Application of the principal components analysis technique to optical fiber sensors for acetone detection

Abstract

Generally, the data analysis of sensors based on spectral measurements, for instance, fiber Bragg gratings (FBG), long period fiber gratings (LPFG) is performed observing the shift or attenuation of a single wavelength. However, sometimes such spectrum variations are not evident or clearly perceptible; therefore, it becomes necessary to use other analysis techniques. One of them is principal components analysis (PCA) which allows to explore the behavior of the complete transmission spectrum pattern. In this work, we show the application of PCA to the response of fiber optical sensors for acetone detection. Two kinds of acetone sensors based upon single long period fiber gratings (LPFG) and Mach-Zehnder fiber interferometers (MZI), formed with a pair of cascaded LPFGs, coated with polydimethylsiloxane (PDMS) were developed. The sensor response was measured inside a Teflon chamber injecting a liquid acetone sample. The transmission spectrum variations at the devices output were measured with an optical spectrum analyzer, stored in a computer and analyzed by PCA. A comparison between the performance of both kinds of sensors was carried out and it was found that the sensitivity was increased up to 13 times (from 0.7 × 10−4 ppm−1 to 9.4 × 10−4 ppm−1) for the MZI sensors compared with that of single LPFG ones, which means, at best, a limit of detection (LOD) of 170 ppm.