A. Yenilmez

Design and testing of an efficient and compact piezoelectric energy harvester

Piezoelectric materials generate electricity when they are subjected to dynamic strain. In this paper compact size self contained energy harvesters were built by considering typical space available for AA size batteries. Each of the harvesters contains a rectifier circuit with four diodes and a capacitor. A series of piezoelectric energy harvesters with circular and square cross-sections were built and tested at different frequency and amplitude levels. On 1M? impedance digital oscilloscope, it was observed that the voltages reached to 16V (round cross-section) and 25V (square cross-section) at 50Hz frequency. The highest power output accomplished was 625µW. The outputs of both types of the harvesters were very similar at low amplitudes. However, the square cross-section facilitates better attachment of the piezoelectric elements with the harvester shell and worked efficiently at higher amplitudes without immediate failure.

Development of Piezoelectric Strain Gages for Structural Health Monitoring Applications

The strain monitoring capability of piezoelectric materials was experimentally studied. Patches and fibers were considered for this purpose. A piezoelectric stripe actuator was used to obtain the typical signal of patches. Two designs were considered for the piezoelectric fibers. The first one was a commercial energy harvesting device with long fibers. The ends of the fibers were loose. In contrast, we designed a small piezoelectric strain gage with short fibers. Fibers were attached to wires at two ends with conductive resin. The piezoelectric strip actuator, and commercial energy harvesting device were connected to digital oscilloscope directly. A charge amplifier was used to condition the signals of the small piezoelectric strain gage with short fibers. All the units were attached to a beam and excited at different frequencies to evaluate the characteristics of their signals.

Representation of the Characteristics of Piezoelectric Fiber Composites with Neural Networks

Ideal sensors for the future should be economical, efficient, highly intelligent, and capable of obtaining their operation power from the environment. The use of piezoelectric fiber composites coupled with a low power microprocessor and backpropagation type neural networks is proposed for the development of a simple sensor to estimate the characteristics of harmonic forces. Three neural networks were used for the estimation of amplitude, gain and variation of the load in the time domain. The average estimation errors of the neural networks were less than 8% in all of the studied cases.

Structural health monitoring applications for space structures

Structural health monitoring (SHM) will be an important component of the next generation satellites and space transportation vehicles. Many SHM systems create Lamb waves and monitor their propagation for detection of structural problems such as cracks and material losses. Use of s-transformation for inspection of the signal, and fitting time delay-attenuation (DA) models by using genetic algorithms had been proposed. Robustness and performance of this approach was evaluated at the noisy and unfavorable conditions in this paper. S-transformation indicated the number of waves and characteristics of the signal. GA estimated the time delay of the DA models with less than 2% error when the waves were 180 degree out of phase with the excitation signal even though the model could not represent the phase information. Estimation of new waves with significant attenuation indicated the development of defects. The time delay information of the new waves indicated the defect location.

Design of Energy Scavengers with the Help of Finite Element Packages

Self‐powering sensors have been desired for future platform sensor networks to minimize wiring and related problems. The selection of the proper area of piezoelectric patches at various operating conditions is an important challenge since the selection of a large patch area increases the complexity and weight. The small patch area could not provide enough energy to operate the electronics continuously. Many Finite Element Method (FEM) packages are capable of estimating the electricity after the stress or strain distribution is calculated. In this paper, the required energy for smart sensors is briefly discussed and the use of FEM is suggested for selection of the size and best location of the piezoelectric patch. The study indicated that the oscillation frequency affects the mode shape and the generated energy drastically. FEM is very useful to determine the mode shapes and the selection of patch locations with maximum dynamic strain.

Evaluation and Tuning of Magnetostrictive Sensors by Using S‐Transformation

Magnetostrictive materials have been successfully used for the manufacture of various sensors. S‐transformation is proposed for evaluation of the characteristics of the signals and tuning of the sensors by considering the requirements of the considered application. The proposed method was used to evaluate the response of a Terfenol‐D rod to impulses. The signals of a coil and magnetic head were compared to the velocity signal of a laser vibrometer. The study indicated that s‐transformation allows the evaluation of the frequency response characteristics and its variation with time. Magnetic field variation of the Terfenol‐D could be effectively used with a coil to monitor the volumetric changes which is directly correlated with surface velocity, and first derivative of an external force. Very compact switches could be designed based on the magnetic head design.

Static Load Estimation for Magnetic Alloy Strips by Using Neural Networks

Magnetoelastic (magnetic alloy) sensors have been used to monitor stress. Use of backpropagation (BP) type neural networks (NN) was proposed for estimation of the load by evaluating the magnetic response characteristics of the system. In the experiments, Metglas 2826MB magnetic alloy strip was located at the axis of two coils. An alternating magnetic field was then generated with the excitation coil by applying a sweep sine wave. The envelope of the signal of the detection coil was given to a BP type NN to estimate the load after the training process. The estimation errors of the training and test cases were less than 0.01% and 2.5% of the loading range respectively. The study indicated that BP type NN could be used for mapping and interpretation of the signals of the detection coil when magnetoelastic sensors are used.

Development of Compact Forward and Inverse Estimators by Using Neural Networks

Backpropagation type neural networks were proposed to represent the forward and backward relationships between the sensory data and critical parameters. Their performance was evaluated by teaching them the relationship between the operating conditions and generated sound pressure levels (SPL). Several neural networks were used in the study to make forward and inverse estimations after they were trained by using the data obtained from the Rocket Acoustic Prediction Tool (RAPT). The average estimation error of the neural networks was less than 3% on simulated cases. Study indicated that neural networks can be used for development of smart sensors which evaluate acoustic loading and fatigue during the launching of rockets.