Rohit Dua

Impact-induced Damage Characterization of Composite Plates Using Neural Networks

Impact-induced damage in fiber-reinforced laminated composite plates is characterized. An instrumented impact tower was used to carry out low-velocity impacts on thirteen clamped glass/epoxy composite plates. A range of impact energies was experimentally investigated by progressively varying impactor masses (holding the impact height constant) and varying impact heights (holding the impactor mass constant). The in-plane strain profiles as measured by polyvinylidene fluoride (PVDF) piezoelectric sensors are shown to indicate damage initiation and to correlate to impact energy. Plate damage included matrix cracking, fiber breakage, and delamination. Electronic shearography validated the existence of the impact damage and demonstrated an actual damage area larger than visible indications. The strain profiles that are associated with damage were replicated using an in-house finite element code. Using these simulated strain signatures and the shearography results, a backpropagation artificial neural network (ANN) is shown to detect and classify the type and severity of damage.

Demodulation of Extrinsic Fabry-Pérot Interferometric Sensors for Vibration Testing Using Neural Networks

Strain level measurement on a periodically actuated and instrumented structure can provide information about the health of that structure. A simple demodulation system employing artificial neural networks (ANNs) is analyzed for an extrinsic Fabry-Perot interferometric (EFPI) strain sensor. The harmonic content of the nonlinear sensor output for the sinusoidal strain case is used to predict the maximum strain level. Implementations of the demodulation system are demonstrated for both simulated and experimental data using back-propagation neural networks. The simulation uses the theoretical response of the EFPI sensor and the experiment uses an EFPI-instrumented smart composite beam to obtain training and testing data. Excitation is provided by a piezoelectric actuator operating from 50 Hz to 1 kHz. System performance is compared for two-stage and single-stage networks and for differing types of data preprocessing. The ANN systems successfully extract the signal harmonics and predict the peak strain levels. Data preprocessing using principal component analysis produces the best accuracy. The architecture of an EFPI-based health monitoring system is proposed.

Smart Truss for Education

This work discusses a laboratory resource and associated lecture material that are implemented in an interdisciplinary course ‘Smart Structures and Sensors.’ Instruction in structural behavior, sensor systems, and experimental systems issues is supported. The Smart Truss Bridge is a reconfigurable aluminum truss structure that is instrumented with strain sensors. It is designed to be representative of a full-scale steel bridge such that person loads produce reasonable strain levels. This multiple-bay truss can be configured with and without redundant members. Strain instrumentation with extrinsic Fabry-Perot Interferometric fiber optic sensors demonstrates the performance and use of sensor systems. Theoretical analysis and experimental measurement may be correlated for different load placements. Observations include elastic strain behavior as well as practical experimental issues such as noise and non-ideal connections. A smart structures case study shows the use of artificial neural networks in a monitoring application. The weight and location of a load are predicted using different combinations of member strain. The Smart Truss Bridge provides educational opportunities for students with different majors to investigate infrastructure technologies and to interact across the disciplines. The topics are relevant to current infrastructure research and problems.