Recently, a variety of sensors using optical fibers have been studied intensively since these sensors offer a compact, robust, lightweight and potentially low cost solution. In this paper, we concentrate an strain sensors using fiber Bragg grating (FBG) because FBG sensors bave proven to be useful for monitoring structural strains (elongation). However, most of the FBG sensors reported so far employ a spectrum analyzer that is usually expensive. We propose here a low cost FBG sensor utilizing artificial neural networks that is used to learn the mapping relalion between strains and normalized trartsmission power spectrum. Numerical simulation based on a back propagation method demonstrates that the FBG sensors with neural networks are useful for the measurements of the strain.
ニューラルネットワークを用いたFBGセンサによる歪み計測 : 誤差逆伝搬法によるシミュレーション
52_toukou08.pdf.pdf
(479.22KB)
