. Near-field microwave NDT&E techniques offer the added advantage of providing high spatial resolution, requiring simple hardware that may be portable, low power, fast, real time, and robust. Additionally, these techniques are noncontact and one-sided. Besides the need for robust detection, electromagnetic modeling of a microwave probe response to a crack is also an important issue. Such a model can be used to obtain optimal measurement parameters and serve as the foundation for extracting important crack information such as its width and depth. In this paper, the utility of open-ended rectangular waveguide probes for detecting surface-breaking cracks in cement-based materials is discussed. Subsequently, the development of a semiempirical model capable of simulating the crack response is presented. The model described here translates the magnitude and phase of the reflection coefficient as a function of scanning distance into the complex reflection plane and takes advantage of the common shape of these signals for predicting a similar signal from an unknown crack. Finally, this empirical model is used to estimate crack dimensions from a set of measurements.

Behavior of piezoelectric friction dampers under dynamic loading

Piezoelectric materials are stressed when exposed to electric field and subjected to a restraint in their motion due to the electromechanical coupling effect. Use can be made of this property to control the motion of civil engineering structures. This paper is focused on the conceptual design of a piezoelectric friction damper and the analytical study on its behavior under harmonic loads. The friction damper takes advantage of the slip mode at the friction surface to endure the large deformation in structures and uses the piezoelectric actuators to regulate the clamped force on the damper. A new algorithm is introduced to determine the friction force for increased energy dissipation capacity. It combines the hysteretic and viscous damping mechanisms. Analytical results have shown, the superiority of the proposed algorithm over others in terms of energy dissipation. The damper is then used to mitigate the dynamic responses of a single-story frame structure subjected to harmonic loads. The structural responses controlled with a friction damper are determined numerically. However, it is found that the structure with the damper can be approximately analyzed with an equivalent linear system. This approximation greatly simplifies the design of friction dampers for practical applications.

A Fe-C Coated Long-Period Fiber Grating Sensor for Corrosion-Induced Mass Loss Measurement

This Letter reports a Fe-C coated long period fiber gratings sensor with a grating period of 387±0.1  μm for corrosion monitoring of low carbon steel in a 3.5 wt. % NaCl solution. An LPFG sensor was first deposited with a 0.8 μm thick layer of silver (Ag) and then electroplated with a 20 μm thick Fe-C coating. The chemical composition of the Fe-C coating was designed to include the main elements of low carbon steel. The resonant wavelength of the coated sensor was correlated with the mass loss of steel over time. Test results indicated a corrosion sensitivity of 0.0423 nm per 1% mass loss up to 80% Fe-C mass loss and 0.576 nm per 1% mass loss between 80% and 95% Fe-C mass loss. The corrosion sensitivity of such a Fe-C coated LPFG sensor was a trade-off for the service life of the sensor, both depending on thicknesses of the inner silver layer and the outer Fe-C coating.

Parametric Identification for a Truss Structure Using Axial Strain

The increasing use of advanced sensing technologies such as optic fiber Bragg grating and embedded piezoelectric sensors necessitates development of strain-based identification methodologies. In this study, a 3-step neural networks based strategy, called direct soft parametric identification (DSPI), is presented to identify structural member stiffness and damping parameters directly from free vibration-induced strain measurements. The rationality of the strain based DSPI method is explained and the theoretical basis for the construction of a strain-based emulator neural network (SENN) and a parametric evaluation neural network (PENN) are described according to the discrete time solution of the state space equation of structural free vibration. The accuracy, robustness, and efficacy of the proposed strategy are examined using a truss structure with a known mass distribution. Numerical simulations indicate that average relative errors of identified structural properties were less than 5% and relatively insensitive to measurement noises.