Abstract
A magnetostrictive vibration sensor has several advantages over conventional piezoelectric sensors. First, due to the high impedance output signal, piezoelectric sensors require a charge to voltage converter and a power supply to process the signal. In the case of the magnetostrictive sensors, the output signal is already a low impedance voltage which can be analyzed directly. A second advantage is its structural simplicity. The sensor can be easily assembled or disassembled. A third advantage of the magnetostrictive vibration sensor can be attributed to the high Curie temperature of magnetostrictive materials, which stands at 380°C as compared to 180°C for a typical piezoelectric material. This study presents a novel application of giant magnetostrictive material to a three-dimensional vibration sensor. The sensor is called three-dimensional because it is intended to measure a vibration exerted along an arbitrary direction. Sensor signals are measured for various directions of vibration based on which the identification of unknown vibration is attempted. A neural network is employed to learn the observed data. The generalization of the trained network is tested on unknown vibrations.
